V-Ray 5: VrScans Clearcoat is blurred compared to Next

[kosso_olli]

Yes, it is.

[Ivaylo Katev]

Well, i tested it with the simplified stene provided, and wasable to reproduce it. Just for information, the blur is bigger than the expected one because the camera is parallel.
Here are the suggested sollutions, starting with the fastest one
1. The problematic material canbe modified to produce the same result as in vray next (already tested this, it works), i can send it to you immediately
2. A command line tool to introduce this correction in any material, can be prepared approximately until the end of the day
3. The materials scanned before the glossines change can be rendered in the old way, you have to wait until the first service pack

[kosso_olli]

3. The materials scanned before the glossines change can be rendered in the old way, you have to wait until the first service pack

Ivaylo, thanks for checking. Solution number 3 is fine for us, because we postponed the update to 5 anyway. Right now there are too many issues that need to be adressed, this was one of them.

[kosso_olli]

The fix is not implemented in 5.00.05, right? Because I just tested that version and it has the same issue.

[hermit.crab]

The fix is not implemented in 5.00.05, right? Because I just tested that version and it has the same issue.

The issue is fixed in the latest nightlies, it is not yet implemented in the stable builds.

[kosso_olli]

Sadly I have to report thas this issue is not fixed!
Right now we are preparing to finally switch over to V-Ray 5 (which we paid a LOT of money for!), and I discover this: Blurry clearcoats on ALL the materials we got. Materials from Chaos own library, scanned samples from a few years ago and samples from 3 weeks ago: They all suffer the same issue: The clearcoat is extremely blurry, as described in the first post.

I am not going to accept this, as it is a major degradation to what we had before.

First is Next (the expected behavior), Second is 5.10.02

[seandunderdale]

Ive resulted to only using the base of a scanned carpaint and making my own gloss coat, then combining it all in a blend shader setup with a custom orange peel also.

[kosso_olli]

Ive resulted to only using the base of a scanned carpaint and making my own gloss coat, then combining it all in a blend shader setup with a custom orange peel also.

I refrain to accept this as a solution, because it defeats the purpose of VrScans.
I just can not wrap my head around why this was done. It is not a realistic behaviour at all. I have the physical samples to most of our scanned materials here, none of them show a blurred clearcoat at all.

[Ivaylo Katev]

Hi Oliver,
I'm really surprised about the old materials , as i promissed their rendering was restored as in VRay 4, i will check this once again and will post the result later.
About the new materials - as i explained above currently we are able to measure the real glossines and by default we set the measured value on every carpaint scan. However, we have control and can specify any glossinesso by demand and beside this glossines control option will be added to the render side as well

[seandunderdale]

Totally agree, its just the only way I can get renders into comp until its fixed. Youve been beating this drum now for almost a year, so Im not holding my breath. Hopefully there is a fix soon though!

[Ivaylo Katev]

none of them show a blurred clearcoat at all.

Can you tell me specific sample name ?

[kosso_olli]

Can you tell me specific sample name ?

Ivaylo, try it with ANY sample. Every sample I tested (around 20) shows the blurred clearcoat in 5.10.02, no matter what.
But try carpaint_grey_2_s. Sharp clearcoat in Next, blurred clear coat in 5.10.02

[Ivaylo Katev]

OK, the first step is done, i did the test for what i promissed and what you agreed, to restore the rendering for the old materials, scanned before the glossiness measurement, and they seem to render really as expected. I made a test with a new material to be sure that really the difference occurs only for new materials But you said that you have new scanned materials, that apparently is not protected by the solution you agreed with. So until we clarify all the points, i suggest to manually correct all your scanned materials to have highest possible glossiness, and to turn off the automatics for the clear coat for all your furhter scans.

[Ivaylo Katev]

But try carpaint_grey_2_s. Sharp clearcoat in Next, blurred clear coat in 5.10.02

OK, trying with the material you mentioned, here is the result, there is no difference, even a single pixel. I'm afraid we are missing something, can you prepare from scratch a very simple scene that reproduces the difference under 4.3 and 5.1?

[kosso_olli]

I think we are misunderstanding each other.
What I want is consistently sharp clear coat, across all the samples, no matter when they were scanned. I can not get that with V-Ray 5, but I can get it with V-Ray Next 4.3
I DO NOT need variying degrees of glossiness based on the date a sample was made. This is not usefull.

Probably the best solution for 5 is to load the measured glossiness specified in the sample when loading the material, but also the ability to override that value to 1, i.e mirror-like glossiness.

However, this argument is still valid: As it is right now, the VrScans are impossible to use for us in 5. We are not able to get consistent results.