Request- Film Response Curves (Comparing Octane to V-Ray)

[grantwarwick]

Can anybody explain what the "Covert to log space before applying LUT" option is doing. I'm assuming log is a mode like SRGB?
I can't seem to get the same result in photoshop applying the same LUT

[^Lele^]

I'm just wondering : When does anyone has asked to BAKE the LUT profile into the render?

DISPLAY luts aren't baked.
Color Mapping which is clipped to 1.0 IS baked, and into a clipped image at that (the curve itself contains no info on how to look up values above 1.0).
Which is what those curves, and Octane, do (you can NOT apply the curves and have an unclamped image. Not in the VFB, not saved on disk. Apply the curves, and the image is clamped.).

As for the luts list, here's a few questions i'd need answered to mock a UI up, ideally with each point backed up by example workflows and proper documentation, so that evaluating how to proceed is clear:

Can you name a few general purpose LuTs you'd like added?
I'm sure i could find and convert them in nuke for you to load in the VRay VFB.
Once those are done, would you still want a dropdown to load them?
Would you want a warning when the loading of a LuT changes your display color space to Log?
Should there be a checkbox to make sure the output is linear?
What should be done when a LUT is loaded into the LUT dialog and another is preselected?
How would you make very clear which of the two is applied, or if they're being applied chained, and if they are, which one is first and which one second (as order matters, very often)?
Would someone not want those presets converted to OCIO, and should we then not make sure we add the dropdown for OCIO descriptors too?
But then, should they match the ones for standard LuTs, or should they be of a different batch, given OCIO is more modern, and has less of an heritage in production to deal with?

Most of all, how would we do all that without making the VFB unwieldy and overcomplicated to use, catering for those users which have been crying out for an UI overhaul and simplification?

Right now, most of the complicated and potentially hazardous color science is relegated to fairly complicated steps across the UI, which insures only those with a precise need will bother, and one can suppose those users know what they are doing.
Add LuTs as a dropdown, and well, you are exposing that to your average joe which needs them like a fish needs a bicycle, and which will likely rue the limitations of such an approach (when mishandled and not backed up by proper post) and wonder why was it put there in the first place.
But the latter are my fears, not actual data.
The questions above are the questions i mulled when the proposal for a DDL of LuTs (1,2,3D whatever.) was suggested, and i still have no answer to any of them, or an answer which isn't a worry, or leads to major risks for very little potential return.

[^Lele^]

Can anybody explain what the "Covert to log space before applying LUT" option is doing. I'm assuming log is a mode like SRGB?
I can't seem to get the same result in photoshop applying the same LUT

I think my case can safely rest.

[^Lele^]

Edit: I am on the beta group and love reading through, but it's for greater minds I feel haha.
I'm not on the nightlies though!

Wot. i was using the wrong bitmap type.
Thanks!

edit: but it's still not finding the tiny sun (6px wide on a 4k dome).
I've attached a scene for you people to try and have that map cast the shadow it should from the HDR map (included, ofc. may need repathing).
Should you do a comparison with VRay (the dome is included) make sure you set the map intensity to 0.1 (up from 0.01, set like so to make the map actually visible in viewport).
White_Roomset.zip

[tsvetan.geshev]

Can anybody explain what the "Covert to log space before applying LUT" option is doing. I'm assuming log is a mode like SRGB?
I can't seem to get the same result in photoshop applying the same LUT

Converts the Base map to a logarithmic color space before applying the look up table.
As I know the logarithmic color space, more closely mimics the natural light response of film emulsions compared to the liner color space.
This is film scan color space which provides better response in the dark end of the spectrum and doesn't waste it on the dark end.

Hope this makes any sense.

[Macker]

Material scanning, on the other hand, is providing for some shockingly accurate matches, through a "simple" RGB color space, of the scanned materials versus the source (See Chaos Group's patent on the material scanner...).

Could you elaborate on this please? Are Chaos Group going to put into production a material scanner, or provide us with scanned materials, etc?

[Paul Oblomov]

I think the easiest route - is to hire C# code-guy, to write you as much as you need

BTW - I think, that there must be anyone, who is comfortable in coding for vray. We can even split the charges among several ppl, so it won't be that expensive. What ya think ? Like a group buy or smth...

[^Lele^]

I can't go too much into details.
What i can say, because it's public knowledge, is that there is a VRayScannedMtl shader in max which is able to load scanned material samples as scanned by the patented chaos group scanner (based on x-Rite devices, I believe.).
In the beta group we tested them, and the results were quite remarkable indeed (albeit with the scanned material approach limitations. Some of which are being creatively tackled.), particularly for very hard to model BRDFs (cloth, very complex, layered metals, and so on.).
How they will choose to market it, is something I don't know.

[3LP]

So let's assume you have a list of LUT presets and you choose one that you really like, then save out the image. You then need to apply the LUT in some other software, so how are you going to access the LUT's that are in the vfb presets list? Are they seperate files that are installed with vray? If so, why not just do a google search for film LUT's and download some free ones, or buy some?

What you're asking for is a set of free LUT's to be shipped with Vray really, rather than a presets panel?

Well as I wrote in my message earlier, I would have seen a folder containing all the presets so you can loads from any other software afterwards and have the same results

Can anybody explain what the "Covert to log space before applying LUT" option is doing. I'm assuming log is a mode like SRGB?
I can't seem to get the same result in photoshop applying the same LUT

I know Lele might laugh again, but did you try out in AFX, PS is really crap to manage colors and bit depth, AFX is the best one of Adobe suits to handle those.

