Aces 1.0.3 in Nuke

[Raphael]

Hello,

trying to use Aces for a wide & sRGB mixed gamut pipeline with various different inputs. (CG in srgb linear; sRGB(lut & gamut) tiffs; AlexaLog, GoPro)

Comparing visually various file formats of Aces 1.0.3 configs VS nuke-default  (on the same monitor with two instances of nuke) there is a significant visual discrepancy that is roughly corrected with a gamma of 1.4 on the sRGB output

  - !<ColorSpace>

    name: Output - sRGBcg

    family: Output

    equalitygroup: ""

    bitdepth: 32f

    description: |

      ACES 1.0 Output - sRGB Output Transform

      

      ACES Transform ID : ODT.Academy.RGBmonitor_100nits_dim.a1.0.3

    isdata: false

    allocation: uniform

    allocationvars: [0, 1]

    to_reference: !<GroupTransform>

      children:

        - !<FileTransform> {src: InvRRT.sRGB.Log2_48_nits_Shaper.spi3d, interpolation: tetrahedral}

        - !<FileTransform> {src: Log2_48_nits_Shaper_to_linear.spi1d, interpolation: linear}

        - !<ExponentTransform> {value: [1.41582377, 1.41582377, 1.41582377, 1]} 

    from_reference: !<GroupTransform>

      children:  

        - !<ExponentTransform> {value: [0.70630259, 0.70630259, 0.70630259, 1]}      

        - !<FileTransform> {src: Log2_48_nits_Shaper_to_linear.spi1d, interpolation: linear, direction: inverse}

        - !<FileTransform> {src: Log2_48_nits_Shaper.RRT.sRGB.spi3d, interpolation: tetrahedral} 

Is there a more accurate way of doing this with the Aces 1.0.3 configs? 

Perhaps a better approach would be to extend the nuke-default config? 

i.e set a gamut target for the 'linear' working space (i.e. arri wide); generate a matching sRGB display; and generate ocio colourspaces for all the scene inputs to this arri wide linear working space?

[Deke Kincaid]

These won't match nor should they.  Nuke-default does a straight linear conversion and has no concept of primaries(they match the source).  ACES does both conversions.  Also the ACES RRT is very different then Nuke's very basic sRGB viewer lut as there isn't any tone mapping.

--
You received this message because you are subscribed to the Google Groups "OpenColorIO Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Raphael]

Yes I understand a mixed gamut workflow is  never going to be the same...
  a gamma shift of approx. 1.4 between the two configs to compensate where midtown grey is?!

I don't understand how, like this, the 1.0.3 config in nuke will work?

If one was going compensate so the two configs are in the same ballpark, side by side on a monitor.
What's the best way of doing this, please?

[Deke Kincaid]

Which IDT were you using on the image?

--

[Kevin Wheatley]

I think there is a miss-understanding about what the 'sRGB' labels means in the two situations (same of Rec 709 only potentially worse) in Nuke's built in case it means the non-linear encoding assuming your linear image has sRGB primaries and gamut.

In the OCIO config case it means, the result of taking a linear image in your reference space primaries and whitepoint and applying an image rendering (the RRT), followed by the ODT, part of which encodes a transfer function appropriate for displaying on an sRGB monitor.

These two things are not the same and are not intended to be, even the 'simple' sRGB setting in the OCIO config has chromatic adaptation applied to it using the simplified Bradford transform.

If you want to use images with and without an image rendering you would need to "assume" to back out the rendering in the sRGB images via the inverse of the RRT, though by the sounds of things this isn't what you want, instead you want the simple sRGB option "Input - Generic - sRGB - Texture" or something like that.

Kevin

​

[ra...@freefolk.com]

On Thursday, 16 February 2017 08:14:57 UTC, Deke Kincaid wrote:

Which IDT were you using on the image?

===========================

IDT:

- Plate A: .dpx (AlexaV3LogC tonemap & primary)

-- Input - ARRI - V3 LogC (EI800) - Wide Gamut

- Plate B: .mov (GH4 V-Log)

-- Input - Panasonic - V-Log - V-Gamut

- CG Renders .exr (linear tonemap; sRGB primary)

-- Utility - Linear - sRGB

- Concept Matte Paintings .jpg (sRGB tonemap & primary)

-- Utility - sRGB - Texture

===========================

ODT

Output - sRGB (adjusted with a gamma offset to match nuke defaults)

===========================

Working space

ACEScg

See our latest showreels

Freefolk 140 Wardour Street, London W1F 8ZT

freefolk.com  / +44 (0)20 7734 9110

@freefolkstudios  /  facebook/freefolkstudios  /  instagram/freefolkstudios  /  linkedIn

[Deke Kincaid]

It looks like you are using ACES correctly on the IDT end.  If you are determined to make a similar “look” to match Nukes default sRGB viewer.  The proper way would be to add an LMT/look transform just before the RRT/ODT.  In Nuke you could use an OCIOCDLTransform, get the vales you want and then make a look in your config.ocio with these values and add it to a viewer/dispay. 

http://opencolorio.org/userguide/looks.html

--

[ra...@freefolk.com]

On Thursday, 16 February 2017 17:29:45 UTC, Deke Kincaid wrote:

It looks like you are using ACES correctly on the IDT end.  If you are determined to make a similar “look” to match Nukes default sRGB viewer.  The proper way would be to add an LMT/look transform just before the RRT/ODT.  In Nuke you could use an OCIOCDLTransform, get the vales you want and then make a look in your config.ocio with these values and add it to a viewer/dispay. 

