Tl’DR
Set flame to rec709
Set macOS on the host to 709-A
set macOS on the XDR macbook Bt.1886 refernce mode
((tested using fluid 2.0 on jumpdesktop teams)
You are out in about in a hotel room and need to fix a grading remote? there are a lot of buttons to go through, so I went on a bit of a mission to find out what I need to be doing to get accurate monitoring, this is not something you want to be doing all day, probably, but its great to have some trust in your monitoring on those occasions.
Newest jumpdesktop beta with fluid 2.0 can do 4:4:4 and has great display res matching but not as good hotkey support as parsec so ill be conectraing on that using the ULTRA color mode which is 4:4:4 10bit
Now, i havent 100% gotten this to be super accurate without doing super weird shenanigans like custom reference profiles because the whole imaging pipeline is SUPER broken between flame and XDR and remote monitoring and display profiles, of course its better to have a actual cleanfeed - no doubt at all! But this is “the least broken” way to get this going quickly, and at least you are in a ballpark. its off by about 1 value in 8bit, so isntead of 72 i get 71 , its super close
Macbooks without XDR are color wise huge trash so i wont be looking at that either.
Host Display settings we are setting the system output to rec709 which is 1.961 Gamma, now idk why this particular profile works much better than the rec709-5 profile, something is special with that , i found it in : /System/Library/CoreServices/Resources/Profiles/HD\ 709-A.icc
Host Flame Settings, super simple leave it all as is, itll be fine wether you tag your files unknown or rec709 or whatever else you do, as long as you mange your media correctly as you should flame will just send that data to macOS so all is well
Macbook display settings:
Of course we are using the correct rec709 reference mode, thats the great thing about XDR!
it will also read jump desktops pixels as sRGB and then do a 1.22 gamma boost on top which just ends up beign where we want it.
My way of testing this in the field is to use a perceptual matching chart that i put in a “known good” state on my macbook, that I know is accurately showing me rec709 gamma 2.4 at 100NIT, then i load the same chart into flame on my remote machine and just simple compare them , if they dont look the same, some setting is trash, even better if you have a known good flanders or something to really try this or you can use a probe and measure patches, but that takes super long. You can also use the digital color meter to compare 2 apps guis or whatever with each other which is super handy,
the error I get is from a 8bit should be value of 72 on a midgrey patch i get 71, while on dark grey i get a 1:1 match , so something as I said seems ever so slightly off. there are many transforms this is just a slight rounding error by encoding the stuff h265 on the way via jumpdesktop so i wouldnt call this “super crucial accurate monitoring” or whatever but its the least shit ive seen .
Here is a additional video comparing known good resolve on my macbook vs flame on the remote machine , switching back and forth.
