NegPy turns film scans into finished positives with a non-destructive, darkroom-style pipeline. It never writes back to your source files. Every edit lives in a local database, so you can experiment freely.
This guide is for new users. It explains what each control does and when to reach for it. For why the pipeline is ordered the way it is, read PIPELINE.md.
If Windows blocks writes to NegPy's default data folder, NegPy suggests %LOCALAPPDATA%\NegPy\data. Click the path to choose another folder, then select Use This Folder. No data is copied or moved: a new folder starts with new settings and no saved edits; an existing NegPy data folder uses its own data. Your original data stays in place. The dialog also explains how to allow NegPy through Windows folder protection and keep using the original folder. NegPy remembers your choice in data-location.json, even if Documents becomes writable. An explicit NEGPY_USER_DIR takes priority. An inaccessible saved folder produces an error instead of switching to other data.
- Left, the film strip: your loaded frames as a contact sheet, plus import, sorting, and triage tools.
- Center, the canvas: the live preview of the current frame. Most tools (crop, white-balance picker, heal brush, dodge/burn masks) are used by clicking directly on it. Scroll/pinch to zoom and drag to pan; a floating toolbar along the bottom holds Fit/1:1 zoom (1:1 is one scan pixel per screen pixel, and lights up while you are at it; below HQ the preview is scaled up to reach it, which a preview res · HQ off pill on the canvas says) plus undo/redo, rotate/flip and more, moving overflow items into an ⋯ menu when the window narrows. What does not fit collapses from the right, and the ⋯ menu keeps every action whatever the row shows. Preferences… in that menu holds every app-wide setting (§14): the interface options, the performance budgets, Edit Toolbar… for which controls sit on the row and in what order, and Persistent Settings… for what carries onto the next file you open. Right-click the image for Reset View and Sticky Zoom (keeps the current zoom level when you switch to another frame, instead of resetting to fit), alongside the picker tools, copy/paste settings, and Unload (removes the frame from the session; its saved edit is kept). With nothing loaded it shows Load some scans to get started; click it for Add Files / Add Folder.
- Right, the controls: a pinned Analysis readout at the top, and below it an icon tab bar. Each icon opens a workflow page holding one or more collapsible panels.
NegPy remembers the window size and whether it was maximized, the panel widths and which edge each panel sits on, which sections were open, and which controls tab was active. Reset Panel Layout in the ⋯ menu puts both panels back at their home edges and widths.
The canvas toolbar's ◑ button (or \) splits the canvas in two. Left of the divider is the auto baseline: the same frame with the same film process, crop and rotation, but with every creative control (exposure, tone, Lab, dodge/burn, toning, retouch and finishing) back at its default. Right of it is your edit. Drag the divider to move the split, or grab its knob in the middle.
The split stays up while you work, so a slider moves the after side against a fixed reference. Press \ or Esc, or move to another frame, to close it. Peek Negative, Peek Flat Scan and the test strip take the canvas over, so they close it too.
The canvas toolbar's film button (or N) shows the scan as it was loaded: the negative, un-inverted, with no metering, no film-base normalization and none of your edits. Use it to judge the scan rather than the print: whether the frame is thin or dense, what color the mask really is, whether the scanner clipped. It changes nothing and closes as soon as you touch a control. Your crop, rotation and flip still apply, so the frame stays where you put it. The frame is color managed and lifted to the display by its own brightest tone, so it reads like the negative on a light table, with a C-41 mask as orange as it is in the file. That lift is what sets the brightness, so a thin and a dense capture of the same negative land at the same level: read density from the picture and the density histogram, not from how bright the peek is. The soft proof stays off: this is the scan, not a print.
The ⋯ menu's Peek Embedded Preview (or P) shows the camera's own JPEG of this capture, at your crop and rotation. The camera's white balance, tone curve and clipping are baked into it, and nothing in NegPy is derived from it. Use it to check the scan against what every other viewer shows for the same file, or to see the frame your camera metered. A file that carries no preview, such as a scanner TIFF or most converter DNGs, says so and the item stays off.
While a peek is up the canvas carries a NEGATIVE, EMBEDDED or FLAT SCAN badge in its top corner, and the menu item shows a checkmark, so a view you left on is never mistaken for a broken render. Esc leaves whichever view has the canvas — any peek, the before/after split, or a test strip.
The right-hand tabs follow the order you work in, which mirrors the processing pipeline:
| Tab | Icon | Panels | What it is for |
|---|---|---|---|
| Setup | cogs | Presets · Calibration · Process · Roll Analysis | Film type, capture-side color corrections, negative→positive normalization, roll-wide baselines |
| Geometry | crop | Geometry · Flat Field | Crop, straighten, lens and falloff correction |
| Exposure | sun | Filtration · Tone · Dodge & Burn | White balance, print density/contrast/curve/saturation, local burns |
| Color | palette | Lab · Alternative Processes · Toning | Chroma, sharpening, effects, lith and cyanotype printing, split and chemical toning |
| Finish | brush | Retouch · Finishing | Dust removal, vignette, border, carrier |
| Favorites | star | Your chosen sliders | Quick access to the controls you use most |
| History | clock | Work prints · Edit history | Keep named versions, step back through every change |
| Export | file | Export settings | Format, size, color, batch output |
| Metadata | tags | Archival metadata | Original camera, lens and film details |
| Scan | camera | Scanner · Camera Scanning | Capture film directly (Linux/macOS) |
You do not have to touch every panel. The defaults are tuned to produce a good print straight away, and most frames need only a crop, perhaps a white-balance nudge, and export.
A small dot on a panel header, and on a tab icon, means you changed something from its default. Every panel header has a reset action and an ⓘ that opens this guide at that panel's section.
Both side panels can be narrowed to give the canvas more room. As the controls panel shrinks, tab icons that no longer fit move into a » menu at the right of the tab bar. The tab you are on always stays visible.
Open a frame you have not edited and it does not start from bare defaults: the settings that belong to the rig and the roll rather than the picture come with it: film process, crop ratio, flips, calibration, paper stock, the Lab polish and your export preferences. The look itself (density, filtration, tone curve, toning, dodge and burn) starts clean on every frame.
Preferences → Session & Storage → Persistent Settings… changes that list. Every setting the copy/paste picker knows is there, grouped by panel; tick one to make it carry, untick one to stop it. Tick the whole group from its header checkbox. Values shown are the ones from your last saved edit, so the list reads as what would actually carry.
A frame you have already edited keeps its own look whatever you tick, since only export and metadata settings reach it. Reset Settings on a frame ignores this list and returns it to bare defaults.
Ctrl in this guide is ⌘ on macOS, and every menu, tooltip and shortcut list in the app shows it that way.
On macOS NegPy has a menu bar. Almost nothing in it is new: apart from Report an Issue, every item runs something a button, a window control or a keyboard shortcut already runs.
- NegPy: Preferences… (
⌘,), which macOS shows in the application menu beside About and Quit. - Window: Minimize (
⌘M), Zoom, Close (⌘W) and Bring All to Front, then every open NegPy window (the main window plus any live view or calibration window) with a tick against the front one. Select a name to bring that window forward. Minimize and Zoom are unavailable while a window is full screen, as elsewhere on macOS. - Help: Take the Tour, Keyboard Shortcuts, Customize Shortcuts, the Analysis panel guide, Report an Issue (which opens the NegPy issue tracker in your browser), and Check for Updates.
Full screen is macOS's own, so NegPy adds no item for it: use the green window button. macOS puts an Enter Full Screen item in the View menu of an app that has one, and NegPy's View menu comes later.
A menu shows a key only when that key uses ⌘. A shortcut bound to a plain key, such as ? for Keyboard Shortcuts, still works, but a menu cannot claim it: macOS gives a menu key priority over everything, so a plain ? in the menu would fire while you type a ? into the film strip search box. Rebind a shortcut and its menu item follows.
⌘W on the main window closes NegPy. Windows and Linux have no menu bar; nothing there changes.
The header shows the NegPy logo and version. The ↻ button beside the version number asks GitHub for a newer release on demand; it becomes a green ⬇ when one is out. When a newer release is out, a green ⬇ Update Available line also appears under the version; click either to read what changed and let NegPy install it (§15). The chevron at the header's top-right folds the branding away to give the frames more room.
Below the header: the toolbar, the search box, and two collapsible sections. Library holds the folders your scans live in; Film Strip holds the frames you have open. Click either heading to fold it away; the one still open takes the whole panel. NegPy remembers which were open.
The Library section is a folder tree of the places your scans live. Press + to add a folder, and point it at the one big Scans directory you keep everything under, subfolders and all. ↻ re-reads it from disk. Each row shows what is inside it ("36 photos", "2 folders"), and subfolders are read when you expand them.
Browsing costs nothing. NegPy opens, decodes and hashes nothing when you add a folder or click through the tree. It only lists what is there.
The Library button (book icon, first in the toolbar, or Ctrl+L) opens the folder your scans live in. The first time you press it, NegPy asks you to pick that folder and remembers it. The panel also goes there on its own: on launch when you do not restore a session, and whenever you unload the last frame. Your rolls are a more useful resting state than an empty sheet.
To point it somewhere else, add another folder with +. To forget them all, use Clear Library in Manage Database. That clears the list of folders only, leaving your images, folders and edits untouched.
- Click a folder to select it, double-click (or Enter) to open it.
- Ctrl+click several folders and open them together to load more than one roll at once. NegPy asks once, for the total.
- Alt+Up moves the selection to the folder above.
- The tree sorts the way the sheet does. Change Sort to Date or Descending and the folders follow.
When you open a folder that contains images, NegPy asks whether to load the roll. Only then does it hash and thumbnail them, which is the part that takes a moment on a big roll. Say no and your open frames stay as they were. Tick Always load without asking in that prompt if you would rather it just get on with it.
Loading a roll replaces what is in the film strip; right-click → Add to Session appends instead. Nothing is lost either way, because your edits live in NegPy's database, not in the list of open files.
NegPy reads the tree straight from disk and never creates, renames, moves or deletes anything in it. Reorganize in Finder or Explorer and the tree shows the new arrangement at the next refresh. Every edit is stored against the image's content, so moving a file between folders keeps its edit, its history and its keep/reject mark.
A note on Nikon High Efficiency raw. The Z 8 and Z 9 can record NEFs in High Efficiency (HE) or HE*, which use a licensed codec NegPy cannot decode. Such a file is still called .NEF and still carries the same TIFF compression tag as an ordinary lossless NEF, so nothing looks unusual until it fails to open. NegPy names the reason rather than reporting a generic unsupported-file error. Re-shoot in Lossless Compressed NEF, or convert with Adobe DNG Converter. Lossless NEFs from the same cameras open normally.
Toolbar buttons, left to right:
-
Add Files / Add Folder: load individual images or every image in a folder. Pick a folder that holds only other folders and NegPy reveals it in the Library section instead of reporting that it found nothing. Dropping a folder on the window does the same.
-
Clear All…: unload everything, or just the selected frames.
-
Hot Folder: watches the current folder and auto-loads new files as they appear, which is handy when a scanner or tethering app drops files into a directory. While it is on, the "Working…" import popup stays hidden so each new frame does not raise a window; the status line over the canvas still reports the import.
-
Trichrome Scan (three-exposure narrowband capture, also called trichromatic capture): treats the folder as red/green/blue exposure triplets and assembles each frame from three shots. Shots are grouped in the order they were taken, read from the files themselves, so filenames need follow no convention. Capture each frame's three exposures back to back, before moving on to the next frame. Where a file states no capture time, the folder falls back to filename order, and the names then have to sort into capture order. Right-click a frame → Edit RGB Triplet… to assign the three files by hand. Three shots are only assembled when they are one of each color and show the same frame; where they do not, they are left as separate frames to pair by hand, rather than assembled wrong. An assembled frame carries the three-dot badge described under Triage.
-
Half Frame: splits each scan into two frames, for half-frame cameras. Each half is edited and metered separately and badged with which half it is. Enabling it opens a rectangle editor on the current scan: drag the green box to crop (everything outside is discarded), drag the orange line to set the split, and use Cut thickness to discard a band centered on the split, which is the physical black separator between the two exposures. Auto-detect, in the editor, re-finds the gutter on the current scan.
The editor's Apply is a split button, like Export's: its ▾ picks what it saves to — Apply to current frame, Apply to selected frames, or Apply to all frames (the roll-wide default, which every frame without its own override follows) — remembered as the default for next time. The tune-icon button beside Half Frame opens its menu: Adjust Split… re-opens the editor on the current scan, for the odd frame a shared split gets wrong on a roll with a bit of irregular spacing; Auto-detect All Splits re-finds the gutter on every loaded scan in one pass and saves each as that frame's own override; Unsplit Diptych (enabled only while the current frame is one) reverts it, the same as right-clicking it. Right-click a half-frame asset for the same editor, defaulted to Apply to current frame, via Adjust Split for This Frame…, plus Reset Split to Roll Default once it has an override.
Recropping or resplitting a frame carries its heal strokes, dust spots, scratch lines and dodge/burn masks along with it, re-anchored to the same spot on the film — painting over a scratch, then nudging the split later, does not leave the repair stranded in the sky.
Turning Half Frame off does not lose the work: each half you edited keeps its own edit. A half you only looked at is not work, so turning Half Frame on and back off again leaves the scan as it was. A scan you did split comes back as one frame carrying both halves, a diptych: half 1 rendered with its own edit, half 2 with its own, joined side by side at the original spacing, with the cut band as a black gap. A scan whose halves hold edits from another folder or an earlier session is not a diptych; it stays one plain frame until you split it again. A diptych carries the both-sides-filled split badge and exports as one file named
<name>-DIPTYCH. Its controls panel is disabled, because the edits belong to the halves: turn Half Frame back on to change either one. A scan where only one half was worked on uses that half's edit for both sides. The filmstrip thumbnail and contact-sheet tile still show the plain whole scan, so a diptych's thumbnail does not match what it exports. An export size set as a long edge applies to each half, so a diptych comes out about twice that wide. To get a plain frame back, choose Unsplit Diptych from the Half Frame menu or its right-click equivalent: it renders and exports whole again, and both halves' edits are deleted, so a later split starts from defaults.Half Frame does not apply to a frame assembled from more than one file: a Trichrome triplet, a stitch, or an HDR merge. Those are never split, and they never come back as a diptych, even when the file they are built around was worked on as two halves earlier.
-
Apply (clone): copy the current frame's settings to selected frames or the whole roll. You choose which aspects in a dialog; crop and rotation are always per-image.
