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

[kosso_olli]

Hey guys,
I found an issue in handling the clearcoat of VrScans materials: In V-Ray 5, the clearcoat seems to have a glossiness value below 1, as the reflections are blurred. In Next though, the reflections are razor-sharp. I do wonder where the difference is coming from. I mean, the VrScans are measured, so why can there be such a deviation in the output?
I can not find anything about this in the changelog, nor in the documentation.
So I really really hope this is a bug, and not a feature...

First is Next, second is 5.

[kosso_olli]

Can anyone confirm that there is a difference between Next and 5?

[tanya_to]

Hello!

I haven't managed to recreate the difference with the clearcoat in a simple scene. Could you send the scene to tanya.todorova@chaosgroup.com for investigation?

Best Regards,
Tanya

[hermit.crab]

Thank you for the report and the provided files. There seems to be a difference, indeed. I have logged the issue (internal bug-tracker id: VMAX-10122) for developer analysis.

[kosso_olli]

Thank you very much. I hope a solution can be found.

[kosso_olli]

Did the analysis reveal anything helpful?

[nikoleta.garkova]

Hi Oliver,

After developer analysis, it has been concluded that this is the current expected result. Before V-Ray 5, clear coat specular highlights were considered negligible, because the clear coat was added on a pre-measured IOR value - hence the CC properties were not fully recreated. Currently, however, we've improved the ability to scan highly reflective materials which rely on the clear coat rendering. We can directly compare the rendering result with the images in the raw scan, and the goal is to have as much similarity as possible.

[seandunderdale]

To be fair, carpaint can be surprisingly shit in terms of blur and orange peel, but clients often don't want that in the paint. If there was a way to override that and have an option for totally clean and crisp coat, then that'd be useful. Cant the coat be disabled in the scan and then added back on in a blend material?

[^Lele^]

It sure can, as additive.
Just watch out for reflection values, keep them lower so to leave room for the underlying VRScan, rather than breaking through the EP ceiling.

[seandunderdale]

So if you use a blend shader for this, with the scan at the base, and a vray material set to be full mirror in slot one, and a fresnel in the mask control, would that not work fine? Rather than using additive? Thats basically how I setup my non vr scan car paints. I assumed additive was the devil in blend shaders, when realism is the goal, and should be avoided at all costs. This way you can then dial in a fraction of blur in the gloss and use a custom orange peel map, and not worry about energy conservation.

[^Lele^]

You can most definitely set up the shader as you please.
Me, i'd rely on the microfacet fresnel of the vrayMtl, rather than the poor fresnel of a texture, and work in a way which will allow full recomping latitude (which blend may not allow, depending on setup.).
If i have to pay for the price of figuring out my coat values.
Also, i *deeply* dislike full-on values in any shader, since time immemorial: masking with a fresnel makes no difference at grazing angles, and then fireflies appear.
Cue people cutting energy values and clamping, when in fact it was always a case of a not well thought out shader set up.

Thankfully, that's none of my business anymore, so no one has to suffer my setups.

[kosso_olli]

Hi Oliver,

After developer analysis, it has been concluded that this is the current expected result. Before V-Ray 5, clear coat specular highlights were considered negligible, because the clear coat was added on a pre-measured IOR value - hence the CC properties were not fully recreated. Currently, however, we've improved the ability to scan highly reflective materials which rely on the clear coat rendering. We can directly compare the rendering result with the images in the raw scan, and the goal is to have as much similarity as possible.

Sorry, this explanation does not make any sense to me. I used a carpaint which was digitized a few years ago, without the "improved" clearcoat. So I expect it to render the same in Next and 5.
Also, I can not sell a blurred clear coat to my clients. Reflections in clear coat just do not look this way. In the examples of the first post, the reflections are just a blurred mess.

To be fair, carpaint can be surprisingly shit in terms of blur and orange peel, but clients often don't want that in the paint. If there was a way to override that and have an option for totally clean and crisp coat, then that'd be useful. Cant the coat be disabled in the scan and then added back on in a blend material?

Well, you can disable the orange peel by setting the bump amount to 0. But you can not control the blurriness. Clearcoat is not blurry, I do not know why it looks that bad in 5. I strongly consider this as a bug, not a feature. One thing more on our list of why to avoid 5 so far.

[Ivaylo Katev]

Hi Oliver,
the old scans were set to very high glossines,and even with the changes the should remain sharp enough.
If you have already sent a scene, i will investigate the material and why is thishappening, if not, you can just tell me the material name

[kosso_olli]

Hi Oliver,
the old scans were set to very high glossines,and even with the changes the should remain sharp enough.
If you have already sent a scene, i will investigate the material and why is thishappening, if not, you can just tell me the material name

Hi Ivaylo,
here is the answer from Aleksandar Hadzhiev: "Thank you for the report and the provided files. There seems to be a difference, indeed. I have logged the issue (internal bug-tracker id: VMAX-10122) for developer analysis."

If you need anything else, please don't hesitate to ask.

[Ivaylo Katev]

just to confirm: the problematic material on the images is carpaint_grey_2_s, right?