Hey Finn
Just ran a test of exporting out of flame and tagging just as rec2020 and in mediainfo it fills the transfer characteristics in as BT.2020 but in yours it fills it as PQ. Just wondering how you managed to do that? The only way I can get the transfer characteristics and everything to match yours is by exporting out of flame with it tagged as rec2100pq.
I just tagged my file as such exported it from flame and it kept its tag like that?
Idk , but Rec2100pq is fine as thtats just PQ with Rec2020 primaries anyhow 
2 Likes
I’ve definitely delivered HDR to Amazon before. Is the issue with your file or a sidecar xml?
Just a small caveat, I am replying whilst on a train after having a few drinks but think my reply should be ok.
It is important that the first clip of your timeline is tagged as P3-D65PQ as the first clip will embed itself as the colour display primaries.
When delivering to Amazon I work in p3-D65PQ until the point of delivery when I apply a P3-D65Pq to BT2100PQ input transform on output. Keep in mind though that the display transform characteristics will come from whatever the first clip in your timeline is. It will also include the Bt2100 colorimetry (Rec2100 is just Rec2020 with a defined gamma curve). You’ll actually hve 2 values in the metadata which will essentially define the output as P3D65PQ residing within a BT2100PQ bucket.
Apologies if this doesn’t make entire sense and my terminology is a bit off. I will blame the alcohol for any confusion. Hope it makes sense. I can create a file and copy and paste what the metadata should look like if that helps at all? That would need to wait until tomorrow
5 Likes
That wpuld be amazing, thank you
1 Like
I should also add that there are 2 trains of thought towards the HDR10 metadata sidecar file. Some people suggest that you should add the MaxFALL and MaxCLL of your graded material whilst others believe it should be the MaxFALL and MaxCLL of the calibrated grading monitor. I do the former after analysis via an automated QC system such as Vidchecker or QSCAN or using cm_analyse within the Dolby Professional Tools. Whilst I do that as that is what the majority of specs dictate as the process, I actually believe that it should be the levels of the grading monitor. That would stay true to what the media would look like on a calibrated monitor as opposed to the MaxCLL being extrapolated to fit the maximum nits of the monitor. From my understanding though, most displays ignore the HDR10 metadata anyway.
1 Like
how do you get flame to write mastering display metadata? i cant get it write anything into a prores?
Is this for an advert or for content for Prime
edit looked at jpeg of spec and saw it’s for VOD.
Ok so If I do this, first clip is tagged PQ-P3D65.
Then export with this →
Then I do not get any Mastering Display Metadata written at all? its just not even present in the Prores export, nothing, this is all the Metadata I get from that:
General
Complete name : /Users/finn/Movies/FlameMasteringDisplay.mov
Format : MPEG-4
Format profile : QuickTime
Codec ID : qt 2005.03 (qt )
File size : 9.05 MiB
Duration : 12 s 40 ms
Overall bit rate mode : Variable
Overall bit rate : 6 302 kb/s
Encoded date : UTC 2022-12-07 13:13:27
Tagged date : UTC 2022-12-07 13:13:27
Video
ID : 1
Format : ProRes
Format version : Version 0
Format profile : 422 HQ
Codec ID : apch
Duration : 12 s 40 ms
Bit rate mode : Variable
Bit rate : 4 763 kb/s
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 25.000 FPS
Color space : YUV
Chroma subsampling : 4:2:2
Scan type : Progressive
Bits/(Pixel*Frame) : 0.092
Stream size : 6.84 MiB (76%)
Writing library : Apple
Language : English
Encoded date : UTC 2022-12-07 13:13:27
Tagged date : UTC 2022-12-07 13:13:27
Color primaries : BT.2020
Transfer characteristics : PQ
Matrix coefficients : BT.2020 non-constant
Audio
ID : 3
Format : PCM
Format settings : Little / Signed
Codec ID : lpcm
Duration : 12 s 40 ms
Bit rate mode : Constant
Bit rate : 1 536 kb/s
Channel(s) : 2 channels
Channel layout : L R
Sampling rate : 48.0 kHz
Bit depth : 16 bits
Stream size : 2.20 MiB (24%)
Language : English
Encoded date : UTC 2022-12-07 13:13:28
Tagged date : UTC 2022-12-07 13:13:28
Other
ID : 2
Type : Time code
Format : QuickTime TC
Duration : 12 s 40 ms
Frame rate : 25.000 FPS
Time code of first frame : 01:00:00:00
Time code of last frame : 01:00:12:00
Time code, stripped : Yes
Title : testTone_finn_h264_from_resolve_macOS_AAC
Language : English
Encoded date : UTC 2022-12-07 13:13:27
Tagged date : UTC 2022-12-07 13:13:27
Even when I add a HDR track , and set my mastering display there, I get nothing in terms of prores metadata in regards to the mastering display, and I dont think I have ever seen this from flame, in the DoVi XML yes - I see it correctly
Yeah ive tried every way i can think out of Flame and never get those tags unfortunately.
I guess it will need to be added elsewhere?
thats what I thought, I just use amcdx video patcher to add them, but adam said the first clip will set the mastering monitor metadata , maybe only for the DOVi XML?
Hello.
Sorry for my delayed response. Finn is correct, that this is important when creating a Dolby Vision master. For HDR10 deliveries I actually have used CineX Meta to add the Mastering display color primary & Luminance information as well as the MaxFALL and MaxCLL values. There is a before and after below and a screenshot of CineX Meta. The metadata needs to look like the after (I put in some random numbers for MaxFALL + MaxCLL. I would usually analyse using something like Cm_analyse, VidChecker or QScan to get the correct numbers)
Apologies as I may or may not have been a tiny bit under the influence of alcohol on my initial response…
Yell out if you want an example of a HDR10 sidecar file as well.
cool thx @AdamArcher , havent heard of cineXmeta before good to have another alternative to amcdx video patcher, but its $200 a year and amcdx is free so…
One Q though, you chose DCI-P3 and not P3-D65, specs say P3-D65 though?
That’s because I fucked up as I was in a rush and not paying attention. It should be Display P3 (which is D65 white point). The irony being that I wanted to get back to thoroughly checking some colour pipeline tests but wanted to reply to this thread. But anyway…
The PQ is irrelevant as the metadata is about the color primaries, not the transfer function/gamma.
1 Like
I literally did the same thing and picked dci p3 but noticed and swapped my screenshots above 
Yea only thing would be the whitepoint that differs, same primaries but D65 vs D60ish WP for DCI-P3