Matching rec 709 QT FLAME / NUKE (Output - Rec 709 (color_picking)

Anyone know how to match the rec 709 output of Nuke?
In Nuke when exporting a QT the colorspace can be “color-picking” Output-Rec709, but not sure exactly what it’s doing…
I’d like to export the QT’s to match the NUke output from Flame (everything else in our pipeline upto this point matches.

1 Like

output-709 and color picking sounds like its aces709 :slight_smile: so you would have to export aces709 and that should be it.

If its the oldschool 709 from nuke-default then you have to use something called “709OETF” in flame its somewhere…

It probably is aces 709 but is this a preset LUT?

Yea you could call it a preset lut, nuke always relies on a working space and read space so you have to set flame up in aces mode if nuke is that allready to get a match.

Basically the aces709 is a “odt+rrt” internally its basiclly doing
Linear to 709 with tonemapping , it is not your straight up linear to display transfom which is only gamma it adds a highlight rolloff, and a gamut mapping from aces to 709 thats why your input should also be aces or be part of the aces workflow,

Nuke imports .exr plate as “scene linear” which means acesCG in aces mode.
then Transfers it to acesCG (working space) and then goes out using the rrt+odt which then is the 709 image you are looking at.

so yea take a look at the whole image pipeline inside nuke and flame its not just the export setting that matters :slight_smile:

1 Like

I believe we have the pipeline matching
Nuke ACES 2065 read input in ACES cg workspace. all fine and matches what we are doing in nuke.
Its just writing out a QT From Nuke (output rec709 colorpicking) and exporting a matching Flame QT that’s the issue.
do you know how to handle the tonemapping and roll off?

I’ll ask Foundry and report back…

they are for sure the same if you have both set to aces. and are doing the same things i am using flame aces everyday in tandem with nuke aces its the same stuff so that rolloff etc will just be there aumatically when you choose 709 on output in flame.

How is the difference visually , can you describe it? If its really minor it might be something like the aces versions used. (if you use something pre aces 1.0 I think the odts where a bit different).

Other than that Could it be just video/data levels mismatch? how are you comparing the QTs, maybe its metadata mismatch when viewing in quicktime player?

I mean thats one of the main points if aces to make your life easier in that regard :smiley:

I can post a picture in a moment but it seems like highlights are off

is your timeline set to 16bit?

I’m not exporting from timeline, just a single shot through media hub from library

hmm does it look like the flame output has softer/clipped highlights for some reason? thats usually happening when something in the chain isnt 16bit float

Just something I have seen a lot and it reminded me of this when you said highlights

You can use these 3 testimages to test your pipeline they are all aces AP0 aka 2065-1

this has a bunch of verified results in Rec709:

2 Likes

Thanks, I’ll check it out.
If I hear anything from Foundry I will post back.
My images look like they match before exporting to media hub so will try with test images
Thanks

Prores from both flame and Nuke match for me

What are your export settings in flame?

Here are my import settings , just dragging them to the libray like that

and then here are my export settings:

and my nuke settings:

2 Likes

Thanks so much Finn for the charts etc. Was able to thoroughly test pipeline.
We are using DNxHR HQ and with the synthetic charts we can see differences but probably down to compression differences, although tested with prores 444 and seems like it’s the same…
will investigate

yea the syntethic testchart is really taxing and will show once any codec falls apart. its the most brutal pixel values youll find anywhere :smiley:
With Prores4444 the difference isnt as much for me at least on the “stilllife” sample the difference is pitch black.

Also remeber that nuke is 32bit under the hood which makes a difference when doing colorspace conversions :slight_smile:

this is when I compare nuke422HQ against flame444 you can really see how 444 makes a lot of difference

This would be the difference between

2 Likes

Thanks Finn, If everyone is getting the same then I’m going stop trying to fix it!
Really appreciate the input!

Craig

1 Like

“as long as its not visible” is kinda the point with 422 image compression :smiley: cheers!

2 Likes

Brilliant thread - very good transferable to other color matching topics. Finn’s inputs always appreciated by its structure and appropriate result. Finn for colour president!

1 Like

aww <3 I love colors but there are way more crazy colorscientist out there that I cant hold a candle to

2 Likes

“We only know what we know when we need to know it.” (Dave Snowden)

1 Like