VfR next bug overview (by Micha)

[Micha]

Hi,

to keep the overview over my found issues I wrote a little list. Maybe it helps the team or other users too. I will remove the bugs from the list if they are fixed at a new build. Some older bugs are forgotten, but I will update the list. The bugs are a collection only and was reported at the forum or per email.

(1) NVLink not working
(2) if a complex scene was rendered at a first Rhino task and a complex scene can't be rendered at a second task - wrong memory limit
(3) small offsets cause errors in parallel views
(4) color sliders not linear
(5) energy preservation mode can't be disabled and cause unrealistic color effects for pearl materials
(6) catmull rom AA filter looks blurry (GPU)
(7) intensity control of textured area lights not working (GPU)
(8 ) AA filter Lanczos 1 cause a black output in progressive mode
(9) GPU bucket size to big, most to big for region render, which cause longer render times as necessary
(10) long strong CPU usage of the NEui.exe (approx. 1min 65%, than 25%), also if the asset editor is closed
(11) IM buckets wrong (size and count), if progressive mode is enabled and LC+IM used
(12) texture filter not working
(13) enable denoise cause effects render element at the frame buffer, also if no effects are enabled
(14) crash if Rhino file is saved during interactive Rendering (!!!)
(15) sampling issue during progressive rendering (one pass is noisier than a previous pass, render time lost)
(16 ) pixelated texures of area light projections (directionality 1)
(19) bad sampling of metallic materials based on metallic values between 0..1 (strong noise effects)
(20) Interactive mode reset non-interactive mode GI to LC+BF always (for example if it was BF+BF before)
(21) Progressive GPU is nearly at 50% speed of bucket GPU
(22) slow LC pass for high res renderings (!!!)
(23) Anisotropy -1 or 1 of reflections is rendered like 0


missed features
(1) ambient light
(2) screen mapping for GPU
(3) spot light textures for GPU
(4) region render per Rhino viewport rectangle selection
(5) material ID color through glass for GPU
(6) advanced UI mode for every Rhino task per default
(7) simple material preview for clearly visible UV range 0..1
(8 ) "no decay" of lights missed for GPU
(9) missed to save only selected render elements (for example only denoise and mat ID)
(10) soft user stop for progressive renderings so that the current pass is finalized and denoised
(11) binding per context menu
(12) GI AO for GPU
(13) Disabling "show calculation" of GI passes missing (only per button script possible)

[Peter.Chaushev]

Hello, Micha,
Thank you for this summary.
To keep the conversation readable, I'll reply after each point:

(1) NVLink not working

Was informed the issue you've experienced was related to specific project(s). Its cause will be fixed with Update 1.

(2) if a complex scene was rendered at a first Rhino task and a complex scene can't be rendered at a second task - wrong memory limit

Again, should be resolved soon.The Support team is still investigating some of the circumstances of this behavior.

(3) small offsets cause errors in parallel views

Was informed this case is already resolved.

(4) color sliders not linear

No decision has been reached yet. For the time being it will remain as it is.

(5) energy preservation mode can't be disabled and cause unrealistic color effects for pearl materials

The Energy Preservation mode only has application in legacy workflows. I was not able to find the thread you have discussed this problem with pearl materials. Please remind me what the issue was.
In any case, it cannot be disabled, only switched between Color and Monochrome (RGB is set by default)

(6) catmull rom AA filter looks blurry (GPU)

This is a known limitation. The sharpening filter is currently not supported on GPU and will probably be excluded as an option in the next update. We’ll update the documentation to reflect this.

(7) intensity control of textured area lights not working (GPU)

This will be amended in one of the upcoming updates.

(8 ) AA filter Lanczos 1 cause a black output in progressive mode

Most filter types produce this bug when their size parameter is set to values below 1.42. The UI will be adjusted to limit inputs below this value.

(9) GPU bucket size to big, most to big for region render, which cause longer render times as necessary

As already mentioned, Bucket size for GPU will be hidden from the UI to avoid causing any further confusion. The GPU engine automatically handles load balancing between GPUs. Dynamic bucket splitting would not improve render time in this case.

(10) long strong CPU usage of the NEui.exe (approx. 1min 65%, than 25%), also if the asset editor is closed

Once the Asset Editor is opened both its process and that of the asset preview continue to run even while the window is closed/hidden. As for the CPU usage of the preview render - it is spread across all cores and is set to low priority which should not inhibit other ongoing processes.
There's an idea how to negate its impact during heavy production rendering which we may put to test soon. In the meantime, the Asset preview can always be manually paused.