DISPLAY luts aren't baked.
Color Mapping which is clipped to 1.0 IS baked, and into a clipped image at that (the curve itself contains no info on how to look up values above 1.0).
Which is what those curves, and Octane, do (you can NOT apply the curves and have an unclamped image. Not in the VFB, not saved on disk. Apply the curves, and the image is clamped.).

As for the luts list, here's a few questions i'd need answered to mock a UI up, ideally with each point backed up by example workflows and proper documentation, so that evaluating how to proceed is clear:

Can you name a few general purpose LuTs you'd like added?
I'm sure i could find and convert them in nuke for you to load in the VRay VFB.
Once those are done, would you still want a dropdown to load them?
Would you want a warning when the loading of a LuT changes your display color space to Log?
Should there be a checkbox to make sure the output is linear?
What should be done when a LUT is loaded into the LUT dialog and another is preselected?
How would you make very clear which of the two is applied, or if they're being applied chained, and if they are, which one is first and which one second (as order matters, very often)?
Would someone not want those presets converted to OCIO, and should we then not make sure we add the dropdown for OCIO descriptors too?
But then, should they match the ones for standard LuTs, or should they be of a different batch, given OCIO is more modern, and has less of an heritage in production to deal with?

Most of all, how would we do all that without making the VFB unwieldy and overcomplicated to use, catering for those users which have been crying out for an UI overhaul and simplification?

Right now, most of the complicated and potentially hazardous color science is relegated to fairly complicated steps across the UI, which insures only those with a precise need will bother, and one can suppose those users know what they are doing.
Add LuTs as a dropdown, and well, you are exposing that to your average joe which needs them like a fish needs a bicycle, and which will likely rue the limitations of such an approach (when mishandled and not backed up by proper post) and wonder why was it put there in the first place.
But the latter are my fears, not actual data.
The questions above are the questions i mulled when the proposal for a DDL of LuTs (1,2,3D whatever.) was suggested, and i still have no answer to any of them, or an answer which isn't a worry, or leads to major risks for very little potential return.

I'll need to re-read this when I have a bit more time to understand all what you just said as it's a bit over my pay-grade for the moment, but what I can say is :
* yes it's worth saving always to linear, and if a checkbox could be added to ensure that, I'd endorse that. Same goes for all the other post effects done in the VFB, for me they should not be saved out/baked in the saved exr, but I can understand why people would like to, so a checkbox could be the right call.
* apply lut should be kind of the same or as easy as going into Arion FX and apply the curves on top of your image, they make you swap between presets like octane does. You say it would be great to mix few of them at the same time? Well yes I didn't even think of that, and yes I guess the order does have importance in that case.
* Just thinking out loud here, but if those luts are just a set of curves of RGB, why can't we apply them in the curves straight then? Would that be better? convert them to *.bcurve and have a preset list of bcurve in stead? Then a simple tool to convert any *.ocio or *.cube to *.bcurve would make the trick.
* No idea of how we need to treat whatever goes over 1.0 float as I understand we don't have that data. But could we not just leave the shape of the curve roll? If you create two points in besier curve, the curve reacts and lives outside the boundaries of those two points.
* Although I understand why you'd want to keep a LUT locked of but some of us, (like me^^) use those just for artistic purpose, and often as a starting point. So once I've loaded one in, if I can move points around (like if they where curves) and be able to save them out to reload them into my next soft down the pipe, it would be even better

Stan

[grantwarwick]

I know Lele might laugh again, but did you try out in AFX, PS is really crap to manage colors and bit depth, AFX is the best one of Adobe suits to handle those.

Thanks!

Converts the Base map to a logarithmic color space before applying the look up table.
As I know the logarithmic color space, more closely mimics the natural light response of film emulsions compared to the liner color space.
This is film scan color space which provides better response in the dark end of the spectrum and doesn't waste it on the dark end.

Hope this makes any sense.

Forgive me if I sound stupid here, I'm tired as hell!
Basically, I load the LUT into V-Ray, it blows out. I untick SRGB (assuming that also plays a part in further distorting things from their intended purpose?)
The LUT now (just for this tests sake, looks pretty good and I'm happy with it) is "correctly" being previewed in the VFB.
So now, in AFX, I load the 32bit source file-
What exactly are the steps involved to reproduce what I've previewed down the bottom?
I feel like I'm reliving the multi layered material research here, it's endless but I'm on the brink of finding serious use for it :/
Cheers for any help.

I think my case can safely rest.

Bastard

[grantwarwick]

Wot. i was using the wrong bitmap type.
Thanks!

edit: but it's still not finding the tiny sun (6px wide on a 4k dome).
I've attached a scene for you people to try and have that map cast the shadow it should from the HDR map (included, ofc. may need repathing).
Should you do a comparison with VRay (the dome is included) make sure you set the map intensity to 0.1 (up from 0.01, set like so to make the map actually visible in viewport).
[ATTACH]25490[/ATTACH]

Can you save out as max 2014 and I'll take a look?

[dubcat]

I know it's kinda off-topic but I wanted to see how that V-Ray sky map behaved in Corona.

[3LP]

Can you post your vrimg and the .cube file please?
I'll see if I can get it working in AFX

[^Lele^]

Bastard

Ahah, i had it coming! :P

[grantwarwick]

Wot. i was using the wrong bitmap type.
Thanks!

edit: but it's still not finding the tiny sun (6px wide on a 4k dome).
I've attached a scene for you people to try and have that map cast the shadow it should from the HDR map (included, ofc. may need repathing).
Should you do a comparison with VRay (the dome is included) make sure you set the map intensity to 0.1 (up from 0.01, set like so to make the map actually visible in viewport).
[ATTACH]25490[/ATTACH]

Couldn't be bothered waiting. It's definitely there!