I am having following issue, and maybe I am doing something wrong, but pls clarify:
- I export all the CC adjustments in the VFB as a .cube LUT file via the “Save all CC as LUT” function in the VFB
- Open the 32 Bit EXR in Photoshop
- Load the previously exported LUT
- Colors now match for the most parts but the highlights get greyish and clamped
What is the correct workflow to replicate the colors in Photoshop? The issue also exists with PDPlayer and result looks equally wrong like in Photoshop.
Basically the desired workflow is to add all renderelements (lighting, gi, spec, etc.) in Photoshop and then in the end should have a LUT so that it looks identical to the rendering in the VFB.
Attached 2 images to demonstrate the issue. Notice in Photoshop all highlights turn greyish and seem clamped.
thanks for your reply. Indeed I am using filmic tonemap, but I just tried to just put only an exposure and curve cc adjustment instead and I am having the same issue.
The colors always seem to match but as soon as something reaches a certain brightness it always clamps to grey.
As said same issue in Photoshop or PDPlayer (which is from Chaos), so I assume it’s not a Photoshop problem but rather something that wrong with the exported LUT or my workflow :)
So you can generate a LUT an get identical results in Photoshop as in VFB even in the bright parts?
Yes, we routinely save the LUT and apply it in Photoshop for identical results, both with V-Ray Next & V-Ray 5.
I have attached a zip containing a VFBL and the exported LUT so you can test at your end.
Let me know if the results match on your environment. cc.zip (89.1 KB)
Hi, thanks for your files. I tested and I initially get identical results with your CC settings and the provided LUT.
The problem is that as soon as I put the highlight burn value in the exposure slider below roughly 0,5 I will get these grey clamped highlights again when I export a LUT, while it looks fine in the VFB.
So it seems to me that if one is trying to compress the dynamic range either with:
- dicreasing the highlight burn value in the exposure CC and/or
- using the filmic tonemap
then those adjustments are not correctly transferred into the LUT and the hightlights are getting clamped. I think the procedure to compress the dynamic range is essential during the tonemapping procedure to get photographic results.
- Exported linear EXR without any colorcorrections
- PSD that uses this file with the created LUT file
- JPG as how the image should look like according to the VFB
- LUT and VFBL
Hi, so for propper tonemapping this step is quite essential:
Basically you are reducing the dynamic range of the linear image to something that can be displayed on your monitor. There are various ways to do this but the minimum tonemapping prodcedure would be to reduce the highlight burn (with 0 being for images with an extreme dynamic range, other images might be fine with much higher highlight burn values). This step will bring the colors/intensities in something that can be correctly viewed on your monitor. Since reducing the highlight burn value will naturally also flatten out your contrast the next step would be to bring the contrast back by using for example a curve correction or a lookup table.
So from my perspective there is hardly any project where I would not touch the highlight burn value, or use the filmic tonemap (which does this internally).
All of this works flawless in the VFB but somehow the transformations are falsified when saving out the LUT file from VFB.
@piotrus3333: Awesome, I think with the approach in your video it seems to eliminate those issues.
The requirement would be that the highlight burn is at the very bottom of the stack. Then, in order to save out the LUT from VFB you would need to disable it and then in Photoshop recreate it by using your ChannelMixer approach. Then load the previously exported LUT on top of it!
Do I see this correctly? From the test I did so far seems it works!
So far so good! But how if using Filmic Tonemap, I think in this case there would be no such easy fix for that or?
I just tried it with a more complex scene that uses filmic tonemap and its possible to get something pretty close with this workflow. Thanks again
So is this a Photoshop issue or an issue with the generated LUT? From how piotrus descibed it, it seems to be more an issue by Adobe rather than Chaos.
I think the idea would be that you can just export the LUT from VFB and load it into PS and it matches out of the box without having to fiddle around with this Channel Mixer and so on…
Again, is this an issue by Adobe or by Chaos?
this is something that was called Exponential color mapping inVRay’s old days. same as VRay’s Reinhard with Burn 0.
y=x/(x+1) where x is raw render pixel value and y is tone mapped result. no matter what the dynamic range this will bring everything into 0-1 range. most of the time it is called basic/simple Reinhard or something along those lines.
Generally this is an issue with the LUT file format itself.
It contains e few things:
1. Domain min and max values - the minimum and maximum values the LUT can correct. Problems here:
- Those values are omitted by default and it is assumed that LUT is for values [0,1]. It is not exactly certain if all softwares actually support other values.
- The problem here is what value to put actually. We can easily determine the current image boundaries but what if you want to apply the same LUT to multiple frames of an animation ?
2. Three dimensional table with a fixed size of N [2,256]. So the file will contain N x N x N sample points. The sample points are equidistantly spaced. This is the key limitation. It means that if the range is [0,1] and we have 32 points - we have values for 0, 0.0322, 0.0644 and so on. If the range is [0-100] we get 0, 3.2258, 6.4516 and if the range is [0,10000] we get values for 0, 322.58, 645.16 and so on..
So if we want good precision and big enough range - we’d have to export huge LUT files that will be quite slow to apply.
Currently we export a 3D grid only in the [0,1] range and use N=32.
