Having initiated several searches on the Internet for what I deem essential knowledge about the colour spaces of the DLOG-M and HLG "colour profiles" that can be used on the DJI Mavic 2 Pro, I cannot find anything definitive, precise, or knowledgeable.

First, I'm using Apple hard- and software to store, manage, and edit: an iMac 27inch 5K Retina 2019 and Final Cut Pro 10.6.5 (the latest version to date). My Final Cut Pro's libraries are set to the "Wide Gamut HDR" colour space. I have a very good reason not to have them in SDR / Rec. 709, and I know how to colour manage the whole workflow from recording, over import and editing,, to output. I know what LUTs are, I know what colour grading is. I get pretty decent looking footage out of the Mavic 2 Pro's Hasselblad camera (although it has a tendency, like all digital camera sensors, to overexpose). So all of that is not a problem.

What is for me, a problem and actually a big mystery that I hope someone here is able to unravel, is the fact that all movie files shot in DLOG-M on my Mavic 2 Pro contain absolutely no metadata information regarding the colour space DLOG-M is using, and nowhere online can I find any information about what colour space DLOG-M is using, either... Therefore, Final Cut Pro, lacking that metadata information, treats all the DLOG-M footage as having the rather limited colour space of Rec. 709. Surprising, to say the least. Or am I missing something?

There are two ways in which one can check this, i.e. the lack of any metadate info regarding the colour space that DLOG-M uses.

• Open a DLOG-M movie file (either .MP4 or .MOV) in Apple's Quicktime Player app and open the Movie Inspector (Command + I). In the "Video Details" section there should be information about three essential data concerning the colour space the movie was recorded in: Colour Primaries, Transfer Function, and YCbCr Matrix. In most, if not all movie files that were recorded by other digital video cameras, these data are contained within the metadata of the file. But not in any DLOG-M movie file coming out of the Hasselblad camera of the Mavic 2 Pro!!! These fields are blank, empty, not filled in!
• Import a DLOG-M movie file in a Final Cut Pro Library set up as using the Wide Gamut HDR colour space (instead of the much more limited Rec. 709 colour space). It is automatically set as using the Rec. 709 colour space, according to the information in the Inspector pane. And... upon opening the metadata information panel in its "Extended" view, the "colour profile" metadata field of any DLOG-M clip is simply blank, empty, not filled in! Final Cut Pro has a feature that allows the user to "override" the colour space setting in the metadata of the movie clip, so I can fill in e.g. Rec. 2020 PQ or Rec. 2020 HLG in the empty colour space field of a DLOG-M clip, but the colours of the clip become totally oversaturated to the absurd, resulting in a totally unusable clip.
  

In other and more simple words: all DLOG-M movie clips recorded by the Mavic 2 Pro can only be edited inside the Rec. 709 or SDR colour space! Where are the 1 billion colours that 10-bit DLOG-M files are supposed to contain???

In stark contrast to this, shooting in the HLG colour profile on a Mavic 2 Pro delivers movie files that do contain the correct and necessary metadata about the colour space the clips were recorded in: Rec.2020 HLG! This, again, can be checked in Final Cut Pro and in Quicktime Player. As a consequence, whenever I edit HLG movie files recorded by the Mavic 2 Pro, I get far better results in colours, contrast, and overall image quality than when I use DLOG-M.

As a summary, my three most urgent questions:

• What colour space is DLOG-M recording its movie files in?
• Why does DLOG-M not write the correct information about its colour space to the metadate of the movie files?
• Why does DLOG-M appear to be using the much more limited colour space Rec. 709 whilst HLG is using the far wider and far more up-to-date Rec. 2020 HLG?
  

Maybe you will find answer here :

LV_Forestry Posted at 1-9 11:19
Maybe you will find answer here :
https://youtu.be/aPDOLwpEVVI

A video about what DLOG-M is and how you can use a LUT to make the clips look decent, does not give any answer to any of my questions about the COLOUR SPACE that DLOG-M is using...

In fact, the LUT being advertised in that video proves that, in order to be able to get any decent result from a DLOG-M movie file, you have to edit it in the Rec. 709 colour space! A colour space that is much smaller than the Rec. 2020 HLG one, which is being used by the HLG colour profile of the Mavic 2 Pro. Where are the 1 billion colours of the 10bit DLOG-M when you have to use Rec. 709?

Pegasos Posted at 1-9 11:27
A video about what DLOG-M is and how you can use a LUT to make the clips look decent, does not give any answer to any of my questions about the COLOUR SPACE that DLOG-M is using...

In fact, the LUT being advertised in that video proves that, in order to be able to get any decent result from a DLOG-M movie file, you have to edit it in the Rec. 709 colour space! A colour space that is much smaller than the Rec. 2020 HLG one, which is being used by the HLG colour profile of the Mavic 2 Pro. Where are the 1 billion colours of the 10bit DLOG-M when you have to use Rec. 709?

"Where are the 1 billion colours of the 10bit DLOG-M when you have to use Rec. 709?"
I have an idea

LV_Forestry Posted at 1-9 11:31
"Where are the 1 billion colours of the 10bit DLOG-M when you have to use Rec. 709?"
I have an idea

Do you or don't you?

