Just out of curiosity, has anything changed about the denoiser in the daily builds? I just jumped to a recent daily from 1.7 HF2 and it almost seems that the denoiser is a little different in its behavior. If I had to describe it it'd be a little more blurry\blotchy at high levels (above .650). Like so many things, could be user error or my imagination...or scene dependent, this is a tough one.
Yes, there were few changes in the denoiser and adaptivity computations (see the changelog for details). Could you provide us with some example where this happens? A scene would be the best, but just sending the render saved in CXR should be enough. At the very least, please post your denoised renders from 1.7.2 vs v2 for comparison :)
How do you override by glass material? Meaning, how does corona know what is a glass material? Do you need to name the material "glass"? I haver a scene where the glass is part of a multi sub material and i cant get it to work.
It is currently some dark magic based on reflection & refraction settings. The name of the material does not matter. This detection will most definitely require some tweaking, so please send us configuration of your glass material and we'll see what we can do.
Also, for the multimap update, how does this work exactly? If I have a tree with a multi sub material where material 1 is the leaf, material 2 for twigs and material 3 for bark, does this allow for a coronamultimap sub material applied to material 1 to be randomized now? If so how? I have tried without any luck.
"Multimap can now be created with just one map allowing to apply hue randomization and other options to just a single map"
The only change is that previously the multimap had to have at least two maps. Now it can have just one. The result should be the same as with 2 maps that are identical.
So I decided to test the noise in the latest daily build, and found I'm getting way more noise than the 1.7.2 release.
This scene is lit by a 1k env map, a 'sun' light (shining behind the surfaces in the images and a few small lights off camera which don't really add much illumination to the scene.
I cropped in to a small area, and I found that the 14 minute 4% noise render produced in 1.7.2 has a similar quality level to the latest daily 30 minute 7.5% noise render. Both using the new light solver. I understand that the different builds may display noise differently, but it's hard to not notice that time-wise, it took over twice as long in the latest daily to get something as good as 1.7.2.
So unless I'm doing something wrong, I'm looking at doubling my rendertimes for the new 2.0 release?
Could you send us the scene for internal performance testing?
Hey guys... when could we see this important feature "Multimap - randomization by mesh elements"?
Please... do this asap ))
thanks
Sorry, we had to sacrifice some features originally planned for v2 so that we won't have to postpone release any further.
Hello!
Found an error in CoronaBumpConverter, the bamp is not correctly displayed when mirroring is mapped.
Whenever reporting such bug, where you have a simple testing/reproduction scene, please try to attach the scene. This will enable us to fix this bug much faster, since we can just jump in the code without having to setup the reproduction case based on screenshots :)