Messages

1

Substance Painter - Showcase - ACESFilm 2.0+ LUTs for Substance Painter (+ 4 filmic tonemappers)

 on: March 01, 2019, 03:33:41 pm 

A set of tonemappers starting from "ACESFilm 2.0" lut which is the same ACES tonemapper found in Marmoset Toolbag 3. The rationale is flawed though, so for correct ACES I recommend following my PBR+ACES workflow here, and use the "ACESFilm - ACEScg"  lut included in this pack.
If you are not into ACES or that's too much of a hassle you might probably better off using the 4 non-ACES tonemappers. One is Marmoset matching Hejl tonemapper, and the other 3 Hable's "Uncharted 2" filmic tonemappers shaped by MJP and Thomas Mansecal (two of them matching Blender's Filmic High and Medium High).

They are divided by bright and dark (depending on HDRI dynamic range -interior/exterior-), and flat or filmic profile more prone for working or showcasing.

Download from my ArtStation.



Originally ACESFilm lut was created to fix an existent problem on another ACES lut (now already fixed)



The 4 new tonemappers back to back. 2 dark, 2 bright; 2 flat, 2 filmic. Using studio05 environment without post-processing. I converted the Filmic tonemappers to .frag files so you can also use them in Marmoset Toolbag (and get a unified look across Substance Painter and Blender if you wish)

2

Substance Painter - Feature Requests - Re: LUT Color Profiles

 on: February 15, 2019, 08:09:43 pm 

That's actually what I did but all within Substance Painter. I initially made a LUT (ACESFilm) from linear sRGB to sRGB passing through the RRT but the RRT always assumes ACEScg space so results turn out a bit darker. A better way is as you pointed to convert textures to ACEScg then pass it through a proper RRT lut (ACESFilm - ACEScg). You can find details here. I also made an automatic "PBR SmartFit" filter that works along the ACEScg filter (for PBR ACEScg textures).

3

Substance Painter - Showcase - Substance Painter: Filters for ACES rendering and Smart PBR conversion

 on: February 06, 2019, 12:12:42 pm 

Two new filters to treat color textures in physical terms. PBR_SmartFit will automatically bring your albedo map into PBR legal range retaining source colorimetry, details and perceptual middle gray, it soft-interpolates the out-of-PBR side so the transform is smooth (no range clamp like "PBR Albedo Safe Color" node). The ACES_ACEScg filter will convert your color maps into the ACEScg color space (AP1 Primaries) which is a space that closely matches and encompasses the Pointer's Gamut, a reference gamut for real surface colors.
These two can be used separately or in conjunction which is the preferred method for physically accurate rendering. By feeding ACEScg maps into the Substance rendering engine the render output is in ACEScg color space so for that matter I made a new ACESFilm lut that converts the render scene-referred linear space to "Output - sRGB" display-referred space. So for all intents and purposes now we are rendering in ACES within Substance Painter. Substance Designer has to wait since currently there is no lut support.
*The ACES_ACEScg filter is based on an implementation of the work of Stephen Hill (@self_shadow).


Some details:

The ACESFilm lut that I shared 2 months ago, even the one updated a week ago, and Brian Leleux ACES lut usually produce dark renders, this comes as no surprise as we are applying the ACES RRT to textures that haven't been passed through an inverse RRT, which most IDTs (Input Device Transform) implement. The ACES_ACEScg filter does the inverse RRT hence we get correct exposure and light interaction.

Take note on your HDRI color space. This is something not usually known but there are hints. If it appears undersaturated most likely the HDRI is not in sRGB but AdobeRGB or RAW. In this case you can convert externally the HDRI from AdobeRGB to ACEScg, this is cumbersome but since the AdobeRGB space is just a bit smaller than the ACEScg if you like what you see you can get away without converting anything, specially when image desaturation is something also frequent when making HDRI in certain applications.


Download with instructions in my Artstation



Own model; 1) Default in SP, 2) with ACESFilm lut (too dark), 3) same but raising gamma (notice the gain in contrast for textures and washed out env), 4) render with ACEScg filter and ACEScg lut (correct exposure, texture and lighting)



Other examples of better light interaction using ACES






PBR_SmartFit: Automatic fit for several Albedo maps



Node tree for the main part of the PBR_SmartFit filter.