Put a small DLOG file on Google Drive, I can check it with Resolve.

I do not work with Filmora or other medium programs, only Adobe and DaVinci....Your HLG are better, as HLG will have the colors (you even need to reduce the saturation) and use a non linear data (DR is pushed on Highlights but remains the same on Shadows). You need to align histogram on right instead of left, in order not to overexpose.
DLOG have a linear push from shadows to highlights that gives a general DR. You need to align histo on left in order not to underexpose. Will be used Curves and HDR controls in order to regain contrast and not directly the Expo/Contrast, as you will make a mess (you will push linear settings with linear modifications creating "stair like" gaps instead of curved continuous editing). Is easy to fix with Color Space Transform where you can define the input and output and a LUT, but this is in DaVinci, no idea how is in Filmora or rest...

Also some editors need to reinterpret manually the footage AND also the RGB range (full or limited), for example in Resolve you put it as input the Wide Gamut and ST2020, but also in editor you need to define the color space manually, or will be misinterpreted. Your output should be REC2020 HLG only if you need an HDR clip...but for this, you need to edit in HDR...
One more thing...a lot of TV sets know to interpret HLG and translate it naturally in "HDR like" view, but none can interpret DLOG....so you need to use HDR container when you save to R2020 HDR mode, as the TV will search for container that say all...

Thank you for all the technical info, most of which I already know. However, my questions are still not really answered. But... you are telling me something about the quality of HLG that I have noticed myself, independently! Until now, I seem to have far better results in overall quality (colours, contrasts, dynamic range, sharpness...) when I shoot in the HLG profile on my Mavic 2 Pro, than when I use DLOG-M.

I'm quite busy today, but I will try to test something out in Final Cut Pro. Putting a DLOG-M file on my Google Drive may not be necessary if my test succeeds. Also, if you want to check yourself about the lack of colour space metadata in the DLOG-M files produced by the Mavic 2 Pro, I have already mentioned that Quicktime Player is the reference to do that on the Apple platform. If you work with Windows, perhaps the VLC Media Player allows you to see the metadata information of a .MOV file.

Also, check out the technical document DJI published in 2017, about DLOG, and which I will attach to this message. Apart from very technical details, it has a diagram with the colour space of DLOG (of which DLOG-M seems to be a derivative, as far as we know or can assume). On that diagram you can see that the colour space of DLOG is 1) far wider than DCI-P3 (which is a colour space available to my iMac 5K Retina display) and 2) exceeds that of Rec.709 even more. The difference between the DLOG-gamut and Rec.709 is huge! Which leads me to the conclusion that, when recording in DLOG-M and using (or actually being forced to use?!) the Rec.709 colour space in a video editor, an enormous amount of colours or grades of colours are lost, being thrown out, and shifted, to fit into the much narrower colour space of Rec.709. The final result, although it may be sufficiently visually pleasing, is technically speaking, a waste of all the advantages of 10-bit recording that DLOG-M offers...

I hope to know more after I have done some more testing in Final Cut Pro later today. In any case, I intend to contact DJI support to ask them about the reason why Mavic 2 Pro DLOG-M files simply lack the necessary colour space metadata, whilst the Mavic 2 Pro HLG files have them.  

As I said in my previous comment, yesterday I was able to do some more testing. And I was able to find a partial answer and partial solution to the question and the problem. I will try to explain this as clearly and simply as possible, but to understand what is going on, you need some knowledge about what colour profiles, colour spaces (or colour gamuts), and metadata are and actually do in the workflow from recording over editing to publishing movie files coming out of any digital movie camera out there.

Apart from the "normal" mode to record the movie files, the Hasselblad camera of the Mavic 2 Pro features two "professional" colour profiles: DLOG-M (most probably a derivative from the somewhat better documented DLOG profile, see the "Whitepaper" pdf file in my previous post) and HLG. DLOG-M seems to be a proprietary profile of DJI or Hasselblad or both (no other company seems to be using it). HLG is a much better documented and known colour profile. The HLG colour profile records the movie files using the very large colour space "Rec.2020 HLG", a much larger colour space that has a number of advantages and possibilities that the much older and until now more common Rec.709 colour space does not have. I will not go into this here.

However, it is important to know that recording movies with the Mavic 2 Pro using either the DLOG-M or the HLG colour profile, gives the user the advantage of - theoretically - having 1 billion (!) colours at his disposal to work with in editing, instead of just +/- 1 million colours when no colour profile is chosen in the camera settings of the drone. This is due to the fact that only in the DLOG-M and HLG colour profiles the 10bit recording mode is being used. In short and much more simply said: DLOG-M or HLG movie files contain much more information about colours, grades of colours, contrast, dynamic range etc. If you want your movies to have the absolute best quality and, for example, take advantage of the upcoming HDR TV screens etc, you shoot in DLOG-M or HLG.

