V-ray Next 4.30.02 - Extreme "dancing" white pixels on any material at a distance with glossy reflections

[Vizioen]

I also have this problem with a default Vraycarpaintmtl2 (just colors changed and I even disabled the speckles because I though this was the issue. In my case it's animation though, haven't tried it with still shots.

When I do animated aerial shots, the cars flicker like a discolight. I've learned to live with it, but it sucks. LC on animation settings, and default vray settings basically.

Is it still the way to go with 2 VraySuns, and big size for specular? It's a bit cumbersome. Latest Vray btw.

[kosso_olli]

Well, the carpaint is a Vrscan, but it happens with all materials having glossy reflections.
The "solution" to use two different suns was not tested, as this just does not feel right or physically plausible.

[Joelaff]

I wonder if Lock Noise Pattern would at least keep them consistent.

We still run into this a lot. It always always ends up being a second light for the highlights. Also, been having much better results with a distant regular Vraylight vs. a VraySun. Not quite as accurate, but tends to work much better. A vignette as a light texture map also seems to help.

Would really like to see a fix like Corona has.

[kosso_olli]

I wonder if Lock Noise Pattern would at least keep them consistent.

Just had to check this, because I did not think of it: Indeed it does!
Side-effect is that you get a noise pattern that looks like hovering over the image...

EDIT: It does only work if the camera is static. As soon as it moved, you get randomly placed hightlights again.

[Joelaff]

Just had to check this, because I did not think of it: Indeed it does!
Side-effect is that you get a noise pattern that looks like hovering over the image...

Yeah, if the sampling is high enough, and the noise is minimized that goes away (or through multi frame denoising, which is what the lock noise pattern is so good for).

I prefer a dancing noise look, but without the dancing highlights!

[Joelaff]

Are you making a still or an animation? You could also try it without the lock nosie pattern, but then average a bunch of frame together. Of course that increases the render time linearly.

[kosso_olli]

Are you making a still or an animation? You could also try it without the lock nosie pattern, but then average a bunch of frame together. Of course that increases the render time linearly.

We are doing an animation where the camera is moving very slightly. And all those highlights are dancing around like crazy. Keeping the camera static and having the highlights in different positions for consecutive frames just seemed strange and we discovered it while debugging the scene.
As I said in the last post, having the noise pattern locked only helps when the camera is static. As soon as it moves, you will get the randomly placed highlights again. I still did not find a cure for this.

[kosso_olli]

Okay guys,
I am stuck with this. It is impossible to get a clean-looking sequence out of V-Ray.

[^Lele^]

do you still have my email, Olli?

[kosso_olli]

Did you send one? Seems I missed it, I can not find it :/
Could you please send it again?

[^Lele^]

Ahah, no i meant email address, for you to send me over the scene so i can check it, and in case ameliorate it, or pass it to the devs to test the advancements we'd need.

[kosso_olli]

Sure, eMail was sent right now. Thank you very much!

[^Lele^]

So to close the topic: the issue was with geometric normals.
V-Ray was doing what it could with what it was given.
Geo fixed, issue gone.

[Vizioen]

So to close the topic: the issue was with geometric normals.
V-Ray was doing what it could with what it was given.
Geo fixed, issue gone.

What exactly was wrong the normals?

[^Lele^]

Unwelded, coincident vertex normals.
Analogous to z-fighting, but for normals, essentially.
Usually it doesn't matter as all the unwelded vertex normals point the same way, but if even just one does not, the random seed across frames may mean a different one is picked.
Combined with the scene's specific lighting angle, it produced the unwanted result.
It's a tricky one to spot as well, because visual inspection of the mesh does not show *big* issues, and if viewed as diffuse, or inside a normals RE, it may just look OK.
One has to get up close, and try things, to notice.