-
Sheet filter (funnel): show All Frames, Keepers Only, or Hide Rejected. The choice is remembered between sessions and applies to every roll you open.
-
Sort: by Name or Date, ascending or descending.
Above both sections sit a filter box, a .* regex toggle and a search-library button. Inside the Film Strip section is a tally, for example "36 frames · 12 keepers · 3 rejected". While a filter hides frames the tally counts both sets and names the filter, for example "3 of 36 frames · Keepers filter". When a filter hides every frame, the strip carries a message with a Show all frames link that clears the filter box and the funnel together.
Type a plain word and it matches the filename. Beyond that the box takes field:value terms, which is how you find a frame by what it is rather than what it was called:
| Term | Finds |
|---|---|
film:portra |
frames whose film stock contains "portra" |
camera:"Nikon F3" |
quote anything with a space |
iso:>=400 |
numeric fields also take >, >=, <, <= (iso, frame, push, devtime, temp) |
date:2024-03 · date:>=2024 |
by file date; a partial date is a prefix |
shot:1998 · shot:>=1998-07 |
by capture date from the Metadata panel, not the file date |
place:tokyo |
by capture city, state or country |
devtime:>=9 · temp:20 |
development time in minutes, and temperature in °C |
roll: developer: dilution: lens: format: scanning: |
the rest of the Metadata panel |
name: path: ext:tif |
file identity |
keeper: rejected: edited: |
frames carrying that mark, or with a saved edit |
-rejected: -film:velvia |
a leading - negates any term |
Terms combine with AND, so film:portra iso:>=400 -rejected: applies all three at once. Metadata comes from each frame's own Metadata panel, so it is searchable once you have filled it in; a frame you have never edited is findable by name, extension, date and mark. The .* toggle switches the box back to a plain regex over filenames, ignoring the field syntax.
The filter box narrows what is already open. The magnifier-over-folder button beside it (or Enter in the box) runs the same search across every library folder and loads what it finds, so film:portra finds your Portra frames in folders you have not opened this month. The status bar counts files as it goes.
This works without opening anything, because NegPy already knows which edit belongs to which file. Film stock, camera and the rest come from frames you have filled in. The folders are only read, never indexed in the background and never modified.
Right-click empty space in the film strip for Add Files, Add Folder and Clear All…, so those tools stay in reach part-way down a long roll. Here Clear All… always means the whole session, never just the selection.
If one negative was captured in overlapping pieces, from a copy stand at higher magnification than the frame, select the pieces and right-click → Stitch Selected Frames. NegPy finds the overlap, matches brightness across the seam and replaces the parts with a single wide composite named a+b (Stitch), badged on the sheet. The parts keep their own edits on file, so right-click → Unstitch puts them back untouched. The registration is saved and replayed, so a composite stays whole across other folders and other launches until you unstitch it, and re-opening it costs nothing.
This works on Trichrome frames too. Turn on Trichrome Scan first so each piece is already assembled from its own R/G/B triplet, then stitch the assembled frames. Each part keeps its own three exposures, and nothing is shared between parts.
A slide's density runs deeper than one camera exposure can record. Expose for the highlights and the darkest parts sit in sensor noise; expose for those and the bright parts blow. Bracket the capture instead, several shots of the same slide a stop apart, then select them and right-click → Merge Exposures (HDR). They collapse into one frame named a +4 (HDR) that carries the whole range, badged on the sheet so a merge is never mistaken for a single capture. Right-click → Unmerge Exposures puts the originals back, with their own edits untouched.
Nothing needs to be set up beforehand. NegPy measures the exposures from the images themselves rather than trusting shutter tags, since several supported scanner formats have none, works out how many stops apart they are, and registers them to each other in case the camera shifted. The result is saved and replayed, so a merge stays whole across other folders and other launches until you unmerge it, and re-opening it costs nothing.
How it lands. Two separate choices, worth keeping apart. The merge is computed in the units of the longest exposure that does not clip: the best reference radiometrically, since every other frame converts into it and nothing can exceed white. Which exposure the picture then opens at is a different question, and not one the software can answer. A slide's brightest point is denser than clear film, so the longest unclipped capture is brighter than the shot you metered, and rendering there pushes the highlights into the top of the transfer curve where it has least gradient left. That looks like lost highlight detail, because it is.
So nominate the frame yourself. Right-click a merged frame → Render exposure and pick the shot that looks the way you intended. The merge opens exactly there, and the other frames contribute only range and cleanliness. The list shows each frame in stops from the reference, with the reference itself marked (as captured).
Only the reference and any shorter exposures are listed. The merge can never open brighter than the reference, the frame defining white, so a longer frame would render identically and is not offered. If a bracket has nothing below its reference, the menu does not appear.
Or set it as a value. With a merged frame selected, the Process panel gains a Render Exposure slider: 0 EV is the reference, the brightest the merge can open at, running down to −4 EV. The menu below snaps to exposures you actually shot; the slider goes anywhere between them, which is usually where the one you want sits. Setting a value clears any frame you had nominated, and picking a frame clears the value, so only one is ever in effect.
Left on Bracket middle (auto) it falls back to the middle exposure of the bracket. That is a reasonable guess only when you bracketed evenly either side of the metered shot. Bracket upward and every frame sits at or above the reference, so the middle lands on the reference and the setting does nothing.
Include the shot that already looks right, and at least one darker than it. It is tempting to bracket only upward, since frames longer than the reference are what buy the shadows. Do not: the menu can only offer frames you actually shot, so the darkest render you can ask for is the darkest exposure in the bracket. The reference typically lands one to three stops above the metered frame and is never the metered frame itself, so (as captured) is not the exposure you took, and the one you want is usually a stop or two below it.
What the merge buys is signal-to-noise in the deepest shadows, worth roughly two stops, with the midtones unchanged. The gain is entirely where a transparency is hardest to scan.
A merge opens with its shadows already lifted, by an amount derived from the range the bracket recovered, so Shadows Density sits off zero. That is deliberate. If your metered frame did not clip, the merge did not add range, it added precision: the tones were all recorded, just with very few levels on top of noise. Precision is invisible at the same tone, so a merge left neutral renders indistinguishable from the frame it was metered on, which is not what you merged for. The lift is measured, not a look: it goes only as far as the recovered precision affords, so the opened shadows are still quieter than the single frame's were. Drag Shadows Density to zero for the render faithful to the metered frame, and that choice is saved like any other edit. Reset Settings brings the seeded starting point back, along with the merge itself and the inherited film process.
Bracket both ways, for two different reasons.
Upward, for range. Longer exposures reach the deep shadows, and they set the reference. Metered, +1, +2, and +3 if the shadows are deep.
Downward, for choice. Shorter exposures add almost nothing to the pixels, because the reference already holds the highlights by construction. Their job is to appear in the Render exposure menu, and a merge is only as adjustable as the frames you gave it. Stop at the metered shot and the menu offers two entries, with no room left to go darker; go two stops below it and there are four or five, reaching −3 and −4 EV.
Metered −2 through +3 costs six frames and leaves the decision to you at the end. If you must economise, economise upward: an extra long frame buys shadow noise you may not notice, an extra short frame buys a choice you cannot make later.
Shorter frames do carry one thing outright: a blown specular, the sliver of water highlight or sun disc above the reference's white.
The merged frame inherits the film process of the exposures it came from, so a bracket of slides opens in Transparency rather than reverting to whatever mode you last set by hand. Stitched composites do the same.
It is named after the first frame in filename order, with an -HDR suffix, so a bracket of _DSC1715…_DSC1719 exports as _DSC1715-HDR.jpg. Not the reference frame, whose identity depends on picture content; and the suffix means a merge never writes over the export of the single frame it is named after.
Merging is for transparencies, so the action appears on Transparency frames only. A color negative holds about 5-6 stops between its base and its densest highlight, and an ordinary black-and-white negative nearer 4, both comfortably inside one capture. A transparency runs to 10-12, which is what the merge exists for. On black-and-white the entry is shown but disabled, because reversal-processed monochrome (Scala, dr5, Fomapan R) really is a transparency and does have the range. It is simply not wired up yet.
Merging is also refused on frames that are already merged, stitched, or Trichrome triplets. Each is its own way of building one frame from several files, and combining them is not supported.
Narrow the panel and the toolbar buttons that no longer fit move into a » menu at its right edge, so you can squeeze the panel down without losing any tool.
Thumbnails are positives from the start. A frame you have not opened yet is inverted straight from its preview, a quick per-channel job rather than the full pipeline, so the sheet reads as photographs while you cull. Open a frame and its thumbnail is replaced by the real render, matching the canvas exactly. Transparencies are left alone, being positives already: a frame whose film process you have already set, or that you have opened once, is taken at its word, and only a frame nothing has decided yet is guessed at from its preview.
Right-click a thumbnail, or use keyboard shortcuts, to mark frames while you review the sheet:
- Keep: a small check badge marks a keeper.
- Reject: a cross badge dims the frame. Rejected frames stay on the sheet but are skipped by batch exports and sidecar writes. The file on disk is never touched.
Marks apply to a multi-selection and persist across sessions. A badge in the top-right corner instead flags a frame that failed to decode.
Each corner of a thumbnail means one thing, so the marks never compete:
| Corner | Badge | Means |
|---|---|---|
| Bottom-right | check | keeper |
| Bottom-right | cross, frame heavily dimmed | rejected |
| Top-right | exclamation | the file failed to decode; click to retry |
| Bottom-left | see below | the frame was built from more than one file |
The bottom-left badge is gray, not red, because it reports what the frame is rather than something you marked. Its glyph says which kind:
| Glyph | Frame |
|---|---|
| Two overlapping panes | a stitched composite (§Stitching) |
| Three stacked bars | a merged bracket (§Merging) |
| Three red/green/blue dots | a Trichrome triplet |
| A split rectangle, one side filled | one half of a half-frame scan; the filled side is which half |
| A split rectangle, both sides filled | a diptych: the whole scan, each half with its own edit |
Hover any thumbnail and the tooltip says the same thing in words, with the frame count: HDR merge of 5 exposures, Stitched composite of 3 frames.
The right-click menu also offers Copy/Paste Settings (with or without normalization bounds), Reset Settings, Apply Settings…, and per-frame export. A copy that took the bounds lists them in the paste picker as Normalization bounds, ticked; untick it to paste the look and keep the frame's own bounds.
Pinned above the tabs, this is your feedback while printing. Drag the divider to resize it, or collapse it entirely. Everything in it describes the frame you are on and updates as you edit. The zone strip is the one part you can also act through. Top to bottom:
The chart is the paper characteristic (H&D) curve NegPy is printing through right now. It models how a sheet of photographic paper responds, and it is not a curves editor. Left to right is negative density, the exposure the paper receives, so dense parts of the negative (the scene's highlights) sit to the right. Bottom to top is the print tone that comes out. A steeper curve means more contrast, which is what Grade moves. The flattening at each end is the toe (shadows) and shoulder (highlights), where the paper runs out of range.
With a Contrast Mask dialled in, a violet band opens between the curve and a dashed edge. The mask shifts each pixel by how far its own value sits from its blurred surroundings, so there is no single curve for it: a large flat area prints on the dashed edge, fine detail prints on the solid curve, and everything else falls between. The band is the mask's reach. Dodge/burn, local grade and CLAHE are spatial in the same way and are deliberately absent from the chart, which plots the global curve.
The crosshair marks the pivot, the density the curve rotates around when you change contrast, so the midtone stays put. While you drag a slider, a faint ghost of the previous curve stays behind for comparison. If cast removal pulls the channels apart you get three separate R/G/B traces instead of one gray curve, and that spread is the color correction.
Two different histograms share the chart. Behind the curve, rising from the bottom, is the output histogram: the tones of the print you are looking at, in R, G, B and luminance. Along the bottom axis is the negative density histogram, which is what the scan contains, before the curve.
Read them against each other. The density histogram tells you which part of the horizontal axis your negative occupies, and the curve tells you what happens to it. If the negative's data sits entirely on the flat toe, no amount of contrast pulls those shadows apart. Move the exposure so the data lands on the steep middle instead.
Peek Negative shows the scan before any curve ran, so the chart drops the curve, the output histogram and the zone strip, and the density histogram itself splits into R, G, B and luminance, since there is no print histogram left to carry color information. Each channel is scaled to its own peak, so a spike pinned to the left or right edge is that channel clipping, and a trace sitting well apart from the others is a strong color cast.
Bottom-right of the chart. It switches the histogram's height axis (how many pixels), not the tone axis. LIN is literal, so a big flat sky dwarfs everything else. LOG compresses the tall peaks so the thin tails become visible, which is where the few hundred pixels of deep shadow or specular highlight live. Use LOG to hunt for clipping, LIN to judge where the bulk of the frame sits. NegPy remembers the choice between sessions.
Small R, G and B triangles in the top corners of the chart. Top-left is shadows crushed to pure black, top-right is highlights blown to pure white. They appear only once a channel passes 0.5% of the frame. A little is normal, since a real print has a black. Watch for a single channel clipping alone, which is a color cast pushing one dye off the end rather than an exposure problem.
The amber wash on the left and the blue wash on the right mark the curve's toe and shoulder, the compressed ends where tonal separation is being lost. The ticks along the bottom are Adams zones I to IX, so you can read straight off the axis which zone a given negative density prints as.
A 21-step Stouffer-style gray wedge printed through your current curve, in even density increments labelled in the scan's own density units. It is a ruler for the curve. Where neighboring patches are clearly different, you have tonal separation. Where they merge into one flat black or white block, those tones are gone. The brackets mark the usable span. It hides while you peek the flat scan, since there is no print curve to wedge.
Ten cells on the Adams scale, where 0 is paper black and V is 18% mid-gray. The last cell (IX) also absorbs paper white. The brightness of each cell is the zone's tone, and how solid it looks is how much of the frame lands there. This is the fastest read of whether a frame is low-key, high-key or sitting sensibly in the middle. The end cells tint red when shadows are blocked up or highlights are blown. Hover a cell for its exact percentage.
It is also where you place a tone. Click a cell, then click that spot on the photo and the print is solved so the spot lands on that zone. See Zone placement below. The armed cell is outlined until you spend it. Click it again, or press Esc, to cancel.
A spot densitometer. Hover the image to read the pixel: per-channel density above film base (ΔD, relative to this scan's normalization, not absolute), the displayed tone's reflection print density, and its print zone (0 = paper black, V = 18% mid-gray, X = paper white). In B&W Negative mode the ΔD channels read the pre-conversion color record.