But... (and this was what puzzled me until yesterday) if you open a DLOG-M movie file in a video editor (I use Final Cut Pro) of which the Library or Working Space is set up as "Wide Gamut HDR" (instead of the much more limited but more common Rec.709 working space), the file is automatically interpreted and edited as having been recorded in the much more limited Rec.709 colour space!!! In other words: The original 10bit colours will be "shifted", as they have to be fitted into a much smaller colour space. And shifted means changed, and no longer true to how they were recorded by the Hasselblad camera of the Mavic 2 Pro. Basically, millions of colours are being lost. Which means that most of the advantages of shooting in DLOG-M can simply not be used in editing, at all!!!

In comparison, HLG movie files coming out of the Mavic 2 Pro are correctly interpreted by Final Cut Pro (and other software!) as using the Rec.2020 HLG colour space, and thus no colours are shifted, changed, or thrown out!

This puzzled me enormously, until I found out what is the cause of this difference in how DLOG-M files are "interpreted" or "read" by the video editing software versus how HLG files are. The cause of this problem is that the software of the Mavic 2 Pro or of its Hasselblad camera, when it is using the DLOG-M colour profile, does not embed the necessary "colour space" (or gamut) metadata into the movie files!, whilst it does embed the necessary colour space metadata when it is using the HLG profile!!! In other words, the video editing software, upon trying to interpret in what colour space the movie file has been recorded in, simply cannot find the necessary information (metadata) in the movie file and automatically and "safely" interprets it as Rec.709. In the case of the HLG files, there is no such problem as the correct metadata about the colour space (such as what are the colour primaries etc.) has been embedded by the software of the drone or the camera.

How did I find this out? When I opened the metadata panel of a DLOG-M movie file in Final Cut Pro in the "Extended" view, I found the "colour profile" metadata field to be empty! That metadata field tells the video editing software, amongst other things, what colour space the movie file has been recorded in. Since there is no such information, Final Cut Pro could not identify the colour space and by default automatically interprets it as being Rec.709, even inside the "Wide Gamut HDR" working space or library. When I opened a HLG movie file, however, the "colour profile" metadata field was correctly showing "HLG" and thus Rec.2020 HLG could be used as the correct colour space to edit the movie file in.

The "solution" to this problem, however, is relatively simple, but it needs user input for every single DLOG-M movie file that is being imported into Final Cut Pro. Final Cut Pro allows you to manually "override" or (in this case) enter the correct metadata in the "colour space override" metadata field (attention: not the "colour profile" metadata field!) in the "Settings" view of the metadata (Info Inspector) panel! It is, however, very important to note three things before doing this:

1. The Library of Final Cut Pro needs to be set to "Wide Gamut HDR".

2. In the "Colour Space Override" metadata field you need to enter "Rec. 2020" (not "Rec. 2020 PQ", nor "Rec. 2020 HLG", and certainly not "Rec. 709"). Apparently, the DJI DLOG-M colour profile is a bit more limited than the latest and greatest colour profiles. But that is not an issue at all because its "colour space" (or gamut) surpasses the P3 and Rec. 709 ones (see the diagram in the White Paper PDF document in my previous post). Yet, this advantage versus P3 and Rec. 709 is annihilated unless we manually enter the correct metadata information ourselves in the "Colour Space Override" metadata field!!! This should not happen, and I have been in contact with DJI Support an hour ago. They are transferring the issue to the Developers' Team and will report back to me. The question was and is: Why does the software inside the drone not embed the Colour Profile metadata into the DLOG-M movie files? Could it be because of the fact that DLOG-M is a proprietary profile format? If that would be the reason or cause, this fact should be documented so that we, the users, know that we have to manually enter or override the colour space metadata in our video editing applications!

3. Enter the "Rec. 2020" metadata in the "Colour Space Override" field before doing any editing and do not apply any "Camera LUT" because the chances are high that that LUT will "override" and thus annihilate the correction we just made by entering the correct "colour space" information into the metadata of the movie file.

This "solution" is not a real solution, as the work has to be done for every single DLOG-M file, cannot be automated (at least not in Final Cut Pro), and s not documented by DJI or anyone else to my knowledge. But it is a solution helping us to benefit from all the advantages of DLOG-M inside our video editors.

Final note: I am using Final Cut Pro, and one could reason that this problem does not arise when using other video editing software (DaVinci, Premiere...). I cannot test it with other video editing applications, but logic tells me that the very same problem arises in every single video editing application out there, simply because the problem lies in the fact that the necessary "colour space" metadata is not being embedded by the Mavic 2 Pro into its DLOG-M movie files. So we can expect that every video editing application out there, will search for that metadata information upon import of a DLOG-M movie file, will not find it because it is missing, and will therefore "safely" interpret (or "tag" it if you want) as having been recorded in the "Rec. 709" colour space, whilst in reality it has been recorded in a much larger one! The result being that an enormous amount of colour and colour grade information is simply not being used, or actually thrown out, when editing any DLOG-M movie file without manually entering the correct metadata first!

