The difference between the 2 images is just the quality slider changed from lowest to second to lowest setting. I thought this might have to do with a few touching bodies, but after removing some of them, even more erroneous regions showed up, specifically in regions that do not touch anymore:
I had a quick look and it seems to me as if there are indeed overlap errors. For example, the green points below show positions that are included in several regions (e.g. AbsR and ABSCOREL).
I would suggest looking into this as this could potentially lead to problems in your calculations.
Yet, when it comes to the generation of the image only, you can still do this at present. If you go to the “Options” dialog under the 3D tab, you can tick off “Show errors”.
yes there are still some touching bodies. I ignored them as numerical overlap should have no meaningful impact on what I want to study. Also the touching bodies don’t coincide with the region I marked in my third image.
By now I have convinced myself that the errors that show up are actually a buggy handling of alpha/see-through portions. Case in point: I found a ghost body that is actually the cut of my viewport with clipped away parts.
errors appear in some overlapping, see-through area. (note the light grey region had it’s alpha value removed here.)
If the see-through-ness of the cut is handled in the same way as alpha values of regions, this would explain the erroneous areas I saw before.
the errors that you see in when switching to the high quality renderer came from a very subtle bug that became apparent only in this mode despite being there also in the fast preview. There was some corruption in the tracking cache, that become apparent when we completely track rays through transparent regions. It is fixed (as you can see below) and the fix will appear in a future release.