The probe made actionable, and what a darkroom enlarging analyser does. Click a zone on the strip above, then click that spot on the photo. The print is solved so the spot prints on the zone you asked for, and you see it straight away. Ask for a second and a third zone the same way. A fourth spot replaces whichever pin is nearer. Each pin appears here as a row: what zone it reads now, and the target it was given, which the − and + buttons trim in thirds of a zone.
With one pin, Print Density is solved so that tone prints on its target. With two, typically a shadow and a highlight, Print Density and ISO-R Grade are solved so both land. A third pin adds one more control for the tone between them: Shadows Grade, Highlights Grade or Snap, whichever can actually move that tone, which depends on where it prints rather than on which zone you call it. A line under the button says which, so nothing moves silently; if the third tone already prints where you asked, no third control is touched.
What you see until you accept is a preview. Place zones, or Enter over the photo, commits it as one undoable edit, turns off Auto Density (and Auto Grade from two pins on) since a meter left on would re-move the placed tones, and closes the tool. One pin is enough to accept. Esc discards instead: first the armed zone, then the pins and the preview with them. The ✕ on a row removes just that pin; remove the last one and you are back to the committed print.
Pins are handles: drag one to move it, and the zone beside it re-reads as it travels (the cursor turns into a hand over a pin you can grab). A pin keeps the zone it was asked for while it travels, and its caption reads 1 · IV⅓ → VI, where it is now and where you are asking it to print, the arrow disappearing once the two agree. Clicking the photo without picking a zone first simply pins a reading and leaves the print alone.
Asking for a zone the paper cannot reach shows an amber → lands … with the closest zone the print can make, the solve pegging at the slider's end like an analyser pegging at grade 5. With three pins the amber also appears when the three asks cannot all be met at once: placing the middle tone nudges the outer two, so the solve alternates between them and reports where they settle.
Pins are proofs, not edits. They go on any other edit and when you change frames. The measured zone reads through the print curve, the same model the chart and step wedge use, so after placing, the pin reads its target by construction; later stages (Lab, toning) can still shade the final pixel the hover probe reads.
The numeric rows at the bottom. Each has the same explanation on hover, and each measures the scan rather than your edit. The first three are always there. The rest appear only when they have something to say:
- Negative: the negative itself, as a relative density range (luminance) plus its development character against a nominal frame: flat (≈N−1), normal, contrasty (≈N+1). It is a relative scale, comparable across a roll, and a heuristic from this scan's normalized bounds rather than a calibrated densitometer reading.
- Exposure: where the frame's midtone sits, in stops from neutral; positive is brighter (high-key), negative darker (low-key). Approximate, read off the metered midtone.
- Clipping: share of pixels crushed to black (shadows) or blown to white (highlights), worst channel. Turns red above 1%.
- Scan clip: share of source-scan pixels at or above sensor white, per channel. In a negative scan the film base and scene shadows sit near sensor white, so clipping there destroys base and shadow separation, and no edit can undo it. Fix it at capture: expose the scan lower. Turns red above 1%.
- Repair: share of the scan that IR Restore, dust detection and painted heals rewrote, shown only once one of them fires. The overlay shows where a repair landed; this says how much. A large number means the threshold is low enough to be redrawing the picture rather than the dust on it. Measured over the whole scan, border included. Turns red above 5%.
- Gamut: share of the frame the proof profile cannot print, shown only while you are proofing to one. Clipping means a tone ran off the end of the paper; this means a color is outside what the profile can make, so printing pulls it to the nearest one available. Zero is the normal reading. Turns red above 2%.
Film mode sits above the panels, because it is the first choice of every edit: Color (C-41 color negative), B&W (panchromatic negative) or Slide (transparency/reversal, E-6 and friends). Each swaps the core conversion math and re-runs the pipeline from scratch. The wand button beside them auto-detects the mode when a file loads.
Everything here corrects the capture, not the look. Three different things sit between the scene and your file: the camera's color filters, the film's dyes, and the light source. Each gets its own control. They are not interchangeable, and none substitutes for another.
Capture, how the file is decoded before any of that:
- Scanning setup (bulb button): a two-question wizard, how do you scan? then what light source?, that sets Linear RAW and Narrowband for you. It runs once after the first-launch tour, and the button reopens it whenever your rig changes.
- Linear RAW (default off): decodes with neutral multipliers for completely raw data. When off, it decodes RAW with the camera's as-shot white balance. Toggling it reloads the file. Let the Scanning setup wizard pick it, or try both and keep whichever gives better results. Locked on for a Trichrome triplet: a narrowband exposure has no full-spectrum scene for a white-balance gain to describe, so every triplet exposure decodes neutral regardless of this toggle. It stays visible and remembered, and unlocks again once the frame is no longer a triplet.
- Narrowband: corrects the oversaturation typical of narrowband (RGB-LED) capture, using a bundled input profile. Leave it off for ordinary broadband scans. An explicit Input ICC in Export overrides it. It is grayed out on Transparency; see Narrowband and slides below.
What the wizard sets, by rig:
| Capture | Light source | Linear RAW | Narrowband |
|---|---|---|---|
| Digital camera | White light (lightbox, CRI LED panel) | off | off |
| Digital camera | Narrowband RGB (Scanlight, RGB LED) | on | on |
| Film scanner | White light (Plustek, Epson, most flatbeds) | on | off |
| Narrowband Scanner | Nikon Coolscan, Kodak Pakon | on | on |
Applying it sets the defaults for newly loaded files, updates the open frame, and rewrites every already-edited frame in the session, undoable per frame with Ctrl+Z.
Single-Shot Narrowband Calibration is for single-shot camera scans under narrowband light. The camera's color filters overlap the light's bands, so a pure red exposure leaks a little into green and blue. That leak is a fixed property of your sensor and light together and has nothing to do with the film, so it is corrected on the linear capture before inversion.
- Profile: the sensor matrix to apply. Custom
.tomlmatrices live in<Documents>/NegPy/sensor/. - Calibrate (vials icon): build a profile from three bare-light R/G/B exposures.
This block grays out unless Linear RAW is on, since profiles are calibrated against neutral white balance and the as-shot gains would misapply the matrix. It also grays out on Transparency, for the reason below. Your selection is remembered either way. It is also skipped for RGB-triplet assets, which never had the leak. It changes what the analysis reads, so re-run Batch Analysis after changing it.
Narrowband and Single-Shot Narrowband Calibration do not apply to Transparency, with or without Normalize. Both stay visible and grayed so you can see what your rig is set to, both keep their values, and both come back the moment the frame is a negative again.
Narrowband is a way to scan negatives. Its payoffs, defeating the orange mask and separating the dyes cleanly before a high-gain inversion, are things a slide does not need, and the bundled profile describes narrowband capture of negative dyes, which a slide does not have. Single-Shot Narrowband Calibration goes with it: the matrix un-mixes a narrowband light against your sensor's filters, so it means something only for a capture made under narrowband light, and there is no way to build one for a broadband capture.
This matters because both are sticky and follow your rig from frame to frame, so a profile set up for your negatives would otherwise arrive on a slide you never touched. If you do scan slides on a narrowband rig, reach for Hue Trim instead: it corrects the hue rotation an unusual light imposes, which is the part that can be corrected.
Crosstalk (hidden in B&W Negative) is a channel unmix applied to the raw densities before inversion. The dropdown lists only matrices for the film you are processing, because a Color Negative matrix does not describe a Transparency's dye set, and a mismatched stored profile resolves to no correction rather than the wrong one.
The film's dyes each absorb outside their own band, but they are not the only cause: your light's spectrum and your sensor's color filters mix the channels too, and in the density domain all three arrive as the same kind of error. So treat the matrix as your whole scanning setup, not just the film. A profile that works beautifully on one rig may be wrong on another with the same stock.
- Matrix: the profile to apply, grouped in the dropdown by where its numbers came from (measured, tuned on a rig, or from spec sheets). Generic C41 is the built-in; drop custom
.tomlmatrices in<Documents>/NegPy/crosstalk/(see CROSSTALK.md). The slider button opens a matrix editor, where a Type control records that provenance and a Process control says which film the numbers describe. Process decides where the profile appears and whether it applies, so a matrix you build for slides needs it set to E-6. Anything created with + is already set to the process you are working in. When the current film process has no matrices at all, the dropdown and Strength are disabled and a hint says so. The editor button stays live, because it is the way to build the first one. - Strength (0.0 to 1.0): how much of the unmix to apply, for richer and cleaner color separation. It changes what the analysis reads, so re-run Batch Analysis after changing it.
The bundled film matrices are derived from published spec sheets, not measured, which is why they are all marked (approx). They describe the film's dyes alone, so they are the whole story only where your capture reads each dye cleanly: a Narrowband Scanner (a Coolscan's mono sensor reads one LED at a time, fully clean; a Pakon's trilinear array comes close, with slight residual bleed), a Trichrome capture, or a Single-Shot Narrowband rig with Single-Shot Narrowband Calibration applied (see above). With a broadband light and a Bayer sensor, the capture adds mixing of its own that a dyes-only matrix does not describe. It may still help, but treat the number as a starting point rather than a correction for your setup.
Worth experimenting with. If a stock or a light gives you trouble, open the matrix editor, nudge the six off-diagonal terms and save the result as your own profile. Name it after the combination, "Gold 200 + Spectracolor", not just the film. A profile you tuned on your own rig beats any datasheet, and profiles that work are worth contributing back, since nobody can guess your light and sensor for you.
Light source:
- Hue Trim (-30° to 30°, default 0): rotates every hue by a fixed angle, to undo the rotation an unusual scanning light imposes. Narrowband LED and odd-phosphor panels sample the dyes away from where the film expects, which turns every color by roughly the same angle, so yellows read orange and greens go olive, while neutrals are left alone. That is why white balance cannot fix it: the error is a rotation, not a cast, so there is no gray to correct. Judge it on a subject whose color you know (foliage, a clear blue sky, skin), and leave it at 0 for an ordinary broadband light. The setting is sticky, because a light source is a property of your rig, so it carries to the next file until you change it. Neutrals are untouched, so it never disturbs the color-balance clip in Normalization.
A color sensor records one color per photosite behind a mosaic filter, and an algorithm fills in the other two. Which one you pick decides how sharp the result looks and how it treats film grain. Preview and export are chosen separately, and both are sticky.
Bayer and X-Trans RAW only: a scanner TIFF, a Pakon scan or a linear DNG arrives already de-mosaiced. Only the algorithms your LibRaw build compiled in are listed.
- Preview / Export (default Auto for both): Auto keeps NegPy's own choice, a fast half-size decode on screen and AHD for export. For the preview, Auto and Linear are the fastest; the others decode at full size. AHD is LibRaw's balanced default, VNG the smooth one, PPG fast with clean edges, DCB and DHT chase fine detail, and AAHD softens edges to suppress artifacts.
How the negative is measured and normalized into a positive. The film mode that decides which conversion runs sits above the panels (§4), and how the scan is decoded lives in Calibration (§4.1).
-
Multi-core CPU rendering (Preferences → Performance, beside GPU acceleration): spreads the CPU rendering kernels across your cores. It takes effect immediately, with no recompile and no restart.
Be realistic about the gain. The kernels run much faster, but a merge is dominated by decoding the RAW files, which this does not touch, so the whole operation comes down by only about a tenth. Ordinary editing changes less again, because the GPU already carries the pipeline. The gain is largest wherever the CPU does the work: merges, exports, and any machine without a usable GPU.
On Windows and Linux this is on. On macOS it is off, pending more evidence: the underlying threading layer terminates the process outright if two threads enter it at once, and while NegPy serialises every such call behind a lock, that has been proven on one Mac rather than on the range of them. If you turn it on and the app ever closes without warning, NegPy notices on the next launch and offers to turn it back off; that is the failure to expect, and it is recoverable. Setting
cpu_parallelunder[performance]inoverride.tomlstill wins over Preferences, for a machine that cannot start.
Analysis window, where NegPy measures the black and white points. The slider takes half the row, the three buttons the other half:
- Analysis Buffer (0.0 to 0.25): insets the measurement window from the frame edge so film rebate, sprocket holes and scanner borders do not skew detection. Raise it on scans with wide borders.
- Analysis Region (square-draw tool): draw a freehand region on the canvas to meter exactly that area, overriding the buffer. Double-click inside to confirm; the ✕ button clears it.
- Lock Bounds (padlock): freezes the analyzed normalization bounds for this frame, so cropping or moving sliders no longer re-analyzes it. Lock it in once you are happy with the bounds.
Normalization tuning:
- Luma Range Clip (-100 to 100): how aggressively the tonal range, the black/white-point span, is set. Neutral already applies a small robust clip. Positive tightens it, which is good for dense or fogged negatives where a few stray pixels would push the bounds to extremes. Negative pushes the bounds outward, for lifted blacks and unclipped highlights.
- Color Clip (-100 to 100): the per-channel color-balance clip (orange-mask removal), independent of the tonal range. Positive tightens channel balance; negative samples nearer the extremes.
- Global / R / G / B selector → White Point / Black Point (-0.25 to 0.25): manual offsets on top of the auto-detected bounds. A positive white point brightens; a positive black point lifts blacks. In R/G/B mode these become per-layer trims: per-dye-layer film-base (Dmin) and Dmax corrections, which is scanner-style per-channel levels. The selector is hidden in B&W Negative, where per-layer trims are meaningless, and in Transparency with Normalize off, where the sliders it scopes are hidden with the rest of the normalization tuning.
Crosstalk, Hue Trim and the sensor unmix all live in Calibration (§4.1). They correct the capture rather than the negative-to-positive conversion.
No Transparency matrix ships with NegPy. On slides the Matrix dropdown starts empty, and it and Strength are disabled until a matrix exists. The editor button stays live, so you can build your own: press +, and it is created for the process you are in. A
.tomlmarkedprocess = "Transparency"dropped into your crosstalk folder works too (the pre-renameprocess = "E-6"still loads). It means something different there: on a negative the dyes' unwanted absorptions are an error to remove before inversion, so unmixing moves the render toward the scene, but a transparency is the finished image, and what you see on a lightbox already includes those absorptions, so unmixing moves it away from the slide's own look. In Transparency, treat it as a color-separation control, not a fidelity correction. Hue Trim is unaffected: it corrects the light source, so it applies to slides exactly as it does to negatives.
Normalize (Transparency only) is the switch between two ways of rendering a slide.