I have just received from DJI Support an answer to my query about why the metadata information of the colour space used by DLOG-M is not embedded into the movie files that are being recorded in the DLOG-M colour profile, whilst it is embedded when using HLG as the colour profile set up in the software of the Mavic 2 Pro. I was happy with how fast I got an answer (from the developers), but I am unhappy with the fact that there is no solution to the problem of the missing colour space metadata inside the DLOG-M movie files. Only manually entering that information for each and every DLOG-M movie file (inside a video editing application) can be seen as a (more or less) useful procedure  to avoid as much as possible (but never totally) the shifting of millions of colours (FYI: the human eye can distinguish 16 million different colours!) from the very wide DLOG-M colour space into the much more limited Rec. 709 colour space. In other and more simple words: Using DLOG-M for the Mavic 2 Pro means using a proprietary colour profile that records the movie files using a colour space nobody outside DJI knows the exact specifications of. In even more simple words: Using DLOG-M in a video editor whilst trying to keep all the useful colour information during the import, edit, and output workflow, is impossible, and converting it to Rec. 709 (which is done by default!) means shifting millions of colours without having any control over it.

Luckily, I have discovered this (although too late to apply this knowledge to my previous movie projects) and will only be using the HLG colour profile from now on!

Here is the reply from DJI:
The HLG of Mavic 2 Pro is a standard HLG, 2020 color space
Dlog-M does not use a standard color space, which can be understood as our own designed, but the color space information of DLOG-M involves technical sensitivity and cannot be provided.
Normally you can convert to Rec 709 by using DLog-M to Rec709 LUT

And here is my reply to theirs:

Thank you!
I do understand that the colour space info of DLOG-M is being kept ‘secret’, but then I do not understand the use of it because if it needs to be converted to Rec. 709, literally millions of colours are being thrown out of the movie files. The only real advantage remaining of DLOG-M inside a Rec. 709 colour and working space, as a consequence, is that one can edit and ‘adjust’ the remaining colours and contrasts "a little bit more" than with a movie file recorded with “None” selected as the colour profile in the DJI software of the drone, as the former’s colour space is reduced to Rec. 709 from a much wider one, and the latter is by default Rec. 709.

Logically, DLOG-M is then not to be recommended to be used as a colour profile to use in the DJI software of the drone.

Pegasos Posted at 1-12 05:39
I have just received from DJI Support an answer to my query about why the metadata information of the colour space used by DLOG-M is not embedded into the movie files that are being recorded in the DLOG-M colour profile, whilst it is embedded when using HLG as the colour profile set up in the software of the Mavic 2 Pro. I was happy with how fast I got an answer (from the developers), but I am unhappy with the fact that there is no solution to the problem of the missing colour space metadata inside the DLOG-M movie files. Only manually entering that information for each and every DLOG-M movie file (inside a video editing application) can be seen as a (more or less) useful procedure  to avoid as much as possible (but never totally) the shifting of millions of colours (FYI: the human eye can distinguish 16 million different colours!) from the very wide DLOG-M colour space into the much more limited Rec. 709 colour space. In other and more simple words: Using DLOG-M for the Mavic 2 Pro means using a proprietary colour profile that records the movie files using a colour space nobody outside DJI knows the exact specifications of. In even more simple words: Using DLOG-M in a video editor whilst trying to keep all the useful colour information during the import, edit, and output workflow, is impossible, and converting it to Rec. 709 (which is done by default!) means shifting millions of colours without having any control over it.

Luckily, I have discovered this (although too late to apply this knowledge to my previous movie projects) and will only be using the HLG colour profile from now on!

Pegasus,

Can't thank you enough for your detailed explanation. Wow, crazy workaround but the immediate usability of the clips is tremendous. Thank you!

Patrick

fans7807bce2 Posted at 4-12 11:05
Pegasus,

Can't thank you enough for your detailed explanation. Wow, crazy workaround but the immediate usability of the clips is tremendous. Thank you!

Dear Patrick,

You're welcome.
But are you talking about the HLG colour profile when you praise its "immediate usability", or do you add the metadata for DLOG-M yourself to the footage inside your editing software?
I've been testing both the DLOG-M and HLG profiles next to each other using almost similar footage from the camera. Both profiles record in 10-bit, as is stated by DJI. But what baffled me from the start was the fact that everybody who is using the DLOG-M profile for recording, ads one or more Rec.709 LUTs to their footage before or during editing. Although the results look OK, the basic fact remains that this procedure basically throws away millions of colours because the Rec.709 colour space is so much smaller than the one DLOG-M is using. One can find diagrams comparing the colour spaces of Rec.709, DLOG-M, and HLG on several websites.
Recording in DLOG-M and then converting that footage (or rather "squeezing" it into the much smaller Rec.709 colour space) is like shooting photographs in RAW and then converting them into JPEGs. The same thing happens: you throw away millions and millions of colours.
If one is satisfied with the results coming out of that procedure (DLOG-M colour space > Rec.709 colour space), then fine with me. But using the HLG colour profile with its much larger colour space (billions of colours!) whilst recording and not converting the footage to Rec.709 during editing, does not throw away any colours. Throwing away colours means shifting colours from their original hue to one that is available in Rec.709, and that in a totally uncontrollable way.
I find it very sad that DJI never bothered to have the Mavic 2 Pro record the metadata into DLOG-M footage (perhaps it's not even possible or would cause editing problems because DLOG-M is a proprietary format), and even more sad that they never explained what happens with those millions of hues when editing DLOG-M inside a Rec.709 colour space.

