In addition to the previously mentioned issues with the rendered view in Cycles under 2.91:
Is anyone else having issues with Cycles sometimes using the Render denoising sample settting for the viewport Denoising sample settings?
Is anyone else having issues with every new Pricipled BSDF material created having emission strength set to one?
So far, aside from one response by Stan Pancakes advising me of a known AMD GPU issue, which does not apply to my hardware, I’ve not had much response about any of these issues.
The render viewport isssues I have been having completely break my lighting and texturing workflow.
Right now, it’s looking as if 2.91 is a hot mess, and I may have to roll back to 2.90, which I would rather not do. I’ve been eagerly awaiting some of 2.91’s new features, but until an unless these problems are solved, 2.91 is simply not very usable.
Well, I believe Emission Strength is a multiplier, so 1.0 would be the correct default value to give the same behavior it had before. The Manual says:
Strength of the emitted light. A value of 1.0 will ensure that the object in the image has the exact same color as the Emission Color, i.e. make it ‘shadeless’.
Interesting. Previously, emission strength of zero was normal for a pricipled BSDF, which is not intended to be an emission shader. Put another way, a Principled BSDF is not an emission shader and should not be emitting light as a default.
Edit: I mean, yes it can do that, but light emission is not its primary purpouse.
Thanks for the response Zoot, much appreciated!
If you don’t report the issue on developer.blender.org there’s a chance it may not even get fixed. I don’t recall recently seeing reports similar to what you’ve shown in the other thread. Maybe I’ve just missed it, or maybe you’re having some sort of a unique problem (driver issue, for example). There’s no way to tell either way if you don’t report it.
EDIT: You can use the Help -> Report a bug option from within Blender so it’d pre-fill some of the required information for you.
Considering that it’s not going to be worth reporting until I find out whether it’s a bug or merely a problem with my system, I’m trying to find out if anyone else has this problem, and get ideas for what is wrong with my system, or what may have been legitimately changed that my system can’t handle. I also want ot make sure it’s not an ID10T error, or a PEBKAC error.
I think it’s advisasble to do this, before I go open a thread over there, get sniped at about all the info I didn’t automagically provide, and then have the thread closed because some dev doesn’t want to deal with it, or because it’s a legitimate change to the software and they don’t want to support older systems, which are the commonly seen response patterns I’ve noted when I have perused threads at the aforementioned link.
I just noticed your edit, about the bug report option. Thanks for pointing that out, it will be useful, if I get to the bug reporting stage.
The worst that could happen is that devs wouldn’t be able to reproduce the issue. They don’t just close reports because “they don’t want to deal with it”, if it is indeed a bug report and not a feature request. I mean, it can happen in some fringe cases where a short discussion about “to bug or not to bug” could occur, but this doesn’t look like it could even remotely be one of such cases Graphics issues especially aren’t just looked at by one dev either.
If you want to exclude a PEBCAK, you could upload (a part of) the offending file from that other thread, and I or someone else could look at it to see if there’s something wrong with that specific file. Maybe also find out if that behavior is reproducible on other systems.
I would happily upload the file, since that would make things easier, but it’s proprietary.
I did trace down part of the problem, and that is a limitation in the number of light emitters, whether they are lights, mesh lights, or just planes with an emission shader. Apparently, something like 1200 light sources is too many. Pity. It wasn’t before.
Mesh lights do seem to be working better in this situationthan light emitting planes though, so that is an improvement. Not usinfg regular lights because they are not compatible with array modifiers and deform modifiers, but this is a known issue, not new.
I have (recently) encountered a transitory error where occasionally it will inform me that viewport denoising cannot be done, because it could not establish a CUDA file because it is a n illegal context (Wut?) or it will claim Optix is not supported. I have no idea what is going on with that, although I strongly suspect that is an Nvidia issue, since there have been so many Nvidia Driver problems earlier this year. Not able to reliably reproduce that particular error.
I appreciate the reassurance abotu the devs, Stan. I have seen some “BlenderBro” behaviour there in the past. Having looked over things today, I’m not seeing any of that now, so it’s a nice change of pace. Mostly I seem to have solved the problem. I’ll keep investigating things on my end. Some of this could have been a bad install.