-
On: auto-stretches the histogram to fill the dynamic range, metered per frame, and prints it through the paper model like a negative. This is a rescue tool for faded or expired slides, which is what it was added for. Where the dyes have lost their range, metering it back per frame is exactly right, and because the stretch is metered, two exposures of the same slide converge on a similar render.
On a slide that was exposed as intended, expect it to look washed out and desaturated. That is not a miscalculation: a slide's density runs all the way to Dmax, but only its top ~1.5 decades carry picture, so a stretch measured across the whole range squeezes the picture into the top of the print curve, where the shoulder compresses tone and color together. Reach for it when the slide needs rescuing, not as a starting point.
-
Off (default): renders the slide as captured. The camera's own color matrix is applied, and the tonal window is fixed to the decoder's white level rather than measured from the frame, so a slide opens looking the way it does in Photoshop, Preview, Affinity or Darktable, and a bracketed set stays a bracket, with each exposure rendering at its own brightness. Use this mode when you exposed the slide the way you wanted it and only need to adjust from there.
With Normalize off, the paper simulation has nothing to act on, so the print-specific controls are hidden (paper profile, Paper White/Black, Auto Density, Auto Grade, split grade, Dye Separation) along with the normalization tuning above, which only shapes a measured stretch. What stays is a plain transfer curve, neutral at its defaults: Print Density (exposure), ISO-R Grade (contrast), Toe / Shoulder and their Width sliders (shadow and highlight roll-off), Shadows Density / Highlights Density (§6.2), the per-layer R/G/B trims, and white balance. Lab, Toning and Finish work as usual.
On a merged bracket, Normalize is grayed out, whatever the switch said before the merge. The merge already decides where the tones land, and Render exposure picks which exposure it prints at; a metered stretch divides that choice straight back out, since below the anchor at which the frame stops clipping, moving it changes nothing. The two are not wanted together in any case, because Normalize rescues faded film and fading compresses the density range a bracket exists to capture. Unmerge the frame if you need the stretch.
Coming from Lightroom's basic panel, the map is: Exposure → Print Density (inverted, lower is brighter), Contrast → ISO-R Grade (inverted, 180 is softest), Shadows → Shadows Density, Highlights → Highlights Density, with Toe and Shoulder shaping how each end rolls off. The Density sliders take the darkroom sign: positive adds density, so negative Shadows Density is what opens the shadows. Whites and Blacks have no equivalent here, because the tonal window is fixed by design, which is what makes the render faithful to the capture.
A source with no camera matrix (a scanner TIFF, a JPEG) is already in the working space and passes straight through.
Linear RAW is grayed out in Calibration here, because it does not apply to an as-captured render: it decodes without the as-shot white balance, which the camera matrix assumes is present, and the multipliers are folded back in, so the render is identical either way. It stays visible, since it is a sticky setting and a hidden one is a setting you cannot see the state of. With Normalize on it is live again, that render being a metered stretch rather than a transfer, and it is live with Positive on too — a positive source has no camera matrix, so a checked Linear RAW would otherwise silently force literal linear data over Positive's own read of the file's profile. An explicit Input ICC in Export replaces the camera's own primaries rotation rather than stacking on top of it, since the ICC supplies its own. The as-shot white balance still folds in as usual.
Narrowband and Single-Shot Narrowband Calibration are grayed out for any transparency, Normalize or not; see Narrowband and slides. Reproducing a slide's appearance is a colorimetric problem, and narrowband illumination samples the spectrum at three isolated wavelengths, so the inter-band overlap the eye integrates is never measured, which is the same reason narrowband scans render oversaturated and hue-rotated. No input profile recovers what was never sampled, and the bundled one describes negative dyes besides.
Positive (default off) sits next to Normalize, slide-only in the same way, and live only with Normalize off. As captured, NegPy reads the source as literal linear data and renders it like a raw capture, with a fixed exposure lift and a filmic highlight roll-off. Turn Positive on when the source is already a finished image, such as a scanned print or an export from other software: NegPy decodes its embedded profile instead (an untagged file falls back to sRGB) and skips the lift and the roll-off, so the Print sliders shape the image directly. With Normalize on it grays out, that render decoding on the source's own profile anyway.
Meter the whole roll once and share the baseline, so frames from the same film match.
- Batch Analysis: scans every loaded file and computes a roll-average density and color balance, discarding outliers. Run it once after importing. (Tip: if you use Batch Autocrop, run it first, in Image only mode, so metering sees consistent crops.)
- Use Luma Average: this frame takes the roll-wide tonal range; color still re-derives per frame.
- Use Color Average: this frame takes the roll-wide color balance; tonal range still re-derives per frame. Enable both for a fully consistent roll; leave both off for per-image auto-exposure.
ROLL, to reuse a baseline across sessions:
- Roll dropdown + Load: apply a saved roll's bounds and balance.
- Save: store the current Batch Analysis as a named roll, useful when you shoot the same stock repeatedly.
- Delete: remove the selected roll (it asks first). The frames keep their current look; only the saved baseline goes.
Save and recall a complete edit, the full workspace, by name.
- Preset dropdown + Load: apply a saved preset to the current image.
- Name field + Save: store the current settings as a new preset.
- Trash: delete the selected preset.
Where the frame gets its final shape: what is inside the print, and whether it sits level. Most scans need a pass here even when nothing else is touched.
Crop:
- Ratio (default
Free): target aspect ratio:Free,1:1,3:2,4:3,5:4,6:7,7:5,65:24,16:9,16:10,11:8.5. There is one entry per shape, because the crop tool auto-orients to portrait or landscape as you drag. OnFreethe crop tool is unconstrained, and auto-crop takes the ratio from the film format it detects, so 6x6, 645, 6x7 and 35mm each keep their own shape. Pick a ratio to force every frame to it instead. - Detect (crosshairs): snap the ratio to the closest standard.
- Crop tool: draw a crop rectangle on the canvas. It opens on the crop already set, including one Auto found, so you can nudge an auto crop instead of redrawing it. Once you adjust it by hand, nothing re-detects over it. Reset clears it and turns auto-crop off.
- Guide: overlay a composition guide while cropping: Thirds, Phi Grid, Diagonals, Golden Triangles, Golden Spiral, Armature, Diagonal Method, Grid or Off. The redo button rotates guides that have orientations; the spiral has 8, the triangles 2.
Auto Crop, to detect the frame edge automatically:
- Mode: Image only (exposed area) or Film edge (full film, including rebate and sprockets).
- Crop Offset (-5 to 100 px): inset the detected edge inward. Positive trims more; negative bleeds slightly outside, for when detection clips too tightly.
- Rebate Trim (0 to 150%): how far into the detected rebate to cut. 0% stops at the film edge, 100% lands on the detected image edge, and above 100% bites into the picture to clear a stubborn white border. Image only mode; it applies to both Auto and Batch Autocrop.
- Auto: detect and crop this frame. Best on clean rebate. The crop is detected once and stored, so the export is framed like the preview. Mode, Ratio, Rebate Trim and the orientation re-detect; Crop Offset moves the stored crop without re-detecting.
- Batch Autocrop: analyze all visible landscape frames as a roll, using confident detections to calibrate weaker ones. A portrait frame among them takes no part in the roll and is cropped on its own, the same as pressing Auto on it. Where no frame in the roll has a readable film edge (film that overfills the sensor, leaving no scanner bed around it), every frame is trimmed from its own measured border instead. A roll of only a few frames, an RGB triplet among them, is too short to pool a border across, so it trims by the edges that read brighter than the picture. It runs in the background with progress and cancellation. Manual, Film-edge and ambiguous frames are left alone. Image only mode only.
- Mixing scans in one batch: allowed, but the roll is what makes batch worth running. A frame that reads its own film edge keeps its own crop, and the roll supplies only what that frame could not measure. A frame that finds no edge at all is placed from the roll instead, so it takes the roll's width and tilt. That is the rescue on a consistent roll, and the risk in a selection of unrelated scans. Frames from one camera, holder and format pool best; mixed formats are safe while each frame reads its own edge.
- When auto-crop leaves a frame alone: the detector reads film against the light of the scanner bed, so it needs the bed to be the brightest thing in the scan. A slide with highlights as bright as the bed does not give it that, and the frame comes back uncropped rather than cropped to a guess. Sprocket-exposed film, and a neighboring frame filling more than a tenth of one side, read the same way. Crop those by hand, or use Film edge mode and trim in.
Alignment:
-
Fine Rotation (±45°): free rotation for tilted scans, in sub-degree steps (positive is clockwise). Applied after auto-crop so the frame stays axis-aligned.
-
Straighten tool (ruler): draw a line along a horizon or vertical edge and NegPy rotates to make it level or plumb.
-
Tilt (±15%): tip the easel about a horizontal axis to straighten converging verticals, the building that leans back because the camera pointed up. Positive stretches the top edge. The unit is per-cent of the frame, what you would measure on the easel, not a tilt angle: the same tilt keystones differently at every enlargement.
-
Swing (±15%): the same movement about a vertical axis, for converging horizontals. A wall shot from one side, or a copy stand not square to the film. Positive stretches the left edge.
Both replicate a wedge along the squeezed edge, as Fine Rotation does; crop it off. Crop before correcting if you can, because the meters read the corrected frame: on an uncropped scan a big correction pulls rebate and surround into the metered area and the print darkens.
-
Distortion Correction (-0.100 to 0.100, in steps of 0.001): radial lens distortion. Positive corrects barrel, negative pincushion. Use the film rebate as a straight-edge reference. Corrected before Tilt and Swing.
Corrects uneven illumination (vignetting or falloff) from your copy-stand or scanner light, using a reference shot of the bare light source.
- Profile dropdown, with + and trash beside it: pick a reference image and save it as a named profile. + reads the reference once and bakes its correction into the profile, so you can then move, rename or delete the original reference file without affecting your edits. The profile is self-contained, stored in NegPy's own
flatfieldfolder like sensor and crosstalk profiles. Trash asks first: the baked gain map cannot be recovered, and every frame using the profile loses its correction. - Apply Flat Field: apply the active reference to this image, enabled once a profile exists.
This is the heart of the print. Three panels shape light, color and contrast, and everything here happens in the "print" stage of the pipeline.
Color timing, like the dichroic filters on an enlarger head. A Global / Shadows / Highlights selector scopes the controls to the whole image, or biases them toward low- or high-density tones.
-
Pick WB (eyedropper): click a pixel that should be neutral gray, and NegPy solves the CMY filtration to make it neutral in the selected region.
-
Roll Lock: re-aims each newly opened frame's temperature to the current target, preserving its own tint. A per-region lock for consistent warmth across a roll.
-
Reset (undo-arrow icon): return the selected region's temperature and CMY to neutral.
-
Temperature: a warm-to-cool lever driving the region's magenta/yellow pair; cyan stays put, as in a real darkroom.
-
Cyan / Magenta / Yellow (-1 to 1): the three filtration axes, Cyan↔Red, Magenta↔Green and Yellow↔Blue.
-
Cast Removal (0.0 to 1.0, color only): balances each color layer against the frame's own grays, so neutrals stay neutral from deep shadows through highlights. The applied strength scales with how many clean near-neutrals the frame has. On Color Negative it defeats the orange mask and starts at about 0.5. On Transparency it starts at 0 and corrects a faded slide's crossover, since a slide's cast can be the photograph, so you ask for it. Hidden for B&W Negative.
It is hidden in Transparency and B&W Negative, because the render ignores it there. What it defeats is the orange mask, a cast the manufacturer built into the film rather than part of the picture. A slide has no mask and its cast is the photograph, so solving for a neutral axis would strip out the light you shot in; a B&W negative has one emulsion and no channels to balance. For a slide's color, use Temperature and the CMY sliders above, or Hue Trim (§4.1) if an unusual scanning light has rotated the hues.
-
Ring-around (target icon, or
Shift+F): prints the frame as a 5×5 mosaic stepping 2cc at a time out to ±4cc on the magenta and yellow axes, so the direction of a color cast is visible instead of guessed. Each patch is a real render of the part of the frame it covers; click one to keep its filtration. The ladder is absolute and centered on neutral, so a ring printed off one frame compares to the next.Escapeor a second press clears it, and any edit drops it. See Rotating a proof below.
The paper's response. A Global / R / G / B selector at the top scopes most controls to the shared curve (Global), or to per-dye-layer trims for crossover correction, meaning casts that differ between shadows and highlights, which filtration alone cannot fix.
Automatic helpers, on by default. They do per-frame work so you do not have to, and turning them off lets the negative print honestly.
- Auto Density: meters each frame's midtone and anchors print brightness there, so dense and flat negatives land consistently.
- Auto Grade: picks the grade from each frame's textural density range, partially, so dense negatives stop printing over-contrasty and flat ones stop printing muddy; it then goes harder if needed so the darkest textured tones still reach a black (Shadow Reach in Set Targets), and burns the brightest textured tones just enough to keep them off paper white (Highlight Hold).
- Set Targets (sliders icon): tune the exact brightness and contrast the two helpers aim for. Applies to every frame and is remembered between sessions.
Test strip (grid icon, or Shift+T): prints the frame as a 5×5 grid, with Print Density rising left to right and ISO-R Grade softening top to bottom, so the diagonals read light-to-dark and soft-to-hard like a split-filter test strip. Both ladders are absolute and centered on their defaults, so the settings you already have are one of the patches. Each patch is a real render of the part of the frame it covers; click one to keep it. Escape or a second press clears it, and any edit drops it.
Rotating a proof: a patch shows only the slice of the frame at its own grid slot, so the part you want to judge is stuck at whichever rung sits over it. While either proof is up, the 90° rotate buttons and [ / ] turn the ladder instead of the image: each press moves the dense or hard end onto a different edge, and the axis labels follow. The image's own rotation is untouched, and turning is instant, because printing a proof assembles all four orientations at once. The orientation you land on is kept for the rest of the session.
Exposure:
-
Print Density (0.0 to 2.0): overall brightness, simulating enlarger exposure time. Lower is brighter, higher is denser.
-
ISO-R Grade (50 to 180): contrast, as a paper ISO-R value. R110 is about classic grade 2; lower R is harder (more contrast), higher is softer. In R/G/B mode a Grade trim rotates one layer's slope about the midtone.
-
Shadows Density (±0.9 ΔD) / Highlights Density (±0.5 ΔD): brighten or darken just the shadow or highlight zone, without reshaping the curve. Bounded by paper black and white, so a burn cannot exceed the print's limits. The ranges differ because density is logarithmic: the same ΔD reads far smaller near paper black than near paper white.