Pegasos Posted at 4-13 01:02
Dear Patrick,

You're welcome.

Hello Pegasos,
I am not sure when did you write it, but it is an amazing explanation and find-out.
I was always puzzled why this D-log did not perform well, basically being more similar to a Flat profile on a camera just allowing  for a bit more control over bright and dark, but otherwise rather poor. I often wondered why the "normal" color space can be so vivid, and on D-log i sometimes push sliders to the max, and all i get is a distorted, but poorly coloured image.
And I tried your explanation - suggestions - and immediately - bingo! Override space to rec.2020 and magically the palette of colours gives so much more detail and life.
So, i basically write to say THANK YOU for your detailed post. It would be great to create a simple script for automated function, to override it with 1 click, like you can perform actions in photoshop. not sure if it is possible in video editors. but even manually - it is so worth it.
Thank you again!

Holoboss Posted at 4-29 10:55
Hello Pegasos,
I am not sure when did you write it, but it is an amazing explanation and find-out.
I was always puzzled why this D-log did not perform well, basically being more similar to a Flat profile on a camera just allowing  for a bit more control over bright and dark, but otherwise rather poor. I often wondered why the "normal" color space can be so vivid, and on D-log i sometimes push sliders to the max, and all i get is a distorted, but poorly coloured image.

You're welcome!

Pegasos Posted at 1-12 05:39
I have just received from DJI Support an answer to my query about why the metadata information of the colour space used by DLOG-M is not embedded into the movie files that are being recorded in the DLOG-M colour profile, whilst it is embedded when using HLG as the colour profile set up in the software of the Mavic 2 Pro. I was happy with how fast I got an answer (from the developers), but I am unhappy with the fact that there is no solution to the problem of the missing colour space metadata inside the DLOG-M movie files. Only manually entering that information for each and every DLOG-M movie file (inside a video editing application) can be seen as a (more or less) useful procedure  to avoid as much as possible (but never totally) the shifting of millions of colours (FYI: the human eye can distinguish 16 million different colours!) from the very wide DLOG-M colour space into the much more limited Rec. 709 colour space. In other and more simple words: Using DLOG-M for the Mavic 2 Pro means using a proprietary colour profile that records the movie files using a colour space nobody outside DJI knows the exact specifications of. In even more simple words: Using DLOG-M in a video editor whilst trying to keep all the useful colour information during the import, edit, and output workflow, is impossible, and converting it to Rec. 709 (which is done by default!) means shifting millions of colours without having any control over it.

Luckily, I have discovered this (although too late to apply this knowledge to my previous movie projects) and will only be using the HLG colour profile from now on!

Hello Pegasos

Thank you so much for all those information! You're certainly not alone! I'm working with an DJI AIR 2S and having the same "problems" as you described. It seems that most (YouTube) creators do not have this problem or accept it or do not have better knowledge. Anyways, this thread is a big step for my research!

I'm using Premiere and DaVinci - when I check the d-log-m files in Davinci (Davinci wide gamut / color managed) it automatically sets the Rechnung.709 (scene) as Input.
As it seems the rec.2020(scene) displays the color more correctly, more defined, balanced etc. - as you said, still a gambling. is it rec.2020? or not...

I'm currently tending to shoot in HLG and therefore the input needs to be rec.2020 as well accordingly to DJIs answer.
But (hello pandoras box) It's a lot and complex to read but very informative.
check this article here and here (links)
https://xtremestuff.net/recording-editing-with-hybrid-log-gamma-part-1/
https://xtremestuff.net/recording-editing-with-hybrid-log-gamma-part-2/

It issues that Davinci is interpreting the HLG footage wrong as well.
the right input - according to the article - is

Gamut: daVinci Wide Gamut (or wide gamut in FCX)
IDT: Rec.2100 HLG (scene) = Rec.2020 HLG ARIB STD-B67

Now I'm a little confused because if I set the input to rec.2020 as the DJI response indicated I get a different image.
So what should we do exactly?

Further more: how to combine HLG footage with non-HLG footage?
Example: editing a short reel with drone footage (HLG) and SDR kind of log footage form a video recorder that is not able to record in HLG.

Really like to work with HLG but with more confident doing the right thing.
Working with d-log-m is - as you said - not really the proper way... it works but is like rotten from the inside.
Funny that there is so much misinformation about HLG... on YouTube and the web.

Thank you again & best regards
Leo

Pilot_Leo Posted at 5-25 13:26
Hello Pegasos

Thank you so much for all those information! You're certainly not alone! I'm working with an DJI AIR 2S and having the same "problems" as you described. It seems that most (YouTube) creators do not have this problem or accept it or do not have better knowledge. Anyways, this thread is a big step for my research!