(11) IM buckets wrong (size and count), if progressive mode is enabled and LC+IM used

Will need more info on this. If you can reliably reproduce it, please reach out to support@chaosgroup.com with a step-by-step description and/or a project file.

(12) texture filter not working

This is not supported on GPU. There are so far no plans to proceed with it.

(13) enable denoise cause effects render element at the frame buffer, also if no effects are enabled

There is no way to disable effectsResult channel from being created if the image is being Denoised, there’s a Lighting Analysis render element or Lens effects applied. You would have to manually remove the channel image file afterwards.
In the upcoming update, however, using the VFB "Save current channel" button will not append the channel name as a suffix.

(14) crash if Rhino file is saved during interactive Rendering (!!!)

We’ve received a few reports about it so far but have not managed to reproduce it on our side. If anyone is able to reproduce it reliably, we would be very interested in a demonstration.

(15) sampling issue during progressive rendering (one pass is noisier than a previous pass, render time lost)

Production or Interactive rendering used? If Interactive, will be resolved in next build. If this occurs during Production rendering, will require more info to be able to assist.

(16 ) pixelated texures of area light projections (directionality 1)

As with (7), this will be amended in one of the upcoming updates.

(19) bad sampling of metallic materials based on metallic values between 0..1 (strong noise effects)

For the current implementation of the Metallic material, consider using values of either 1 or 0, based on the result you wish to achieve.

(20) Interactive mode reset non-interactive mode GI to LC+BF always (for example if it was BF+BF before)

The current behavior will be investigated.

(21) Progressive GPU is nearly at 50% speed of bucket GPU

I was informed this report of yours is still being investigated by my colleagues from the Support team.

(22) slow LC pass for high res renderings (!!!)

Fixed. Will be amended in Update 1 or an early stable nightly build.

(23) Anisotropy -1 or 1 of reflections is rendered like 0

Already noted. UI will be adjusted accordingly in one of the next updates.

missed features
(1) ambient light
(2) screen mapping for GPU
(3) spot light textures for GPU
(4) region render per Rhino viewport rectangle selection
(5) material ID color through glass for GPU
(6) advanced UI mode for every Rhino task per default
(7) simple material preview for clearly visible UV range 0..1
(8 ) "no decay" of lights missed for GPU
(9) missed to save only selected render elements (for example only denoise and mat ID)
(10) soft user stop for progressive renderings so that the current pass is finalized and denoised
(11) binding per context menu
(12) GI AO for GPU
(13) Disabling "show calculation" of GI passes missing (only per button script possible)

Some of the requests were considered edge cases and will not be considered while others are already logged in our system. As always, detailed feedback is tremendously appreciated!

Kind regards,
Peter

[Micha]

(15) sampling issue during progressive rendering (one pass is noisier than a previous pass, render time lost)....
Production or Interactive rendering used? If Interactive, will be resolved in next build. If this occurs during Production rendering, will require more info to be able to assist.

It can be seen during the production rendering. I reported it some weeks/months before at the beta forum and the bug is still alive. Konstantin confirmed the bug. You can see the progressive rendering in passes, one of the first passes is very noisy, next one is finer, than the very noise level is rendered again, than the finer pass comes again and now the next passes works like expected. So, if only a few passes should be used, this additional passes can cause a quite notable slow down. Instead 1 min render time it can be that 2 min are needed for a specific noise threshold.

[Peter.Chaushev]

It can be seen during the production rendering. I reported it some weeks/months before at the beta forum and the bug is still alive. Konstantin confirmed the bug. You can see the progressive rendering in passes, one of the first passes is very noisy, next one is finer, than the very noise level is rendered again, than the finer pass comes again and now the next passes works like expected. So, if only a few passes should be used, this additional passes can cause a quite notable slow down. Instead 1 min render time it can be that 2 min are needed for a specific noise threshold.

The cause of this issue has already been found. It will be resolved one of the next updates for sure. We hope to be able to cherry-pick it for the upcoming Update 1.
In the meantime, if you are able to provide a scene we can investigate, I may be able to provide you with a workaround until then.

Kind regards,
Peter