The way layering is implemented in ren2 keeps track of the highest layer and scales accordingly. So, if the layers 0, 1, 2 and 1000 are used, then 0, 1, and 2 are pushed to the extremes of the positive z axis (0.9) while 1000 is pushed to the extreme of the negative z axis (-0.9). The entire middle is left unoccupied. With a big enough discrepancy, it is theoretically possible that the precision of floating point numbers could create layering issues (i.e. the low end of the layers all getting pushed to 0.999999... until the computer had no more space to store the minute differences between the layers).
Is this issue practical or problematic enough to warrant a more bulletproof algorithm?
The way layering is implemented in ren2 keeps track of the highest layer and scales accordingly. So, if the layers 0, 1, 2 and 1000 are used, then 0, 1, and 2 are pushed to the extremes of the positive
zaxis (0.9) while 1000 is pushed to the extreme of the negativezaxis (-0.9). The entire middle is left unoccupied. With a big enough discrepancy, it is theoretically possible that the precision of floating point numbers could create layering issues (i.e. the low end of the layers all getting pushed to 0.999999... until the computer had no more space to store the minute differences between the layers).Is this issue practical or problematic enough to warrant a more bulletproof algorithm?