Deke. Thanks for the Look suggestion. That improves the situation greatly (I made a simple 1d lut to match - with identical RGB curves... a CDL was not doing it.) 

But I'm not sure that a lut before the RRT is the best place for this as it does give me some slight gamut shifts. Wouldn't it be best to put this lut between the RRT and the ODT, or after the ODT?

Here is a screen grab visually describing the difference between the ACES 1.0.3 config and the nuke-default: 

The nuke default config visually matches IDT/ODT in most other VFX applications when they are using their defaults - it's the ACES 1.0.3 display that's the odd one out.

[Deke Kincaid]

Aces only allows an LMT before the RRT, otherwise you are not working in legal/reproducable ACES anymore.  

Just curious, why do you want to repro Nuke’s basic(crappy?) sRGB?  It has no concept of tone mapping for the display so it doesn’t roll off highlights. It just blatently clamps it which is quite ugly to the eye.

--

[ra...@freefolk.com]

On Tuesday, 21 February 2017 07:03:23 UTC, Deke Kincaid wrote:

Aces only allows an LMT before the RRT, otherwise you are not working in legal/reproducable ACES anymore.  

Just curious, why do you want to repro Nuke’s basic(crappy?) sRGB?  It has no concept of tone mapping for the display so it doesn’t roll off highlights. It just blatently clamps it which is quite ugly to the eye.

...because like beauty, truth lies somewhere in the middle. Although it be a lie:  until all other apps in the vfx pipe use OCIO.... We have to match the look to the lowest common version of what is right. Creating a OCIO ACES pipe that visually matches the common colour pipe, is a pipe that everyone is prepared to use. And that is progress.

besides that caveat. The default sRGB ODT is still not correct in my point of view. Compare the Alexa image - that's using the Alexa lut (not nukes default gamma display). Just like all the other IDT images, the ACES ODT is also too dark in the mid & dark greys.

here is my WIP attempt at a fix (transform space is ACEScct) - attached LUT

[ra...@freefolk.com]

just discovered that OCIO display 'looks' don't work in nukestudio 10.5v2 doh! bug reported to the foundry

(they do work in nuke comp)

[Raphael]

for some reason the attached images in this thread have disappeared?

so here are external links:
comparison with look adjustment
comparison without look adjustment

[Deke Kincaid]

What do you mean it doesn’t support 1.0.3?  Just download it from the AMPAS github.

https://github.com/ampas/OpenColorIO-Configs

On Fri, Nov 17, 2017 at 15:56 <ariel.w...@gmail.com> wrote:

Hi guys, I'm new to ACES. I received some test images that are encoded in ACEScct. How do I work with that in NUKE? My OCIO does not support aces 1.0.3.

best regards

Ariel

--

You received this message because you are subscribed to the Google Groups "OpenColorIO Users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+...@googlegroups.com.

[Andy C]

This footages are exported from the FilmLight Product.You could install a FilmLight Nuke Plugin

To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "OpenColorIO Users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+unsubscribe@googlegroups.com.

[ariel.w...@gmail.com]

Thanks Deke, I'll download the new version. 

[ariel.w...@gmail.com]

Thanks Andy. I'll check into that.

To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpenColorIO Users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+...@googlegroups.com.

[ariel.w...@gmail.com]

I've check the nuke plugin, but this is for importing grades done in Baselight into nuke. That is not what I need. I just wanted to know the proper workflow for working with ACEScct inside nuke. 

From what I found I just need to use this workflow, but with ACEScct instead of ACEScg:

http://www.nukepedia.com/written-tutorials/aces-for-nuke-via-ocio-part-1-initial-setup

best regards

Ariel Wollinger

[Diogo Girondi]

In theory it isn't recommended to use anything AP1 as a interchange or archival format. But if you must download the OCIO aces_1.0.3 config from GitHub and load it up in Nuke.

Then perhaps go from ACEScct to ACEScg and then back. But personally I would just ask for ACES 2065-1 encoded files instead.

cheers,

Diogo Girondi

To unsubscribe from this group and stop receiving emails from it, send an email to ocio-users+unsubscribe@googlegroups.com.