Sorry, forgot to mention I used the Disney BSDF, and tweeked it a bit for SSS.
Here is the Filmic default & the sRGB default below it, using the default Disney BSDF (except for color).
I left the Display Device as is…“sRGB/BT.709”…is that right? I must have accidently changed it to “None”
to get the dark sRGB image in the first comparison.
Anyway, let me know if I got it right (or not) this time.
And when I asked if anyone wanted to tweek it for the “best” render,
didn’t mean the default, but to see what settings an experienced artist
would use with just this simple scene.
I still need try HDRI lighting & studio multi-light setup.
BTW, here’s a wiki article (doctoral dissertation?) by troy_s that may clear it up for me, IF I have, LOL, " the patience (to) get to the other end of the tunnel without a large degree of duress." Will need some serious caffeine.
The disney shader is of course “compatible”. Filmic is just an image transform, there are no compatability issues with anything. You can even render the scene on an install of Blender that doesn’t have filmic installed at all (eg a renderfarm).
BUT if you created the scene with filmic it will not look right if rendered to an srgb transform. Render to exr and apply filmic on your machine oder grade however you want it.
Taking an image like that and just comparing srgb to filmic mapping shows nothing. Try, using the filmic transform, really pushing a light source, until you get red color in the false color view. Maybe use a very strong colored shader in a primary color on the object. That should give you an image that you can compare to srgb/gamma 2.2 transform where you will distinctly see the advantage of filmic.
Oh sorry. I didnot do that.
I simply render image in Blender with default colormanagement settings (without having filmic stuff installed). I thought that was the comparison you wanted.
You mean SSS on Disney shader?. I do not know much about BPT configuration. If you want, you modify the scene shared by ‘3dcal’ configuring for BPT in a similar way to PT, send me it with a private message and I then do the test. Then I will edit this thread with the result to not continue off topic
I think whatever looks good to you is best for you. And same for me or anyone else.
Maybe there is an industry standard, but in the long run and the Big Scheme, it’s very subjective as to
what looks good. “Beauty is in the eye of the beholder.”
They look almost the same because light condition doesn’t bring up sRGB critical issue. Try to raise up exposure (or lamps power) and you’ll see where sRGB flaws.
“Beauty is in the eye of the beholder” sure, but all this imho is a technical matter, where sRGB is prone to screw color data, while Filmic Log allows to preserve colors under a wider light/exposure conditions.
Last off-topic on this, promise, cross on heart (i’d continue on the disney shader’s post or pm edit: done!)
you don’t need to do any special thing: render path, then switch to branched, raise AA samples and render again. In plain path you’ll see color desaturation in SSS zone, while in branched there’s only base color green as if it is ignoring the SSS color
They look almost the same because light condition doesn’t bring up sRGB critical issue. Try to raise up exposure (or lamps power) and you’ll see where sRGB flaws.
“Beauty is in the eye of the beholder” sure, but all this imho is a technical matter, where sRGB is prone to screw color data, while Filmic Log allows to preserve colors under a wider light/exposure conditions.
It’s hard to put your finger on what it is exactly, but Filmic renders definitely look better, and closer to insane stuff from Octane etc… Here’s a test I made.
Photox, I guess one could say art is subjective … yeah I guess that could be said. I actually prefer your default Cycles render here or sRGB in this render. It’s plausible realism and much more interesting to my eye. Which is surely no knock on Filmic Blender.
Here’s a thought if I might. Send these two off to friends and family who know nothing about CG and ask their opinion. In other words I would think the vast majority of average viewers of any of our efforts. That would interest me way more, and others here, then any opinion of mine.
Both images have its pluss, filmic one looks more natural, but some of the texture detail is lost. The sRGB one looks unatural strong (materials) but the texture detail is better preserved.
Hi troy_s.
The details of the floor in sRGB version look stronger (the texture is more noticeable).
Does that mean that starting from Filmic version saved in openEXR format, with postprocessing you could get exactly what you see in sRGB version without loss of textures?
Also the feeling of depth that can give the lights and shadows changes, this is noticeable in the green monkey, filmic version looks much flatter.
I agree with ‘esimacio’ about both images have its pluss
That power slider ideally is only used to check composites. Personal opinion, avoid those two sliders in the CM panel except when using them temporarily to test values. They should live at 0.0 and 1.0 respectively.
Adjust the lighting. Use False Colour as a light meter as desired.
The Filmic variant is overexposed. The rest is confusion.
Oh, ok. Would it be something similar to when you have an overexposed photo? Is there loss of information/data in overexposed areas? So according to I understand it is not possible to make direct comparisons between Filmic and sRGB without setting the lights appropriately for each case, right?
In this case, the data is sort of there, but the detail is compressed together near the upper end of the display referred encoded output.
You can still clip data, but here it is equivalent to comparing exposures versus comparing details at a given exposure. Part of the details will also be anchored in the aesthetic contrast chosen, plus any additional CDL tweaks one might put into the output.
Exactly.
If you do try this, you will find that your light ratios will need to be extended quite wide. In doing that, you will quickly find the sRGB EOTF falls apart entirely as a view transform. If you try to massage values to effectively be an entirely display referred bit of work, you will also find that your effort and consistency will suffer tremendously.
That is, it is far easier to simply live in the scene referred domain, keep your mental model there, and tweak under that context. When you do this, you will get plenty of yield for “free”.
Watch the Alex Fry video on ACES for a great demonstration. Though his discussion is specific to ACES, which has nothing to do with the actual transforms in Filmic, the conceptual framework is identical.
