Thanks for fixing. Is build with the fix downloadable somewhere? I've tried daily builds, but seems it doesn't have a fix and I don't know how/when fixes are implemented. Could someone explain me?

In T100106#1398360, @Jesse Yurkovich (deadpin) wrote:

I can reproduce here using an older 1070 - but only when using OptiX, CUDA and CPU seem fine.

In T100106#1398339, @Pratik Borhade (PratikPB2123) wrote:

Thanks for the file. I've not been able to notice any performance drop with either texture connection.
Looks like some textures are missing from the file (all objects are in magenta color when switched to cycles)
Only sketchbookCover.jpg texture is available in file.

In T100106#1397629, @Pratik Borhade (PratikPB2123) wrote:

Thanks for the report. Please upload .blend publicly if possible

Is there some workaround I could use? I've noticed only this instancer in green frame causes problems, but other instancers in GN graph weren't causing problems. Could it be that problematic instancer spawned too many instances?

Will it be possible to disable horizontal margins? Viewport + outliner already make setting panel cramped horizontally, often causing texts to be cut.

Found the culprit via trials/errors:

I thought I noted it there, but seems it slipped out of mind. But I've managed to get that super-bright rim back by adjusting Clamp, IIRC, to either 0 or very high value.

I've noticed also jagged artifact, camera height position and viewport zoom in/out affect it - there's recording:

@Clément Foucault (fclem) That's fantastic, just checked out .blend file and I don't see artifacts originally reported in first post. However, I've noticed some strange outlines - their appearance depend on how you zoom in/out in viewport and on value of F-Stop in camera settings. Is it an known issue? Enabled Jitter in DOF settings solves it, if it helps.

I've checked out 2.93 Beta committed 20 April 15:45, the white-blown out artifacts on wet floor seem to be gone, but there's still regression/change on how bright blue light passes through glass - comparison pics are still up to date in the original post. Can I ask for some update/triaging/dev comment? I don't know if it can't be reproduced, or if it's bug or side effect rework of reflection/translucency code.

In D10835#279951, @thin soldier (thinsoldier) wrote:

In D10835#279548, @Harley Acheson (harley) wrote:

I'm not understanding how this has changed from before the button. Wouldn't you open the same file, also not know that it is on, move the camera, and have the same situation?

Correct. I'm just trying to bring attention to a problem I consider more annoying than having to dig into the sidebar menu to find the "lock camera to view" setting.

Wouldn't it better UI-wise to place spread parameter like that? I'm not sure why separate parameter section is done, unless there are planned several new parameters in future for beam shape?

Great simple button, very much needed. Now only we need to add camera to current viewport view with 1 click (instead of adding camera, then aligning view to it through bunch of menus)

It's missing blue glow on glasses. Agreed on white reflection artifacts though, you don't have those, unlike me. But I've noticed too your glass look very different from me, I can't see water droplets and there's some large-scale warping noise going on.

In T86583#1134241, @Hadrien Brissaud (hadrien) wrote:

I don't know, doesn't this seem more complicated than the problem it tries to solve ? Don't frames and reroutes already provide all the organizing functionality ? Is scrolling actually an issue ? Maybe a minimap could help in that regard ?

No idea, but =<2.92 didn't have these artifacts, it's new. Dev mentioned tiles not being calculated correctly, so maybe it's a screenspace thing?

In T85845#1129410, @Florian Kamenz (TimBrown) wrote:

this is normal. that are the different distances from the camera.
in the old system it was the same

In rB000a340afa67a12a38c103e19f7d52ed42c4492e#295555, @Clément Foucault (fclem) wrote:

In rB000a340afa67a12a38c103e19f7d52ed42c4492e#295546, @nameless (tonpix) wrote:

It's amazing improvement, in my cinematic short clip, it pretty much solved all DOF issues I've had, except there's an one artifact that's looks odd to me - I've recorded it, please look above the mouse cursor, the artifact shows/disappear depending on viewport zooming in/out (camera stays same) https://streamable.com/5p9s0h

I feel like it's the return of the same bug I fixed many times (tile clasification not being conservative enough). Can you try to isolate the 2 objects that create the issue, the camera setup, create a bug report ticket with the simplified file? That would be very helpfull to fix the issue.

It's amazing improvement, in my cinematic short clip, it pretty much solved all DOF issues I've had, except there's an one artifact that's looks odd to me - I've recorded it, please look above the mouse cursor, the artifact shows/disappear depending on viewport zooming in/out (camera stays same) https://streamable.com/5p9s0h

After seeing info that LANPR would be used as backend only, while frontend would be GreasePencil, I have some doubts - would really production ready LANPR compete visually with solution there? No offense, but I don't feel that GreasePencil has intricate visuals like custom material node there. (it's just vector curves that are very opaque and vary in thickness, not much more? I'd be happy to be corrected though)

I think I might have found a culprit, hopefully possible to reproduce. If I assign pin vertex group, it prevents displaying tooltip from hovering in middle (with UI scaling x1.24). When I remove pin vertex group, tooltip pop out works correctly. Sharing clip + .blend file. Seems to be very obscure UI bug cause, I'm not sure how I even found it. Possibly to affect other areas of panel with group selectors?

How can I help, then?

5s+ doesn't make tooltip pop. But your hint about resizing UI gave different result, tooltip appears like in 80% cases - I'm not sure when it stops popping, it felt random, I've tried clicking around, but couldn't find pattern.

What's the status of this bug? Is there some ETA for fix? I had the same issue as well, parented Collection Instance to Plane that had Array + Curve + Mirror. Instances on mirrored side had incorrect rotations sadly. I've attached an simplified .blend if it helps anyone

I meant to ask if this could be hotfixed in 2.81. Especially if there's already an fix. 2.82 alpha isn't production ready and it will be months before reaching this point. Thanks.