Am 18.09.2013 16:42, schrieb Thomas de Groot:
> In a long render, I twice used +C. In each case it seems, a small
> artefact remained from the previous render session (see image in
> p.b.i.). Could this be a bug?

I can now confirm that this may happen, that it is a bug, and that it 
actually /is/ radiosity related.

What happens is that the continued render starts the pretrace for the 
missing blocks again; during the early stages of this pretrace, the 
"pixels" may be far larger than the render blocks, and may overpaint 
some already rendered blocks.

I've just coded a hotfix that will prevent pretrace results from being 
written to the image buffer. This way the artifacts will still appear in 
the preview window, but not in the final image file.

Avoiding the artifacts entirely seems a bit tricky, and is probably not 
worth the pain as the pretrace step is scheduled for a redesign anyway.

Post a reply to this message