Hello Pilot Leo,

I'm afraid I don't have any experience with DaVinci nor with any other drone but my Mavic 2 Pro. I can't answer all your questions but I will explain about the colour spaces that I mentioned in my previous messages on this thread.

Again:

1. Recording in HLG is done in 10 bit.
2. Recording in D-LOG-M (which is more limited than the newer D-LOG of the newer Mavic 3 drones) is done in 10 bit.
3. Recording without any colour profile ("Normal") is not done in 10 bit

10-bit recording allows for many more colours to be captured by the drone's camera.

But...

1. Recording in HLG uses a very wide colour space (and allows for HDR editing) that allows for editing in Rec. 2020 without any loss of colours (there is no compression done from the wide colour space into a much smaller one - the Rec.709 one).
2. Recording in D-LOG-M uses a colour space that is wider than Rec.709 but a LOT more narrow than HLG / Rec.2020!!!
3. Recording without any colour profile selected, means you need to edit in Rec.709

Now here is the "strange" thing, which made me search for explanations about D-LOG-M in the first place: everyone who records in D-LOG-M seems to find it "normal" to throw away millions of colours when converting/importing and editing D-LOG-M files in the much more limited Rec.709 colour space!

When recording in HLG, it is totally possible to edit the files in Rec.2020, thus preserving the FULL amount of colours recorded.

Furthermore, D-LOG-M files don't contain any metadata telling the editing software in what colour space they were recorded! That is logical (but very annoying) because it's a proprietary file format, designed by DJI. It is possible in Final Cut Pro to edit the metadata of D-LOG-M files and enter for example Rec. 2020 as the colour space. Then the colours won't change. Why not? Because it's like putting a small shoe box into a much larger one, if you equal colour space to shoe boxes (small ones, medium ones, large ones). No colours are shifted when you put a small box in a larger one.

If you scroll up on this thread, you will find a comment of mine with a link to a PDF document, published by DJI, about the D-LOG gamut (or colour space). In it you will find a diagram showing the difference in gamut (or space) between D-LOG and Rec.709. It is a very substantial difference. Although D-LOG-M is different from D-LOG in the sense that D-LOG-M has a smaller gamut than D-LOG, the diagram helps to understand how converting D-LOG-M files into the much narrower Rec.709 gamut, will "throw away" millions of colours or shades of colours by shifting them to colours that are present in the narrower gamut of Rec.709.

In a way, recording in D-LOG-M and then editing the files in Rec.709, makes little sense unless you are only going to use/view the final video files on standard monitors or older TV screens that can only render Rec.709 because that is their maximum gamut.

In short: if you want to keep things simple and only record, edit, and view files in Rec.709 environments, recording in D-LOG-M is "fine", but in a way it's overkill.
I still keep on recording in D-LOG-M but more and more I'm switching to HLG. Mainly because I'm stuck with my Mavic 2 Pro. The Mavic 3 camera is much more powerful, but beyond my budget for now...

To answer your question about combining HLG and non-HLG footage: Try entering the Rec.709 footage into a HLG workspace in your video editor. I'm not sure that it will work in DaVinci.  I use Final Cut pro, as I said above.

Pegasos Posted at 5-25 13:56
Hello Pilot Leo,

I'm afraid I don't have any experience with DaVinci nor with any other drone but my Mavic 2 Pro. I can't answer all your questions but I will explain about the colour spaces that I mentioned in my previous messages on this thread.

I am so thankful for this post, and all the effort you put into providing clarity, giving a FULL explanation of the issue. I have a new Mavic 3 Pro Cine, which I bought partly because of the 3 cameras with different focal lengths and the 10 bit color.
One of the limitations of the Mavic 3, even the Pro Cine version - is the 3rd camera (70mm) will only shoot in D-LOG-M, whereas the 24mm and 30mm cameras will also shoot in DLOG. So if using all 3 cameras, one needs to pick a shooting profile consistent to all 3 (either HLG or D-LOG-M - after the recent firmware update in September 2023). I came to the same conclusion you did after watching countless videos on color grading D_LOG-M footage - "... just use the DJI LUT ...", but I was less than impressed with the results I was getting.
This week I reviewed some very informative videos on how to use Color Space Transform in Davinci Resolve, where Daria Fissoun explained the system and workflow. I got excited at the thought of an automated workflow for my new crispy footage! Sadly, this automated workflow is critically dependent on knowing the exact color space of the camera and setting the project settings to match it. I went on a search for the correct input color space for the expensive Hasselblad camera I just bought, and eventually wound up here.
Unless DJI decides to reveal their precious D-LOG-M color space magic, we are left with manually entering the input color space for each clip. I will probably use one of 3 workarounds moving forward:
1 - Record in D-LOG using just the 24 and 30mm cameras. I will have to decide before filming whether or not I will need the 70mm camera.
2 - Record in HLG and use the Rec2020 - HLG input color space (automated in DR).
3 - Record in D-LOG-M as I am now, and just manually input the color space for each clip. I think if I add the information prior to editing/cutting the clip, it will save me some work in color grading later.

Again, thank you for posting this information.