These two also work in Transparency with Normalize off, on the same tones (the centers are mapped by position on each curve's own scale, not by raw density), and there they are the only mid-sparing controls: Shadows Density opens the quarter-tone with the highlights unmoved, where Grade and Toe drag the whole scale with them and cost the highlights.
-
Shadows Grade / Highlights Grade (split grade, ±50 ISO-R): rotate contrast locally in the deep shadows or highlights, the digital equivalent of split-grade printing.
-
Contrast Mask (±0.5, hidden in Transparency): sandwich the negative with a blurred, low-contrast mask, as the darkroom does with a mask film and a spacer. Densities add, so the mask's own polarity decides which way the range goes, and its gamma decides how far. The value is that gamma, signed.
Positive is the ordinary masking case: a blurred positive, contact-printed straight off the negative, so it is dense where the negative is thin. That squeezes the range by (1 − gamma) and a harder grade then fits the paper, while the blur keeps fine detail out of the squeeze. Use it on a scene too contrasty for the grade you want, then bring Grade back down in R. Past about 0.4 a soft halo appears along strong edges, as it does on a masked print.
Negative is a mask of the same polarity as the negative, dense where the negative is dense, which stretches the range by (1 + gamma) instead. The broad tones expand while grain and texture stay put, where a harder Grade steepens both together. It also works on a negative too flat for Grade, whose slope has bottomed out. Past about −0.4 highlights start to clip, and the Analysis panel's Clipping row says so.
-
Mask Spacer (2 to 6%, dead without a mask): the clear sheet that holds the mask off the negative, as a per-cent of the frame. It sets the scale above which tones are masked, so it reads backwards from a blur radius: thick works on the broad masses only and leaves detail alone, thin reaches down into the detail and bites harder. Thin also lifts shadows sitting next to something bright and hazes them, which is the mask line a printer would see on the print. Below the minimum the mask stops being unsharp at all and becomes a plain contrast change, which is Grade's job. 4% is a conservative default.
Both controls read only your crop. A masking film is neutral, so there is no per-layer trim and both gray out in R/G/B mode.
-
Dye Separation (0.5 to 1.5, hidden in B&W Negative): saturation in density space. It pushes the print's three dye densities apart before the positive is decoded, in the same matrix the paper's own dye crosstalk uses, so it responds to the paper profile you picked and eases off automatically where the curve is already compressed at toe and shoulder, instead of forcing color into tones that have none left to give. Below 1.0 it pulls the dyes together toward neutral. 1.0 is off. Contrast this with Chroma in the Color tab, which scales color evenly after decode.
-
Separation Damping (0 to 1, hidden in B&W Negative): decides where the Dye Separation push lands, rather than adding a push of its own. At 0 every color gets the same treatment. Turn it up and muted color keeps the full push while color that is already saturated gets the opposite, so a hard push puts color into the tones that had none instead of driving the strongest colors until they flatten into a slab. Below 1.0 separation it mirrors: pastels go gray while the vivid colors survive. It is dead at Dye Separation 1.0, where the slider grays out, because it has no look of its own. This is not the same as backing Dye Separation off: a lower value takes color from everything, including tones that had little to start with, where turning damping up takes it only from the colors that already have plenty.
Paper Response, the characteristic-curve shape:
- Paper profile: a bundled darkroom-paper profile, RA4 color papers in Color Negative and tonal B&W papers in B&W Negative. It re-shapes the curve as a baseline; Grade, Density, toe and shoulder still trim on top. Neutral reproduces the defaults. Each B&W paper also carries its own lith color path, which the Lith panel picks up: Fomatone liths warm and colorful, while Neutral and Ilford Multigrade stay nearly colorless.
- Paper White: simulate paper base density, so whites print at about 0.93 instead of pure white, like a real print.
- Paper Black: show the paper's true, slightly milky Dmax instead of compensating it to pure display black. Off (default) applies black-point compensation so the adapted eye reads black as black.
- Snap (-0.5 to 0.5): midtone gamma, steepening or flattening the S-curve around the reference tone while paper white and black stay put.
- Toe (-1 to 1) + Toe Width (0.1 to 5): the shadow roll-off into paper black. Positive toe lifts shadows for a gentle film toe; negative deepens them and, with Paper Black off, makes exact black reachable. Width sets how far the knee reaches into the midtones.
- Shoulder (-1 to 1) + Shoulder Width (0.1 to 5): the highlight roll-off into paper white. Positive compresses highlights (film-like); negative extends them and risks clipping.
In R/G/B mode the sliders become per-layer trims on top of the global value, for that dye emulsion: Grade (±30 ISO-R), Toe / Shoulder (±1), Toe Width / Shoulder Width (±2), Snap (±0.5) and Dye Separation (±0.4).
Draw masks and lighten or darken just those areas. Three shapes, one per darkroom move:
- Draw Mask (the cut card): click to place vertices; double-click, press Enter, or click near the start to close the mask; Esc cancels. To edit an existing mask, select it in the list, then drag a vertex, click an edge "+" to add a point, or right-click a vertex to delete it.
- Oval (the hole in the card, or a dodging wand): drag out an oval. Three handles: the center moves it, the other two set each axis, so it can be stretched and tilted. It has a fixed three points, with no adding or deleting.
- Card Edge (the graduated burn): drag from the edge that gets the full exposure (solid line) to where it fades out (dashed). This is the printer moving a card across the paper: a sky burn, a corner held back. The gap between the two handles is the softness, so Feather does nothing on this shape.
Mask handles can go outside the picture, and a tilted Card Edge usually needs that: its line must start past the corner it burns, or the tilt cuts that corner off the full-exposure side. Drag into the gray area around the frame.
- Mask list: each mask shows its shape icon and Dodge (lighten), Burn (darken) or Grade (contrast only), with the values it carries. Click the shape icon to enable or disable the mask's effect on the render; it keeps its shape and values, and its row grays out. The eye toggles its outline; the trash deletes it.
- Each mask is tinted on the canvas so its extent is visible. Hold down Burn, Feather or Grade, or drag a vertex, and the tint drops away until you let go, so the value is judged on the picture rather than through the tint. Any mask that intersects the one you are working on drops its tint with it, since stacked tints are what hides the area worst; masks clear of it keep theirs. A Card Edge covers its whole side of the frame and an inverted mask covers its surround, so both count as intersecting anything in that area.
- Burn (-2 to 2 stops, default 0): print exposure for the selected mask, signed the way the rest of NegPy signs light on paper. Positive burns (longer exposure, darker paper), negative dodges (held back, brighter paper), the same direction as Print Density and the Finishing edge burn. A freshly drawn mask sits at 0, so it changes nothing until you give it a value.
- Feather (0.0 to 0.15): edge softness for the selected mask, as a fraction of the frame's short side. Inactive on a Card Edge.
- Invert: acts everywhere except inside the selected mask, so it is the card itself rather than the hole cut in it. Burn the surround and hold the face with one shape.
- Grade (-40 to 40 R): prints the selected mask at its own contrast, in ISO-R points off the frame's Grade, negative being harder. This is the darkroom's burn-in through the hard filter: burn a sky at −20 R and it darkens without the highlights beside it flattening; dodge a face at +15 R and the shadow opens without going chalky. The rotation happens about the region's own midtone, so a mask with Burn 0 and a Grade set changes only contrast, not overall density. Overlapping masks add their grades, and the result is clamped to the ISO-R ladder (R50…R180) like every other grade in NegPy.
Printing Notes (Export tab, or Shift+N) turns the frame into the printer's marked-up work print. Each mask is outlined and labelled with its number and its value in stops; a Card Edge has no outline, so it is marked as the side of the frame that gets the full exposure. A card in the corner carries the print recipe: paper, Print Density, ISO-R Grade (with the split-grade trims when they are set), filtration, toe and shoulder, Snap, edge burn, and the dodge/burn list.
Two conventions are worth knowing, both borrowed from the darkroom rather than from the sliders:
- Burns are hatched, dodges are left open. Shading marks where the paper gets extra exposure.
- The numbers are exposure, not brightness, the same convention the Burn slider uses, so a mask at +1.00 st is written
Burn +1. Values land on ⅓, ½ and ¼ fractions where they are close enough, and otherwise print as decimals.
A mask with a local Grade also carries the grade it actually prints at, not the trim: a burn of +1.00 st at −20 R on a frame graded R115 is written Burn +1 @ R95, and a grade-only mask reads Grade @ R95.
Every mask is on the map, including ones whose outline you hid with the eye: that eye is there to unclutter editing, and a printing record that quietly omits a burn would be wrong. A disabled mask is the exception: it burns nothing, so it is left off the map and the recipe. The overlay steps aside while a test strip, either peek, the before/after baseline, or the crop and analysis tools own the canvas. Both the preview and its export live in the Export tab's Printing Notes section.
Mimics what a lab scanner (Frontier or Noritsu) does automatically. Color controls hide in B&W Negative mode.
Color (hidden in B&W Negative):
- Chroma (0.0 to 2.0): a color scale applied after the print is decoded, even across every tone, so it is a retouching move rather than a density-space one. 1.0 is unchanged, 0 is grayscale, 2.0 is double. For saturation that behaves like a print instead, reach for Dye Separation in the Exposure tab. Below 1.0 is a flat scale; above 1.0, pixels that would clip the display gamut get a soft per-pixel knee toward their own in-gamut headroom instead of a hard per-channel clamp, since clamping only the overshooting channels shifts the hue that the flat scale itself preserves.
- Skin Protection (0.0 to 1.0, default 0.5): holds skin-hued color under a chroma ceiling so faces do not go sunburnt. Hue and lightness are untouched, and chroma is only ever pulled down, never added, so asking Chroma for 0 still gives you grayscale. It is independent of Chroma and works with it at 1.0: skin that arrived over-saturated from the print curve or the filtration gets reined in just the same. Higher values lower the ceiling: the 0.5 default catches only genuinely excessive chroma, 1.0 leaves skin matte, 0 is off. The mask is warm hue and skin's own chroma and mid lightness together, which is what keeps a red coat, a saturated sunset, brick or autumn color out of it. What it cannot separate is warm objects sitting at the same chroma as skin (bare wood, tan leather, sand), which soften along with it. The same bound cuts the other way: skin that arrives really excessive, a sunburn, is only partly caught, so reach for Chroma or the Filtration panel for that.
- Chroma Denoise (0.0 to 5.0): smooths color noise, especially in shadows, while leaving luminance grain intact.
Sharpen:
- Method: Unsharp Mask (boosts edge contrast) or Deconvolution (Richardson-Lucy, which reverses the scanner's optical blur; set Radius to the scan's blur width).
- Sharpening (0.0 to 1.0): amount, on the L (lightness) channel so there are no color halos.
- Radius (0.5 to 3.0 px): blur width in output pixels, small for fine grain and larger for soft scans. Sharpening acts on the pixels of the exported file, so a fit-to-window preview shows less of it than the export carries; judge it at 1:1 with the loupe or at 100% zoom.
- Masking (0.0 to 1.0): restrict sharpening to edges, which protects flat areas like sky, skin and grain. Independently of it, both methods sharpen the deepest shadows at a third of the amount, since that is where negative grain is coarsest.
Detail:
- CLAHE (0.0 to 1.0): local contrast without blowing global highlights or crushing shadows. Use it sparingly, since near 1.0 can look cartoonish. It runs before dust removal, so healing operates on the final rendition.
Effects:
- Glow (0.0 to 1.0): lens bloom, where bright highlights scatter across all channels for a dreamy softness.
- Halation (0.0 to 1.0): the red glow of light scattering back through the film base. Highlights only, strongly red-dominant.
Two printing processes that are not ordinary silver-gelatin enlarging. Pick one with the None / Lith / Cyanotype buttons at the top, and only that process's controls are shown. Both are B&W Negative only, and both are off by default.
Lith printing is the darkroom process of massively over-exposing a lith-capable paper, developing it in a very dilute low-sulphite developer, then pulling it out part-way through. You get creamy warm highlights and an abrupt drop into hard, sooty blacks, with very little in between.
There is no color control here. The paper chosen in the Exposure panel sets the whole path, from peach highlights through an olive transition to neutral blacks. Neutral and the Ilford papers lith almost colorlessly; Fomatone is the one that gives you the peach and the olive.
While Lith is selected, Sepia, Iron Blue, Copper and Vanadium gray out in the Toning panel, since they do nothing distinctive on a lith print. Selenium and Gold stay live, and behave differently here (see 7.3).
- Exposure (0 to 5 stops, default 2): print over-exposure. Real lith printing runs on two to four stops more light than a normal print. More light means warmer, more colorful highlights and softer gradation.
- Snatch Point (0.0 to 1.0, default 0.55): how long the print stays in the developer before you pull it. Higher drops the point where the shadows go black further up the tonal scale, giving deeper, colder blacks and a wider band of undifferentiated shadow. Lower keeps the print high-key and warm, with weak blacks.
- Abruptness (0.0 to 1.0, default 0.6): how suddenly the shadows go black. High turns the transition into a step, so the next zone down blocks up with no separation left in it. Low leaves a gentle roll into the blacks. In the darkroom this is the hydroquinone-to-alkali ratio of the developer.
A cyanotype is contact-printed in UV onto paper brushed with iron salts. There is no development to time and no silver anywhere in it: the image substance is Prussian blue, which absorbs red light around 700nm, so the print never goes black, it goes blue. Highlights come out green, where the blue mixes with the yellow sensitizer left in the paper.
Two things make it look unlike an enlargement. It holds only a short density range, so a negative that prints normally on paper clips at both ends. And it compresses the midtones, so the middle of the scale is flatter than the ends.
While Cyanotype is selected, every chemical toner grays out in the Toning panel, because there is no silver for those baths to react with. Use Bleach and Tannin instead. Split toning still works.
- Sensitizer (Classic or New, default Classic): Classic (Herschel) is the original ammonium ferric citrate mix. It loses a lot of its pigment in the wash, so it tops out at a fairly light blue and keeps a strong green stain in the highlights. New (Ware) is the ferric oxalate formula, which holds far more pigment through the wash, so it goes much deeper and cleaner.
- Exposure (-2 to 4 stops, default 0): time under the UV source. More light drives more of the scale into blue; less leaves the print pale and high-key.
- Exposure Scale (0.8 to 2.8 log D, default 1.4): the negative density range the sensitizer can print, which is the contrast control. Ware measures about 1.0 to 1.2 for the traditional formula against 2.4 for the new one, and his Simple Cyanotype comes in variants at 1.8, 2.3 and 2.7. A short scale gives a contrastier print that clips both ends.