ACES test render in Arnold (surface shader) between illegal texture range (HDR), illegal PBR range (0-1), and legal PBR range. IDT and ODT "Output - sRGB"

4

Substance Painter - Technical Support - Re: Pixel Processor filter with position functions into Substance Painter

 on: January 18, 2019, 12:09:14 pm 

I had to explicitly override as you said. I exported it as Inherit from Parent thinking it would read Painter's RGB16F setting.
Since cooked filters don't pass Parameter labels I thought on defining them on Description but this is single lined, is there a carriage return command to sort parameters by lines?

Also do you know any paper where I can read a bit more insight about PBR ranges? For simplicity I stuck at 30 as minimum sRGB but never understood what 50 meant, is it more scientifically accurate?

5

Substance Painter - Technical Support - Re: Pixel Processor filter with position functions into Substance Painter

 on: January 17, 2019, 03:52:28 pm 

I made a version of "PBR Albedo SafeColor" but with a refit function instead of clamping out-of-PBR-range values. To detect the Min and Max pixel values I reused some nodes from AutoLevels and everything works fine within Designer. When published my filter to Painter it doesn't work (too dark or bright Albedo, no functional behavior). It works (or seems to work) when presented with "raster" data like plain colors or textures. Layers or complex materials fail to work.

To test that the offending nodes were the ones computing the Min/Max I explicitly defined some random Min/Max values and more or less (some banding issues aside) it behave as expected. A good way to test is to publish the AutoLevels filter and apply it to Color channel in Painter, it doesn't work correctly, so I'm wondering if there are compatibility issues with some nodes.

6

Substance Painter - Technical Support - Re: Pixel Processor filter with position functions into Substance Painter

 on: January 15, 2019, 07:12:11 pm 

I know how it works I made an ACEScg and PBR SafeFit filter, works fine in SD but not in SP so I think that some Pixel Processor nodes are unsupported in SP. This is the last bit refraining me from release. Is there a compatibility list somewhere?

7

Substance Painter - Technical Support - Pixel Processor filter with position functions into Substance Painter

 on: January 14, 2019, 12:56:55 pm 

I'm trying to port a PBR filter I alread made in Designer using the Pixel Processor, it works there but when imported into SP it renders white. I suspect after some tests it has to do with position arguments, but hopefully someone here can confirm or if there are workarounds.
It works if I use a fill layer with a base color or texture but not with the Material Mode swatch.

8

Substance Academy - Substance Academy - Beginner - Tiling Substance (yes again)

 on: June 27, 2018, 10:05:28 pm 

In Substance Share there's a Substance called Rock Moss, the presentation ball depicts what seems like a cliff substance, but when you open the sbs what you see is basically a composition of 2 rocks, so my question is how people tile substances?

"Make it Tile" etc reuse the texture to scatter it randomly so this is not an option since I want procedural tiling, I checked on the substance options and there's not an exposed parameter to tile it either in the sbs or sbsar. So the only option is to go to each node (starting on the Tile Random node) and start raising the tiling parameters, but this is very time consuming and not elegant. Any help is welcome.

9

Substance Painter - Discussions - Configuring Settings

 on: June 16, 2018, 10:26:06 pm 

I'm having some issues saving settings in Painter, first I want to default the library view as large, another issue I have is hotkeys keep reseting themselves, for example I use another hotkey for symmetry and L for LazyMouse, when I reopen now symmetry defaults back to L overriding my LazyMouse hotkey (which keeps the L hotkey but has no priority on it).

Also as a side question, is it possible to work on the smooth areas (opposite of curvature), I tried inverting the curvature mask and used levels to adjust further but it doesn't change much. I'm using the AO map now but I don't like the fact that it creates a smooth gradient mask.

10

Substance Designer - Feature Requests - Re: Better Color Management

 on: May 26, 2018, 05:34:51 pm 

Same here, I would like to keep a Rec2020 space workflow, you know real world pointer's gamut. sRGB is like capping all the goodness that you created in SD.

11

Substance Designer - Technical Support - linear workflow in Designer

 on: May 20, 2018, 11:02:27 pm 

Hello, I'm a bit confused on color spaces in Designer, for color nodes you've got sRGB (2.2 sRGB I assume?) and float. Enabling or disabling this has no effect on the texture output, it looks the same, in the other hand if I use a "Convert to linear/sRGB" node just before output then the appearance changes. Well, then what is the color space applied on color nodes to begin with?
My workflow would be float+sRGB enabled (linear sRGB I suppose as in Maya file node), export as is directly and apply linear sRGB in Maya (texture is 16bit), but the texture looks washed out, unless I select sRGB (sRGB 2.2) on Maya or add "Convert to linear" node in Designer, isn't this redundant?