Guitars555 Posted at 9-11 04:33
I am so thankful for this post, and all the effort you put into providing clarity, giving a FULL explanation of the issue. I have a new Mavic 3 Pro Cine, which I bought partly because of the 3 cameras with different focal lengths and the 10 bit color.
One of the limitations of the Mavic 3, even the Pro Cine version - is the 3rd camera (70mm) will only shoot in D-LOG-M, whereas the 24mm and 30mm cameras will also shoot in DLOG. So if using all 3 cameras, one needs to pick a shooting profile consistent to all 3 (either HLG or D-LOG-M - after the recent firmware update in September 2023). I came to the same conclusion you did after watching countless videos on color grading D_LOG-M footage - "... just use the DJI LUT ...", but I was less than impressed with the results I was getting.
This week I reviewed some very informative videos on how to use Color Space Transform in Davinci Resolve, where Daria Fissoun explained the system and workflow. I got excited at the thought of an automated workflow for my new crispy footage! Sadly, this automated workflow is critically dependent on knowing the exact color space of the camera and setting the project settings to match it. I went on a search for the correct input color space for the expensive Hasselblad camera I just bought, and eventually wound up here.

You're welcome.
D-LOG-M and D-LOG will never be standards. They are proprietary format created by a drone factory, not a camera factory... HLG, however, is a standard, and since quite some time. Why would you not simply shoot everything in HLG, which has a far wider colour gamut/space than D-LOG (D-LOG-M has an even smaller one). I find HLG far easier to edit and my iMac 5K 27inch from 2019, which isn't even the latest and greatest computer out there, has no problems with it using Final Cut Pro.
If you shoot HLG, you have a very wide colour gamut/space, and you can easily convert the video clips from rec.2020 to rec.709 if ever needed. Rec.709 will become obsolete one day, perhaps sooner than we think.
I would know which "workaround" to choose from the three you propose if I had the Mavic 3 Pro or Cine. My Mavic 2 Pro, however, is still my trusted friend up in the air, and its Hasselblad camera gives me wonderful footage, in HLG...

@Pegasos I want to add my thanks for your research years ago.  I'm actually going to request this again of DJI and link this thread.  

I recently picked up an OSMO Action 4 camera that shoots DLOG-M.  In my experimentation in Davinci Resolve's Color Transforms I found that the best I can do to get "close" to what the camera sees is to use:

Input Color Space: DJI D-Gamut
Input Gamma: Rec. 2100 HLG

Now I find this thread and I understand why that's likely the best settings.  Interestingly, the following Input Gamma settings appear to be nearly identical:  Rec. 2100 HLG, ARIB STD-B67 HLG, and Rec. 2100 HLG (Scene).

That could have been 1:1 from me.

A DJI proprietary, undisclosed log format has no future and doesn't make any sense. It's a contradiction in itself. What does the manufacturer want to protect here, anyway? That people can't handle it properly? Laugh.

The HLG standard is openly accessible, has a very wide color gamut, can store a dynamic range of 17.6 stops, the metadata is correct, and so on.

It doesn't make sense to continue working with Dlog, which originated from the limitations of 8-bit. The world keeps turning, and we now have new, modern formats that no longer need it. It was a good idea back then, but it's no longer necessary.

If these people would only turn on their brains once...
I've often received the same answers as you :-) And I also said again and again, the question was unfortunately not answered. But it doesn't matter what others don't do. Everyone should be happy with what they like. You can walk through the world with open or closed eyes. It would be nice if people talked less and gained their own experiences.
Dlog-M no longer has a future.

I find you very likable

Not sure how old this thread is, but for anyone that may come upon it in the future....

This myth that DLOG is some sort of wonderful profile that "serious pro's and colorists" use must end. It has to stop!

Sometime back in the old Phantom days I recall started to experiment with HLG vs DCinelike I think. I remember coming to the conclusion, rather easily, that HLG was a superior profile.  Now that I am on the Pocket 3 Osmo, which has HLG and DLOG-M, I am again reminded that DLOG is trash. I color all my footage, from all my cameras (canon, RED, DJI, etc) in ACES. There is no supported DLOG IDT, and I doubt there will ever be one. This "secret proprietary" formula that DJI insists on is ridiculous. HLG is standard, and can pretty much translate from any camera. It's 10-bit, records billions of colors with a high dynamic range. It is simple to convert inside an ACES space.  In just a few clicks, I can match an HLG shot to the identical "normal" shot, with the added bonus of much greater flexibility to recover highlights, grade, etc.

DJI's insistence that we must use their "LUT" to accurately convert DLOG-M footage is absurd. Professional color pipelines do not use LUT's.

Stick to HLG. It is far superior.

mrryan76 Posted at 11-7 11:59
Not sure how old this thread is, but for anyone that may come upon it in the future....

This myth that DLOG is some sort of wonderful profile that "serious pro's and colorists" use must end. It has to stop!

Indeed!

Hi, really thanks to all for this thread!