- Bleach (0.0 to 0.5, default 0): washing soda. It strips Prussian blue out of the print, highlights first. Take it far enough and only the deepest shadows keep any pigment.
- Tannin (0.0 to 0.5, default 0): tea, coffee or tannic acid. It re-develops the bleached iron as a brown iron tannate, which covers more than the pigment it replaced, so the print goes browner and a little deeper. Bleach first for a full brown; use Tannin alone for a split blue-brown.
Color the print itself rather than the scene: chemical toners that convert the silver (B&W Negative only), and a split tint that works in any mode. Lith silver is much finer than normal print silver, so the toners bite harder and differently on a lith print. With Lith on, only Selenium and Gold stay enabled; with Cyanotype on all six gray out, because there is no silver in the print at all.
Chemical Toning (B&W Negative only), simulated as sequential toner baths, in the order shown, each strength 0.0 to 2.0:
- Selenium: deeper blacks, cool eggplant shadows. On a lith print it reaches much further down the scale, lifts Dmax hard and turns the green-black shadows magenta.
- Sepia: warm highlights first; partial strength gives split-sepia.
- Gold: cool blue-black on untoned silver; over sepia it shifts highlights orange-red. On a lith print it works on every density evenly instead of the highlights first, and pushes the print towards blue-violet.
- Iron Blue: Prussian-blue shadows deepening to navy blacks.
- Copper: pink to brick-red shift, with the classic Dmax loss.
- Vanadium: greens the mids and highlights while deep shadows keep their black.
Split Toning (all modes), an additive tint in Lab space, so grain and detail are preserved:
- Shadow Hue (0 to 360°) + Shadow Strength (0.0 to 1.0).
- Highlight Hue (0 to 360°) + Highlight Strength (0.0 to 1.0).
Spotting, the way it was done with a brush on the finished print. There are three ways to find the marks (local contrast, the scanner's IR channel, or by hand) and they stack. However a mark is found, it is repaired the same way: the film under it is rebuilt from the clean film around it, with the frame's own grain transplanted back, and anything too wide for that goes to a fill that follows the structure through.
An Overlay button cycles the detection overlay (Off → Marked → IR) so you can see what is being caught: green for what Optical Removal found, magenta for IR and for defects sent to the structure-following fill.
Optical Removal finds specks and hairs on the visible scan by local contrast, with no IR needed:
- Toggle Optical Removal on, then set Threshold (0.01 to 1.0; lower catches more, at the risk of false positives; above the default the bar rises faster, so the top end leaves sharp highlights and dense lines alone) and Size (3 to 8 px; max spot radius). A mark covers the whole speck or hair, not just its darkest point. The bar is measured against the film's own grain, so the same setting means the same thing on any scan, and it rises inside busy image detail, where a thin dark line cannot be told from a hair; dust on textured film may need a lower Threshold, or the IR or Heal tools.
IR Removal uses the scanner's infrared channel to remove dust invisible to the color dyes, and is enabled only when the scan carries an IR plane.
- Toggle IR Removal and set IR Threshold (0.05 to 0.95; lower catches more).
- Method picks how the film under a defect is rebuilt. Both use the same IR plane and the same threshold slider.
- NegPy (default) divides semi-transparent dust back out, fills opaque cores with a weighted average of the clean film around them, and transplants grain from the nearest clean pixel.
- OpenICE works in log density and restores detail rather than averaging it away: at each scale it adds back the picture detail that beats the infrared's own contrast at that scale, so texture under a speck survives. Where a defect was solid there is no detail left to restore, so the repair gets Digital ICE's own synthetic grain instead, strongest in the midtones and fading out at both ends of the scale. It measures clear-film level and dye-to-infrared crosstalk from each frame, and leaves film it judges clean untouched bit-for-bit. Better on fine detail and gentler elsewhere in the frame, but less proven across scanners, so try both on a frame you know.
- The IR plane is read from 4-channel TIFFs and DNGs (VueScan, NegPy's own scanner output), SilverFast's iSRD TIFFs and 64-bit HDRi RAW DNGs, and
_IR.tifsidecars. Scan to HDRi, not plain HDR, if you want IR data in the file. B&W and Kodachrome block infrared like dust does, so those frames are skipped automatically.
Manual Heal (the header shows the current spot count):
The brush marks a search area, not a stamp: only the pixels that actually stand out from the film around them are rewritten, so you can paint generously over a speck and the clean grain inside the brush is left exactly as it was. Marks are caught in both directions: dust, which prints light, and scratches, which print dark. If the brush finds nothing wrong, it does nothing.
-
Heal Tool: click dust spots in the preview to paint them out one at a time, or drag to paint over a run of them.
-
Scratch Tool: click points along a scratch or hair, then double-click or press Enter to finish. Esc cancels, Backspace removes the last point. Right-click an overlay to delete it.
-
Transport Line: for the long straight marks film picks up running through a camera or lab, the ones that cross the whole frame, usually in the same place on every shot of the roll. Click once anywhere on the scratch and the whole line is traced and repaired; there is nothing to paint or drag.
These are the marks the brush is worst at, and not for want of care: spread along its length, a transport scratch is far too faint to pick out from film grain at any single point. The line tool reads the evidence along the whole scratch at once, which is what makes it visible at all. It follows the scratch's own angle (film is rarely square to the sensor), widens the repair to match the scratch, and covers only the stretches where the scratch is actually present, so one that fades in and out is left alone where it fades. If a click finds nothing, it says so rather than touching the frame; click directly on the line.
Hovering shows a guide: the line that would be traced and the band it would repair, before you commit. Committed lines stay drawn the same way, so you can see what each one covers; right-click one to delete it.
-
Line Sensitivity (0.05 to 0.95, shown while the Transport Line tool is active): how readily a scratch is followed. Lower catches fainter lines and repairs a wider band; raise it if a line starts picking up film either side. It applies to lines already placed as well as new ones, so you can trace first and tune after.
-
Brush Size (2 to 16 px): diameter of the manual brush, matching the on-screen cursor, shown while a heal or scratch tool is active.
-
Undo Last / Clear All: remove the most recent or all manual heals and traced lines; auto-detected dust is unaffected. Right-click a line to delete just that one.
How the print is presented: edge burn, a filed-out carrier's black rebate, and the paper margin around it. Applied at the very end of the pipeline, after everything else is settled.
Vignette (printer's edge burn, in stops):
- Burn (-2.0 to 2.0 stops): positive darkens the edges, negative holds them back and lightens. 0 is off.
- Size (0.0 to 1.0): falloff radius. Small keeps it tight in the corners, large spreads it into the frame.
- Roundness (0.0 to 1.0): 0 is radial (lens-like), 1 is a rectangular card burn following the print edges.
Filed Carrier, a filed-out negative carrier: the clear rebate prints max black, framed by a margin of unexposed paper.
- Width (0.0 to 5.0 mm): black rebate frame thickness. 0 is off.
- Roughness (0.0 to 1.0): how raggedly the aperture was filed, on the paper-side edge of the black frame. The picture-side edge is the camera's film gate and only ever wobbles slightly.
- Flare (0.0 to 1.0): light reflected off the bared metal of the filed bevel, a glow that lifts the black just inside the filed edge and stains the paper just outside it. Colored on color film, with the hue drifting along the edge, because the stray light never passes the orange mask; neutral in B&W. 0 is off.
- Corners (0.0 to 1.0): how far the aperture's corners round off, since no file cuts a sharp inside corner.
The paper margin takes the mat color, so it runs into the border with no seam.
Border:
- Width (0.0 to 2.5): border thickness as a fraction of the image. 0 is no border.
- Bottom Weight (1.0 to 2.0): thickens the bottom border, for window-mat proportions.
- Color swatch: click to pick any border color.
- Paper White: tint the border with the toned paper-white instead of the picked color.
The sliders you reach for most, gathered in one place, so a routine edit no longer costs a tab switch and a scroll. Empty until you fill it.
- Edit Favorites: opens a picker. Tick sliders on the left, drag them into the order you want on the right, then press Apply.
- The panel then shows those sliders in your chosen order. They are the same controls as in their home panels, so moving one here moves it there and the other way round. Nothing is duplicated or moved out of its own tab.
- A favorite hides itself when its original does. Favorite a Filtration slider and it disappears while you are in black & white, where it has nothing to act on.
- Your selection is remembered between sessions.
Two lists: the versions you chose to keep, above the running record of every change.
A work print is a named version of this frame, the darkroom habit of keeping the prints you made on the way to the final one, so you can go back to the third attempt after deciding the fifth went too far.
- Save Work Print (Ctrl+Shift+S) keeps the current edit under a name; NegPy offers Work print 1, Work print 2 and so on. Saving over an existing name asks first.
- Click one to make it live. That counts as an edit, so Ctrl+Z puts back what was on screen before; you cannot lose your place by looking at an old version.
- Right-click for Export This Version…, Rename… or Delete. Delete asks first, and a rename to an empty name is ignored.
Work prints differ from history steps in the way that matters: they are never pruned and never thrown away by a later edit. The undo history keeps the last 100 steps and drops the branch above you when you edit after stepping back; a work print survives both. The list appears only once you have saved one.
They belong to the frame, not to your presets: a preset is a look you apply to other images, a work print is one version of this print. Both live in NegPy's database; work prints are not written to .negpy sidecars.
A scrollable list of every edit step, the last 100 kept, newest on top. The current step is bold.
- Click a step to jump to that state.
- Right-click → Export This Version… to export a past state directly.
-
Print (default): the full creative look you see on screen.
-
Flat: a flat, neutral, low-contrast master that keeps maximum tonal and color information for editing elsewhere (Lightroom, Darktable, Photoshop). It skips the print look, effects, toning and vignette, and writes a 16-bit TIFF, or a lossless JPEG XL when JXL is selected and the color space is taggable (sRGB, P3, Rec 2020 or Grayscale). Your in-app preview is unaffected.
- Preview Flat: temporarily show the flat master on the canvas without changing your edit.
- Roll Baseline: measure every visible frame and share one exposure baseline, so flat masters are consistent across a roll. Recommended before a flat batch.
-
Linear: bypass the entire darkroom pipeline and dump the scanner's or camera's decoded buffer as a linear 16-bit file. The output format is selectable: TIFF (default, zlib-compressed, genuinely untagged) or JPEG XL (lossless). JPEG XL has no untagged state, so it comes out asserting sRGB primaries and a linear transfer regardless, which is not true for camera or scanner-native primaries; use TIFF if an unasserted file matters. An Effort slider (1–9, default 7) controls JPEG XL encoder speed against compression. No normalization, exposure, color management, flatfield or sensor correction, just the raw data with lossless geometry (rotation and flip) applied. Supported sources:
- Pakon RAW: 4× expansion by default (14-bit sensor range scaled into 16-bit). F335 files (16-bit sensor) default to no expansion.
- LinearRaw DNG: SilverFast HDRi (3-channel) and VueScan (4-channel RGB+IR). IR is written as a separate grayscale file with an
_irsuffix, in the same Format you chose for the main output. - Camera RAW: demosaiced with unity white balance (1,1,1,1), using the algorithm chosen for Export in the Demosaic panel (§4.2); the resolved algorithm is recorded in the description. The camera's as-shot WB is written into XMP (
RAW-WB: R G B) so downstream tools can apply it. Source device and timestamp are preserved. Trichrome triplets (narrowband R/G/B exposures) are merged into a single combined TIFF. Stitch composites are assembled with flatfield and sensor correction applied per-part for clean seams; stitch and triplet combinations are also supported. - Coolscan NEF: Nikon Coolscan scanner files. Despite the name, these are not raw sensor data: the content depends on the Nikon Scan settings used at scan time, so linear, unprocessed output needs the right settings before scanning. The full-res RGB SubIFD is read directly, and any extra channels beyond RGB are dropped, since Coolscan has no separate IR channel. No expansion.
- Flextight FFF: Imacon/Hasselblad Flextight scanner files, including both standard uncompressed 16-bit RGB exports and SGI LogLuv compressed raw files (
.3fr/.fff). LogLuv files are decoded through a LogLuv → XYZ → linear sRGB pipeline with per-channel percentile normalization; LogLuv is HDR, so normalization is part of the decode, and without it the data would be truncated, not raw. The largest image IFD is selected by pixel count. Data is linear scanner transmittance. Embedded FlexColor metadata (film stock, film type, scan date, scanner serial) from the proprietary plist (tag 50457) and the firmware blob (tag 46279) is carried through to the output TIFF headers. No expansion. - Noritsu RAW: headerless BGR 16-bit scanner dumps. Frame dimensions are auto-detected from file size against known Noritsu scan dimensions. 16× expansion by default (12-bit sensor data in 16-bit range).
- TIFF: generic scanner TIFFs. If the file has a 4th channel tagged as IR (ExtraSamples = UNSPECIFIED or missing), it is written as a separate
_irfile in the same Format as the main output. Sidecar IR files (_ir.tifnext to the source) and IR stored in secondary TIFF pages are also detected. Input gamma lets you select the gamma encoding of the source (linear, 1.8, 2.2 or sRGB) so the data can be linearized before export. Expansion is available, off by default. - Expansion: scales the linear data before writing. The combo box shows source-appropriate options: Pakon F135/F235 default to 4×, Noritsu to 16×, F335 and LinearRaw DNG to off. Camera RAW, Coolscan NEF and Flextight FFF have no expansion option. Leave it at the default unless you know why you need to change it.
- Apply ICE dust removal (visible when an IR channel is available): applies IR-based dust and scratch correction to the linear output before writing. Off by default.
- Corrections (camera RAW only): three optional toggles that bake corrections into the linear output before writing. All default to off, following the raw-dump philosophy. Apply white balance multiplies by the as-shot WB gains; it grays out for a Trichrome triplet or a Single-Shot Narrowband capture, since as-shot gains have no practical use against a narrowband exposure. Apply flatfield applies the flatfield gain correction. Apply sensor correction applies the sensor crosstalk unmixing matrix. For stitch composites, flatfield and sensor correction are always applied per-part regardless of these toggles, because clean seams require it.
Linear Output writes where the Destination section says, the same as a print or flat export: folder mode, subfolder, export path and Filename Pattern all apply.
_linearis always appended to the rendered filename, so a dump written next to its source cannot overwrite that source. Without Overwrite, an existing file makes the next one_linear_2,_linear_3and so on.Linear Output runs in the background like any other batch: the progress popup shows which frame is being written, Abort stops it after the current one, and the finish message counts any frames that failed.
