Do you know about RRT of ACES 1.3 Technical Documentation

ACES Resources & Education
1

Is there technical documentation for RRTs up to ACES 1.3?
I looked at the CTL, but could not understand what operation was being performed.

2

I am not aware of detailed technical documentation. But the comments in the CTL describe each step.

You don’t need to worry too much about the glow module and red modifier. These apply subtle subjective changes for preferred rendering, and slightly desaturate red hues.

The next step simply converts to AP1 primaries. And the next applies a slight overall desaturation. Then a tone curve is applied independently to the RGB channels. Then the image data is converted back to AP0 (now OCES - output referred AP0).

1 Like
3

There may be various home-brewed documentation but there is not a formal documentation of the 1.3 RRT.

The steps are relatively simple, and are segmented in the CTL:

  1. Glow module: This is a negligible modifier intended to boost colorfulness in dark colors - a la “film”
  2. Red modifier: This targets a specific hue region (reds), aiming to reduce colorfulness slightly because reds were deemed “too hot”
  3. AP0 to AP1: AP1 is the “rendering space” in which the tonescale is to be applied
  4. Global desaturation: A slight desaturation of all colors (0.96 factor)
  5. Apply the tonescale independently to RGB in AP1 space
  6. Re-encode the output back to AP0

Steps 1 & 2 could have been and probably should have been encapsulated in a “default LMT”.

1 Like
4

Thank you.
It helped me because I didn’t particularly understand what the Glow module meant.

LMT.Academy.ACES_0_1_1.ctl
LMT.Academy.ACES_0_2_2.ctl
LMT.Academy.ACES_0_7_1.ctl
Are these LMTs previous version RRT?

ODT.Academy.Rec709_100nits_dim.ctl
I looked at the following above code.
it used 48nits for the tone scale, why not 100nits?
And, The ssts function is used in some ODTs(HDR…etc), is this the same function that does the same thing as segmented_spline_c9_fwd function in ACESlib.Tonescales.ctl?
custom parameter version?

5

Those LMTs replicate the appearance of the earlier RRTs when using the 1.3 RRT.

Because ACES 1.x uses 48 nit cinema as the reference target for SDR. 100 nit video SDR is derived from that by applying a “dark to dim” (viewing environment) modification.

It broadly serves the same purpose, but is not the same thing. The SSTS is a “Single Stage Tone Scale” because the HDR Output Transforms are self contained, rather than being made up of the RRT plus an ODT. The combined effect of the segmented splines in the RRT and ODTs creates the overall tone scale for SDR Output Transforms. The result of using 100 nits as the peak white in the SSTS function is similar, but not exactly the same.

ACES 2.0 will unify all the Output Transforms, so there are not the separate approaches for SDR and HDR which exist in ACES 1.x.

1 Like
6

Because ACES 1.x uses 48 nit cinema as the reference target for SDR. 100 nit video SDR is derived from that by applying a “dark to dim” (viewing environment) modification.

Is below code also related to “dark to dim”?

    // Apply desaturation to compensate for luminance difference
    linearCV = mult_f3_f33( linearCV, ODT_SAT_MAT);```

It broadly serves the same purpose, but is not the same thing. The SSTS is a “Single Stage Tone Scale” because the HDR Output Transforms are self contained, rather than being made up of the RRT plus an ODT. The combined effect of the segmented splines in the RRT and ODTs creates the overall tone scale for SDR Output Transforms. The result of using 100 nits as the peak white in the SSTS function is similar, but not exactly the same.

so, the CDL contained in the ODT folder mean that RRT cdl is required and the CDL contained in the OutputTransform folder mean that RRT cdl is not required right?

7

Effectively, yes. Although strictly it is because of the change in peak luminance (see Hunt effect) from 48 to 100 nits, and the dark-to-dim is for the the change in surround environment.

For ACES 2.0 it was decided that because the effect of the dark-to-dim and luminance based desaturation are small, and their presence just causes confusion, it was best to remove them. If people feel that they need a difference between the 48 nit theatrical and 100 nit SDR video versions of a show, it becomes a creative choice to be added in a trim pass. So, for example, this and this are the same except for the encoding EOTFs (2.6 gamma and piecewise sRGB).

Correct