Notice of Meeting - ACES Output Transforms Architecture VWG - Meeting #3 - 1/13/2021

ACES Output Transforms Architecture VWG Meeting #3

Wednesday, January 13, 2021
11:00am - 12:00pm Pacific Time (UTC-7:00pm)

Please join us for the next meeting of this virtual working group (VWG). Future meeting dates for this month include:

  • 1/20/21
  • 1/27/21

Dropbox Paper link for this group:

We will be using the same GoToMeeting url and phone numbers as in previous groups.
You may join via computer/smartphone (preferred) which will allow you to see any presentations or documents that are shared or you can join using a telephone which will be an audio only experience.

Please note that meetings are recorded and transcribed and open to the public. By participating you are agreeing to the ACESCentral Virtual Working Group Participation Guidelines

Audio + Video
Please join my meeting from your computer, tablet or smartphone.
https://global.gotomeeting.com/join/241798885

First GoToMeeting? Let’s do a quick system check: https://link.gotomeeting.com/system-check

Audio Only
You can also dial in using your phone.
Dial the closest number to your location and then follow the prompts to enter the access code.
United States: +1 (669) 224-3319
Access Code: 241-798-885

More phone numbers
Australia: +61 2 8355 1038
Austria: +43 7 2081 5337
Belgium: +32 28 93 7002
Canada: +1 (647) 497-9379
Denmark: +45 32 72 03 69
Finland: +358 923 17 0556
France: +33 170 950 590
Germany: +49 692 5736 7300
Ireland: +353 15 360 756
Italy: +39 0 230 57 81 80
Netherlands: +31 207 941 375
New Zealand: +64 9 913 2226
Norway: +47 21 93 37 37
Spain: +34 932 75 1230
Sweden: +46 853 527 818
Switzerland: +41 225 4599 60
United Kingdom: +44 330 221 0097

1 Like

Hi,

To go back to the posterization topic that I quickly jumped on and so that we agree on what we are talking about, here is the Wikipedia definition:

Posterization or posterisation of an image entails the conversion of a continuous gradation of tone to several regions of fewer tones, with abrupt changes from one tone to another. This was originally done with photographic processes to create posters. It can now be done photographically or with digital image processing and may be deliberate or an unintended artifact of color quantization.

With that in mind, I took @ChrisBrejon’s lightsaber image which I think was not processed correctly:

ACEScg + sRGB ODT

Incorrectly Processed?

And then I extracted a cross-section because even if my eyes could not see posterization, maybe the analytical or scientific method if you want to call it like that would!

Well, I’m happy to announce that my old eyes are still in good shape!

And with a zoom on the red character region, just in case there is a trap:

Mostly noise from the render, nothing-to-see, moving on.

Cheers,

Thomas


Bonus Point: Here is the above signal quantized to 7-bit, i.e. posterized:

Thanks @Thomas_Mansencal ! Your insight, as always, is much appreciated.

No they aren’t. It is just a really poor naming convention. They are AP0 exr files. Sorry about that.

If we have a look at the name of the file : cg_ACEScg_lightSabers_001_aces.exr, I’ll explain what I tried to attempt (obviously an epic failure on my end) :slight_smile:

  • cg -> prefix to indicate that this image has been generated using computer graphics
  • ACEScg -> primaries used for the lights
  • lightSabers -> the “topic”
  • aces -> the colorspace alias (it could have been “lin_ap0”)

Since AP0 is the format for exchanging files, I tried to follow this rule. Obivously the fact that “ACEScg” is present in the name of the file is a terrible idea. New file are on their way with a better naming convention.

Having said that, I do not think that interpreting this exr as AP0 (instead of AP1) would change your results, right ?

Just wanted to say that I agree about the need of precise terminology to name properly the errors the VWG OT needs to fix. I’ll try to be more accurate in my next posts, both in terminology and ctl processing.

For the record, we did manage to generate some ctl reference render for the light sabers with the help of @jedsmith

Reference render

Nuke implementation by Jed Smith (perfect match)

OCIO config

I think it is very interesting to observe the differences between OCIO and CTL on the blue neons.

On a side note, here is a couple of thoughts I had during this meeting :

  1. define exactly what posterization, banding, gamut clip, hue skews and mach bands are, with clear examples illustrating them (no ambiguity !)
  2. try to list where these issues may come from
  3. only provide examples using CTL or the ACES nuke nodes from Jed Smith (and not using OCIO V1)
  4. One “advantage” we have with full CG renders is that we are not IDTs dependent in a way and I thought it would be worth mentioning. The ACEScg primaries used in this render will show perfectly the issues we’re trying to fix I think (I just need to name them properly).

Regards,
Chris

So in your terms, posterization is purely numerical, and not the perceptual implication of the numerical output?

If we have a 64 bit magical display, and all the values oscillate at a threshold below the perceptual implications, it’s not posterized under your definition.

Not sure where you read that, I merely re-posted Wikipedia definition which is the one commonly agreed on, not one I invented. Does not require much more than looking on Google Images actually. :slight_smile:

It is simply asking the question as to what the defining baseline is for posterization.

If we had an infinite bit depth, could an image be posterized?

IE: Is it purely bit depth or the perceptual implications?