The output file is always written clean: no ICC profiles, no EXIF color space tags, no XMP color metadata from the source. Both formats carry raw pixels plus device metadata (Make, Model, DateTime) from the source file, and a description recording the source format, expansion, the demosaic algorithm (camera RAW only), white balance and any corrections applied, including ICE. JPEG XL also carries the forced color tag noted above, which is not from the source; the format cannot leave it unset.
The primary Export action. Its chevron menu picks the scope: current frame (Ctrl+E), selected frames, or all visible frames. Every scope uses the settings below. To deliver the same frames in more than one format or size in a single run, use Export Presets.
- Format:
JPEG,TIFF,PNG,JPEG XL, orWebP, with quality or effort options per format. JPEG XL supports onlysRGB,P3 D65,Rec 2020orGrayscalefor Export profile: it tags color with compact enumerated values rather than an embedded ICC profile, and NegPy's JXL encoder cannot carry an arbitrary one, soAdobe RGB,ProPhoto RGBand a custom Output ICC are rejected with an error. Pick a supported space or a different format. - Bit Depth:
8-bitor16-bit, for TIFF, PNG and JPEG XL. JPEG and WebP are 8-bit formats and hide the row. A flat master is always 16-bit and hides it too. - Compression (TIFF):
Uncompressed,LZWorZIP. All three are lossless; ZIP is usually the smallest. - Compression (PNG):
0–9. Lossless either way: higher is slower and smaller. - Progressive (JPEG): render the image in passes while it downloads.
- Input ICC: treat the source as this profile, for when a scan's profile is known but untagged. It overrides primaries only: the tone curve is always the pipeline's own, so a matrix-style profile's declared TRC is ignored (two profiles with identical primaries but different TRCs render identically); a LUT-style profile's own input curves are still honoured.
- Export profile: what the exported file is converted to and tagged with. Either a color space (
Same as Source,sRGB,Adobe RGB,ProPhoto RGB,P3 D65,Rec 2020, orGrayscalefor true B&W output), or an imported ICC profile for a printer or paper.Same as Sourcenames the space it resolves to for the open frame; for an untagged scan that is the Adobe RGB the pipeline encodes to. An imported profile is not a preview-only proof: it is converted to and embedded in the file. Not available for JPEG XL output; see the Format note above. - Import ICC (the folder button on the COLOR MANAGEMENT header): copy a
.icc/.icminto~/NegPy/icc/, where it joins both lists at once, with no restart. A profile named after a built-in space (sRGB.icc) replaces that space's profile everywhere rather than appearing as a separate choice, and asks first. - Proof on screen lives in the Soft Proof section below, with the intent and paper-simulation controls. A warning appears here when the preview cannot predict the exported colors, either because nothing is being proofed or because the proof is aimed at a different profile than the export writes.
- Paper Aspect Ratio: final print ratio, or Original (no resize).
- Resolution: Original (full RAW resolution), Print (long-edge Size in cm plus DPI), or Pixels (long-edge px; the short side follows the paper ratio). Every format is tagged with a resolution, so a print or layout tool opens the file at the intended size. Print uses the DPI you set and Pixels the DPI its own long edge implies; Original resamples nothing, so it keeps the source file's own resolution, read from its EXIF or from the file's own record such as a JPEG's JFIF density, and falls back to the DPI field only when the source declares none. Linear output follows the same rule.
- Destination: Filename Pattern (a Jinja2 template with export settings plus Metadata fields such as roll, camera and film; see TEMPLATING.md), an Overwrite toggle, and the output location (subfolder of source, same as source, or an absolute Export Path with a browse button). Destination applies to all three output intents: with Linear selected, Format, Size and Color Management hide (a raw dump has no use for them) and Destination stays.
A checklist of export presets, each a saved Format/Size/Color Management/Destination/filename recipe. Manage edits them; Export Presets renders the frames with every enabled preset at once, and each preset uses its own destination, not the sidebar Destination above.
The printer's record for this frame: the dodge/burn map with every mask outlined and numbered, and a recipe card with paper, density, grade, filtration and the burn list. Preview shows it on the canvas (also Shift+N); Export writes it as an image beside the print. The conventions it draws with are described under Dodge & Burn.
Save on export writes a .negpy edit sidecar next to each source on every export; Export Sidecars writes them for all visible frames now, and reports how many failed if a source folder is read-only. Edits always stay in the database too; sidecars are optional archival copies.
Render all visible frames into a single sheet. Choose a Template or set Cell / Gap / Margin / Max tiles by hand, pick an output Path, then press Export Contact Sheet. The sheet is a JPEG at the JPEG Quality and Progressive settings above.
Simulate the print on screen. See below.
These controls change the preview only. The exported file is never proofed.
A soft proof shows what the picture becomes when a given printer puts it on a given paper. Paper is dimmer than a screen and makes fewer colors, so a fair proof looks worse than the unproofed preview. Judge one in the room's own light, and give your eyes a moment to settle.
- Preset: a saved printer and paper set-up. Picking one restores the profile, intent and toggles together, so glossy to matte is one click. None is the baseline: proof whatever the export targets, simulate no paper. Save names the current set-up, the bin removes the selected one. The box shows which preset your settings match, and goes blank once you change one, so it never names a set-up you have edited away from. These are separate from the Presets section above, which saves delivery recipes for the file.
- Proof on screen (
Shift+P, on by default): the master switch. Off, the preview is your edit at full gamut and the rest of the section grays out. - Profile: what the preview is proofed through. It follows the Export profile until you name a printer or paper here, so you can proof a print while exporting a web JPEG. Only imported ICC profiles are listed. Import one with the folder button on the COLOR MANAGEMENT header.
- Intent: how colors the paper cannot make are fitted into the ones it can. Relative Colorimetric leaves every printable color where it is and clips the rest to the edge of the paper's range, which is accurate until a saturated area flattens into a patch. Perceptual squeezes the whole picture inward so the relations between colors survive, at the cost of moving colors that would have printed fine. Printer profiles carry their own table for this, so it is worth trying on a saturated frame. Saturation favors vividness over accuracy and suits charts, not photographs.
- Black point compensation (off): scale the darkest tone in the picture onto the darkest the paper can make, instead of clipping everything below it.
- Simulate paper white (off): show the paper's own white instead of the screen's. The picture dims and picks up the paper's tint, which is most of why a print looks flatter than a screen.
- Simulate ink black (off): show the paper's deepest real black instead of mapping it to the screen's. Shadows lift and lose separation, as they do in the print.
- Gamut warning: flatten every color the profile cannot print to gray, so unprintable areas are visible rather than only counted. The Analysis panel's Gamut row counts the same pixels. The mark fades over a short distance instead of stopping at a hard edge, so read it as a region rather than a pixel mask.
- Display: the monitor profile the preview is shown through, auto-detected. Set it by hand if detection fails. It is the other half of the proof chain.
Archival metadata for the original analog capture (camera, lens, film, process), written into exported files as EXIF and embedded XMP, so DAMs like Lightroom show your film gear rather than the scanner.
Every export format carries it: JPEG, TIFF, PNG, JPEG XL and WebP. A TIFF holds the capture position in XMP only, and EXIF text is 7-bit, so typographic punctuation is transliterated (4×5 is written 4x5).
- Protect original metadata: copy the source file's EXIF/XMP to exports unchanged, adding nothing. When it is on, the fields below are ignored and the source's resolution is copied exactly: the same numbers, axes and unit, whether the source states it in EXIF or in its own header, even where the export was resized. A source that declares no resolution stays that way in every format that can leave it out. TIFF cannot, so it states the export's own resolution rather than the unit-less value readers report as 1 DPI.
- Sync custom metadata to all files in batch export: batch and preset exports write this frame's capture, gear and process values to every file, instead of each file's own.
A saved set of metadata values, stored in ~/NegPy/presets/metadata/, separate from the edit presets on the Setup tab:
- Preset + Load: write the selected preset's fields onto this frame. Only the fields the preset stores change; everything else on the frame stays. Hover the field for a list of what a preset holds.
- Manage…: the library, with a page each for Cameras, Lenses, Film Stocks, Process, Scanning and Presets. A Process entry is a development recipe (developer, dilution, push/pull, time and temperature); a Scanning entry is a digitizing setup. On the Presets page, + stores the current frame's metadata under a name you pick, the pen renames a preset or changes which fields it stores, and copy and trash duplicate and delete. The fields a preset stores are then editable in place: swap its camera, lens, film stock, saved process or saved setup, or retype a developer, dilution, push, time, temperature, scanning note, roll or exposure. No frame needs to be open. Picking from the library refills everything read from it; typing over a filled value unlinks the pick. A stored capture date, place, description-field set or flag is shown but not editable here, being a per-frame decision. Notes is free text. Starter data seeds into
~/NegPy/gear/on first launch.
Gear travels as one unit: camera, lens, film stock, the film format and every other value read from them. So a loaded preset fills the dropdowns below and the exported EXIF with the same pick, and a preset for a 120 stock cannot leave the frame claiming 35mm. Picking a film stock sets the format either way, so set a frame format such as 6×7 after choosing the stock. The frame number is never stored in a preset.
Searchable; type in any field to filter the library.
- Camera / Lens / Film stock: pick from your library. Empty means not set. Clear empties all three.
- Date: when the frame was shot. Give only what you know:
1998,1998-07,1998-07-14or1998-07-14 16:30, with an optional offset such as+02:00. An impossible date turns the field red and is not saved. EXIFDateTimeOriginalpads the missing parts; XMPphotoshop:DateCreatedkeeps the truncated form andnegpy:CaptureDatePrecisionnames it. The scan file's own timestamp moves toDateTimeDigitized. - Place: the capture location. The map-pin button opens a map to search a place name, click a position or paste coordinates, the ✕ beside it empties the place, and the field itself accepts a pasted coordinate pair or an OpenStreetMap/Google Maps link. Coordinates are written to the EXIF GPS tags and XMP
exif:GPS*, the names to XMPphotoshop:City/State/Country; a TIFF carries the location in XMP only, and a place you set replaces the source file's GPS block whole, rather than leaving its altitude or heading beside your coordinates. A geotagged source with no place set here keeps its own coordinates on export, and the map opens centered on them. Where the frame was digitized is a starting view, never the capture place. Zoom the map with its + / − buttons, a scroll wheel or a trackpad pinch, and drag to pan. Opening the map contacts OpenStreetMap; typing coordinates needs no network.
- Saved process: pick a development recipe from the library to fill Developer, Dilution, Push / Pull, Time and Temperature. Typing over any of them unlinks it, so the picker never names a value that is gone.
- Format:
—(not set),35mm,120,4×5,8×10,110, orOtherwith a free-text field. - Developer and Dilution: the developer, for example
D-76, and its working strength, for example1+1,1+50orstock. The two join in EXIFImageDescriptionasD-76 1+1; the dilution also goes to XMP asnegpy:DevelopmentDilution. - Push / Pull:
Push +3…Normal…Pull -3. - Time and Temp (°C): development time as
9:30or plain minutes, and the temperature it ran at. An unreadable time turns the field red and is not saved. Both are written to XMP asnegpy:DevelopmentTimeandnegpy:DevelopmentTemperature, and searchable asdevtime:(minutes) andtemp:. - Clear: empties the saved process and everything it fills: developer, dilution, push/pull, time and temperature. Format stays, since the film stock sets it.
- Saved setup: pick a digitizing setup from the library to fill Scanning. Typing over it unlinks it.
- Scanning: scan method or notes. EXIF
Softwareis alwaysNegPy. - Clear: empties the saved setup and the scanning note. Roll and Frame stay, since the scan stamps them rather than the setup.
- Roll / Frame: Scanlight capture roll name and frame number, stamped automatically on capture and editable here. Available in export filename templates as
{{ roll }}and{{ frame }}, and written to XMP asnegpy:CaptureRollandnegpy:CaptureFramewhen set. Not the Roll Analysis normalization name.
Optional original shutter, aperture and ISO. Click the lock to edit a free-text string, for example 1/125s f/2.8 ISO 400.
A live view of exactly what will be embedded, grouped by capture, scan, process and file. The Scan group shows the source file's own timestamp and coordinates, so you can see what you are replacing. Description… opens a checklist of which fields join into EXIF ImageDescription. The defaults are camera, lens, film stock and ISO; format, developer, push/pull and scanning are off until you enable them. Confirming Description… sets that frame's selection and becomes the sticky default for other frames that do not have their own, so the last confirm on the roll wins. Sync metadata and Sync settings can also copy a frame's selection with the rest of the metadata.
When you set capture gear, it is written to standard EXIF, and the digitizing rig is preserved separately in negpy:Scan* XMP tags. Leave gear unset and your scanner or DSLR stays visible in EXIF instead.
Capture film directly into NegPy. Two collapsible sections.
Drive a film scanner. Choose a Backend: SANE (Linux/macOS; Coolscans and other SANE devices), Nikon Coolscan (nkscan) (a direct driver for Nikon Coolscans on Linux, Windows and macOS) or pyOpticfilm (Plustek) (OpticFilm 8200i SE and 8100 V2; Windows, macOS and Linux). Controls are grouped in the order you decide them: Film (what is on the film), Quality (resolution, depth, extra passes), Framing (which frames, and the window) and and Output (format, folder, filename template). A group's header disappears with the whole group when the device has nothing in it. Format writes TIFF or TIFF (mono), which is one 16-bit grey plane for film with a single record, such as a black-and-white negative. Frames takes the frames to scan as a list: 1-6, 1,2,5, or empty for every frame on the film. The strip preview writes its picks there, so a selection can be changed without previewing again. The line above Scan says what pressing it will do: how many frames, at what resolution, which extra passes and roughly how much disk it takes. Depth appears only when the device offers more than one bit depth, so it is hidden for the OpticFilm 8200i SE, which is 16-bit only. Autofocus and hardware Auto-exposure appear only when the connected device reports them, so typically on Coolscans and not on the OpticFilm 8200i SE. Prescan appears for devices that support a low-DPI full-window preview, such as the OpticFilm 8200i SE: run the preview, drag a crop rectangle, and the next Scan uses that hardware ROI. When the scanner exposes a scan-exposure-time option, as some genesys devices do, an Exposure slider appears; set it to override the scanner's default exposure time, and the value shows in µs, ms or s as appropriate. A device without the option hides the slider, so a saved value never breaks a different scanner.
pyOpticfilm (Plustek) notes: the OpticFilm 8200i SE (07b3:1825) and the 8100 V2 (07b3:1824) are scan-ready. Other OpticFilm models may appear in the device list but cannot scan until pyopticfilm marks them ready; on Linux and macOS, switch Backend to SANE if that backend lists the scanner. Use Prescan to grab a 1200 dpi full-window preview, set a crop, then leave with Apply Crop or Scan Frame. Either way the next scan reads that hardware ROI at the chosen DPI, not a software crop. Multi-exposure (8200i SE, 8100 V2; off by default) merges short and long color passes for more highlight and shadow detail; the long pass exposure is chosen per frame, and the scan takes longer than a normal pass. Scans from pyopticfilm 1.1.2 onward match SilverFast orientation; rescans older files if left-right matters.
With IR checked, color and infrared come back in one scan pass; pyopticfilm aligns the IR plane to the color frame. Color scans apply ASIC shading measured at home before the film feed, the same order as SilverFast, so the strip may stay loaded. The table is cached per DPI, so later scans only re-upload it.
The default Full window includes a little holder chrome top and bottom; host-path scans clamp those near-white margins to the film highlight so auto exposure is not skewed. Raise Analysis Buffer or crop if a frame still looks off. Autofocus and hardware Auto-exposure controls stay hidden, because the SE does not report those capabilities. On Windows, bind the device to WinUSB with Zadig before use, since the stock vendor or SilverFast driver conflicts. The driver is the optional pyopticfilm package: install it with uv sync --group plustek or pip install negpy[plustek]; Windows release builds bundle it. See PLUSTEK_WINDOWS.md.
Nikon Coolscan (nkscan) notes: the driver talks to the scanner directly, so it needs no SANE backend. It measures the loaded film instead of counting frames: Preview strip… reads the whole strip in one pass, finds every frame on it, and cuts every tile out of that same pass. The search starts when the dialog opens, and its result is kept until the film is ejected. Detect frames runs it again when the film has moved. The tiles appear as the frames turn up, and there is no preview resolution to choose. Check the framing before scanning; a measured boundary can be nudged with Offset (±10 mm, either way, since the frame is re-addressed rather than fed past) and Drift, and because the tile comes out of the strip pass, a nudge re-frames without going back to the scanner. Scan with nothing picked scans every frame on the strip, measuring it first if no preview has. To scan a subset, type it in Frames, or untick frames in Preview strip…. Each tile carries its own tick, All and None move the lot, and the count next to them says how many will be scanned. Either way the selection shows in Frames, and ejecting the film clears it, since the frames and their crops describe the piece of film that just came out. Offset and Drift survive an eject, because they register the transport rather than one strip. Four controls appear only on this backend:
- ICE: remove dust and scratches with the infrared channel while scanning. Permanent, because it is baked into the file, unlike the Retouch panel's IR Restore, which stays editable. Color film only: silver grain blocks infrared, so the mask on a black-and-white negative is the picture again. ICE and IR exclude each other, because they read the same pass: ticking one unticks the other. Tick IR to keep the plane and clean the file later in Retouch, ICE to have the scanner do it now.
- Samples: reads per line the scanner averages (1–16). Higher settings cut shadow noise and cost proportionally more time.
- Superfine: read one line per pass. Slower, and free of the line registration the faster three-line mode owes the host.
- Film: what is on the film. Color negative, B&W negative, Slide or Kodachrome. It decides three things: which way the frame boundaries read when the strip is measured, whether IR and ICE are offered at all (B&W and Kodachrome stop infrared with silver and dyes, so the mask comes back as the picture rather than the dust on it), and how the frame is metered. A color negative is metered one channel at a time, which takes the orange mask off before the converter instead of quantizing the blue record through it; every other film keeps its factory balance, because there the cast is the picture.
- Film format: the frame length on the loaded film (135, 66, 645 and so on). Leave it on Auto where the holder narrows it, and set it for loose film in a masked carrier. It appears only where the transport measures the film to find its frames. A holder with its own frame table fixes the format, so there is nothing to choose.
Every control here follows what the unit reports. An LS-50 shows neither Samples nor Superfine: it reads one CCD line at a time whatever you ask, and it ignores repeated reads of a line, so both stay hidden and a setting saved from another scanner is never sent to it.
The driver is the optional nkscan package (0.11 or newer), which ships as a wheel: If running from source install it with uv sync --group nkscan or pip install negpy[nkscan]. On Linux a Coolscan on USB needs a udev rule for Nikon (vendor 04b0), and one on FireWire/SCSI needs the sg kernel module.
SANE scan window: on a roll/strip feeder (a live frame count reported), Preview strip… previews every frame, sets a per-frame window, and picks which frames to scan. On a SANE device with a single manual holder and no feeder, the button reads Preview… instead: it previews just the current holder position and lets you drag one crop window, reused for the next scan (the pyOpticfilm backend's equivalent is Prescan, above). Either way, the window narrows the scanner's own hardware scan area, so the real scan only reads that region, rather than reading the full frame (holder margins and film rebate included) and cropping in software afterward.
A preview holds the scanner for the whole pass, so while one runs a progress bar tracks it, Cancel reads Stop Preview, which abandons the pass and keeps the tiles already in hand, and the Apply and Scan exits stay dark until the pass ends. Previews read the way the Film setting says: negative stock is inverted, Slide and Kodachrome are not.
DSLR or mirrorless copy-stand capture (macOS/Linux). Put the camera in PC Remote mode and plug it in over USB; NegPy detects it on its own. With a NegPy Scanlight connected it captures narrowband R/G/B triplets from saved film-stock presets; without one it takes a single white-light exposure. Captured frames land in the hot folder and flow straight into Trichrome Scan mode.
- Live View & Scan: frame and focus. Click the image to aim the camera's focus magnifier at that spot, click again for the full frame. In white-light and camera-only scanning, ISO, shutter and aperture are set live from the toolbar; a calibrated RGB preset locks them instead, so the scan cannot drift.
- Preset: a selected preset shows its RGB levels, ISO, shutter and aperture read-only, and the scan forces that exposure before every frame. + calibrates a new one: place the small rectangle on clear film base (the rebate between frames), name it and run it. It solves one shutter plus a per-channel LED level so each channel lands just under clipping; if the target is out of reach it stops and says which way to adjust, and saves nothing. Create a manual preset… unlocks the sliders and steppers to dial one in by hand.
- Scan and Retake: Scan shoots the frame into a per-roll subfolder, auto-numbered, and imports it; Retake re-shoots without advancing the counter. Delay between exposures pauses between the R, G and B captures for bodies that lock up when the next command arrives too soon.
- Narrowband: RGB-lit scans render more saturated than white-light ones. The Narrowband toggle in the Normalization panel corrects this with the bundled input profile.
Camera scanning needs the optional python-gphoto2 dependency (pip install gphoto2; no Windows build). CAMERA_SCANNING.md has the setup, the macOS camera-daemon note and troubleshooting.
Every preview dialog ends the same way: Cancel, then Apply (keep the framing and go back to the panel) and Scan (start the scan from here). The Apply button names what it keeps: Apply Framing on a strip, Apply Window on a single holder, Apply Crop after a Prescan.
- Cropping: drag on a previewed frame. A corner resizes, inside moves. Each frame keeps its own window, and Clear Crops drops the lot.
- Frame outline: a red box marks the detected frame on each tile. Offset, Drift and the per-frame slider are measured from it.
- Offset: slides every frame along the film to clear the inter-frame gap. Frames shift left as it grows, live. On a measured strip the tiles are cut again from the strip pass when the slider stops, with no new scan. The shaded band on the right is film past the frame boundary the transport cannot deliver, so offset past the gap costs frame tail. A feeder cannot back up, so there it only goes one way.
- Drift: adds progressively more (or less) offset per frame position, for a strip whose gaps creep along its length.
- Per-frame offset: the slider under each tile corrects that frame alone, on top of Offset and Drift. The tooltip shows its value; double-click resets it.
- Size: the tile size. The grid reflows to fit the dialog, with no new scan, and the size is remembered. Double-click resets it.
- Which frames: each tile carries its own tick; All and None move the lot, and the count says how many will be scanned. On a measured strip the ticks and crops describe the piece of film in the transport, so ejecting clears them, per-frame offsets included; Offset and Drift survive, because they register the transport.
Settings for the whole application, not for one photo. Open them from the canvas ⋯ menu → Preferences…, with Ctrl + ,, or on macOS from the application menu. Changes apply as you make them; the rows NegPy reads at startup say so and light a restart notice.
- UI scale (80% to 120%): scales the whole interface. Applies after a restart.
- Canvas background: four swatches (black, dark gray, mid gray, white) for the surround the image sits on. Mid gray is the neutral judgement surround; white reads like a print on a light table.
- Immersive canvas: the toolbar floats over the image instead of sitting below it, so the frame gets the whole canvas.
- Sticky zoom: keep the current zoom when you switch frames, instead of resetting to fit. Useful for checking the same magnification across a roll.
- Reverse scroll zoom: scroll up zooms out.
- Show slider values: keep every slider's value box open, instead of revealing it on hover.
- Customize Shortcuts…, Edit Toolbar… and Reset Panel Layout: the shortcut editor, the picker for which controls sit on the canvas toolbar, and a return to the default panel sizes and positions.
- GPU acceleration: render the pipeline on the GPU. The active backend is named below the row. Off falls back to the CPU pipeline, which is slower but produces the same image. If the GPU viewport itself fails to start, a warning toast says so at launch and an amber line here repeats it; the display then runs on the CPU.
- Multi-core CPU rendering: see §4.3. It takes effect at once, with no restart.
- Preview size (512 to 8192 px): long edge of the interactive canvas. Higher is sharper at 100% zoom, and costs proportionally more VRAM and CPU per frame, so lower the cache limit and the rendered-frame count to match. RAW files decode at half sensor size for the preview, so there is nothing to gain past half the long edge of your scan.
- Preview cache and Preview cache limit: how many recently-viewed photos stay decoded in memory, and the memory ceiling for them. Lower both on a machine with little RAM.
- HQ buffers: full-resolution HQ preview buffers kept in memory. Each is large (a 60 MP scan is about 700 MB), and keeping the previous frame makes going back instant.
- Rendered frames: rendered frames held for navigating back with no re-render.
- GPU texture cap: largest GPU texture dimension, including HQ preview loads. 0 lets the hardware decide, except on an integrated GPU, where a conservative default applies automatically to avoid a VRAM crash; set it to 4096 if exports run the card out of memory.
Every row from Preview size down is read at startup, so a change needs a restart. A value set in override.toml wins over these, and its row is grayed out and says so.
- Persistent Settings…: which edits carry onto the next file you open (§2).
- Manage Database…: stored row counts and sizes, clearing saved edits, and your library roots.
If NegPy crashes on launch or has rendering glitches, force the backend without touching code. On first run NegPy creates Documents/NegPy/override.toml with defaults for your OS. Edit it and restart. The [performance] numbers above are here too, and a value in this file wins over Preferences, which is what makes the file usable on a machine that will not start.
| Setting | Values | Effect |
|---|---|---|
rendering.backend |
"auto", "vulkan", "dx12", "metal", "cpu" |
GPU backend for image processing. "cpu" disables GPU entirely. |
display.qt_rhi_backend |
"auto", "vulkan", "d3d12", "metal", "opengl", "software" |
Qt UI rendering backend. |
display.qt_platform |
"auto", "xcb", "wayland" |
Window system plugin (Linux only). |
performance.force_hq_preview |
true / false (or absent) |
Overrides the saved HQ preview toggle. |
performance.cpu_parallel |
true / false (or absent) |
Multi-core CPU rendering kernels. Defaults on, except on macOS. |
logging.level |
"debug", "info", "warning", "error" |
Log verbosity. Use "debug" when reporting issues. |
max_texture_size, preview_render_size, preview_cache_max_bytes, preview_cache_max_entries, preview_cache_max_full_res_entries and render_memo_max_entries take the same values as their Preferences rows.
Common fixes:
- Crashes immediately on Linux →
backend = "cpu"orqt_rhi_backend = "opengl". - Black or blank preview on Windows →
backend = "dx12"orqt_rhi_backend = "software". - Wayland rendering issues →
qt_platform = "xcb"to force X11. - GPU out-of-memory during export →
max_texture_size = 4096.
NegPy asks GitHub for the newest release once at startup. If there is one, a green ⬇ Update Available: vX.Y.Z line appears under the logo in the left panel. Click it to open the update window with the release notes, the download size and one button.
To check yourself, press the ↻ button next to the version number in the left panel's header. If nothing is new, NegPy says so; if something is, the button turns into a green ⬇ and the update window opens.
Install Update downloads the build that matches how this copy was installed, then closes NegPy, installs it over the old version, and reopens on the new one. You do not download, uninstall or reinstall anything by hand. Nothing is replaced until NegPy has exited, so a failed download or a refused permission prompt leaves your working install exactly as it was.
What happens per platform:
| Install | What NegPy fetches | How it installs |
|---|---|---|
| Windows | the -Setup.exe installer |
Runs it silently over your existing install. Windows asks for administrator approval first, because the app lives in Program Files. Approve it before NegPy closes. |
| macOS | the .dmg for your chip (Apple silicon or Intel) |
Mounts the image and replaces the NegPy.app bundle where it currently sits, then reopens it. |
| Linux | the .AppImage |
Replaces the AppImage file you launched, keeps it executable, and relaunches it. |
Your edits, presets, settings and library are untouched: they live in Documents/NegPy and the local database, not in the installation folder.
When the button says "Open Releases Page" instead, NegPy cannot swap itself and sends you to GitHub. That happens when you run from a source checkout, when the release has no build for your platform, or when the app was moved somewhere it no longer matches its installer's layout. The same happens if the folder holding the app is not writable by you (an AppImage in a system directory, an app bundle in an /Applications you do not own); NegPy says so rather than failing halfway.
You can ask for the check again at any time with the Check for updates action, unbound by default, so give it a key in the shortcut editor.
- GPU acceleration: NegPy uses your GPU for near-instant previews and responsive sliders. The Normalization panel's analysis (bounds, white/black point, normalize) runs on the CPU. Turn it off in Preferences → Performance if you suspect a driver issue, or force the backend in
override.toml. - Database: all edits live in a local SQLite database keyed by file hash, so you can move or rename files without losing your work. Optional
.negpysidecars mirror edits next to your sources. - Saving edits: edits are written to the database on export, when you switch frames, or when you save explicitly. Closing the app mid-edit without any of those loses unsaved changes.
- Keyboard shortcuts: KEYBOARD.md
- Filename templating: TEMPLATING.md
- The pipeline in depth: PIPELINE.md