I've been trying to fight D-Log M in Davinci for quite a while, without success. Neither the various combinations in CST nor the YRGB color managed approach led to a successful end, not to mention 'crippling' it with DJI LUT. I really thought that from the sorting logic in the app settings, D-log M would provide the highest quality (DJI M4P), but it's really nonsense.

I am 'spoiled' by the adroid app, where I can set both color space and gamma curve separately in 10bit LOG, in this case, the subsequent post-process is easier.
So for DJI, at least from my point of view, HLG is the only functional solution, at least as far as post-processing is concerned.

Test shots on Pocket 3 in all 3 color profiles, "normal", "HLG", and "D-Log". When brought into Davinci, they all show as "10bit" depth in the Inspector. I know this can't be possible, so maybe it is because of the H.265 container? DJI specifically list's their DLOG-M as 10-bit (literally named that way in the camera profile), but now I am wondering if the HLG is 10-bit or 8-bit??

I have been unable to find any technical info about this, so if anyone can link or point to a resource or whitepaper that can confirm the true bit-depth of HLG, that would be awesome!

Thanks

mrryan76 Posted at 11-9 12:21
Test shots on Pocket 3 in all 3 color profiles, "normal", "HLG", and "D-Log". When brought into Davinci, they all show as "10bit" depth in the Inspector. I know this can't be possible, so maybe it is because of the H.265 container? DJI specifically list's their DLOG-M as 10-bit (literally named that way in the camera profile), but now I am wondering if the HLG is 10-bit or 8-bit??

I have been unable to find any technical info about this, so if anyone can link or point to a resource or whitepaper that can confirm the true bit-depth of HLG, that would be awesome!

The DJI Mavic 2 Pro Hasselblad camera, which I'm using, records HLG in 10bit. Because of the large colour space (aka gamut) of HLG, 10bit recording is a necessity, in any camera IMHO. HLG is a colour space or gamut. 8 or 10bit recording has to do with how much information is stored inside the video clips/images/photos.

LV_Forestry Posted at 1-9 11:19
Maybe you will find answer here :
https://youtu.be/aPDOLwpEVVI

peace of ... this video

mrryan76 Posted at 11-9 12:21
Test shots on Pocket 3 in all 3 color profiles, "normal", "HLG", and "D-Log". When brought into Davinci, they all show as "10bit" depth in the Inspector. I know this can't be possible, so maybe it is because of the H.265 container? DJI specifically list's their DLOG-M as 10-bit (literally named that way in the camera profile), but now I am wondering if the HLG is 10-bit or 8-bit??

I have been unable to find any technical info about this, so if anyone can link or point to a resource or whitepaper that can confirm the true bit-depth of HLG, that would be awesome!

Short answer : everything is possible !

Longer one : everything can be recorded in 8-10-12-14 bit... But depending on how the signal was processed and how it is aimed to be used, it is not necessary to record in a higher bit depth.

An image recorded in log should have the higher bit depth possible to get the most precise shades possible in the image, enabling better grading in post. A log format is generally a storage format, aimed to be processed later, not to be dispalyed as is.
I said generally because HDR diffusion formats (PQ and HLG) are log format, but aimed to be displayed.

Here is the white paper for HLG and PQ.

Both should be recorded as least in 10 bit, and share a common color space : Rec2020

I just discovered this thread so here are my 2 cents, from personal experience.

DLogM on the Mavic 2 Pro is different from the version present on the Mini 4 Pro, Action 4, Osmo 3 (I have all of them). I don't when they transitioned from one to the other.
DLogM on th Mavic 2 Pro was a modifed HLG curve, keeping the log part and adding a knee/linear part to the bottom (very common on log curves in general), instead of the gamma curve section of the HLG.
The color space was not something common as far as I know.

The new DLogM curve is very different. Still a log curve with a linear part at the bottom but no knee. The colorspace used is Rec709 as far as I've tested.

Now, here is what I find utterly annoying : HLG was correctly implemented on the Mavic 2 Pro, but on the Mini 4 Pro and Osmo Pocket 3, the curve is very strange (and in a bad way) ! The Action 4 doesn't support HLG, only DLogM (WHY ?!).

Why is it annoying you ask. Well, HLG is a standard with a white paper (so, know curve and color space), which make the transformation to whatever format you like working with relatively easy (when you know how this things works I mean). I, for one, prefer HLG over the new DLogM because the colorspace is wider (Rec2020 instead of Rec709), which avoid color related in some pretty common cases (like deep blue sky).

HLG curve first and DLogM curve in second, both from the Mini 4 Pro. The red line is the middle gray.
As you can see where the white arrow points, there is a slight bump in the HLG curve (actually, it raises the signal and then lowers it for the remaining high part of the curve).
I tried a lot of different ways to correct it and nothing gives good results. Everytime, I either get artifacts or colors go nuclear.

The same thing can be seen on the Osmo Pocket 3 HLG.


That is very frustating as I can't understand or find a good reason as to why they did it !
So I could just use DLogM, you might say. Well...

This is a picture from a blue sky, shot with a Blackmagic Pocket 4K. Note the slight skew on the vectorscope in the deepest blues.

Here is the same thing shot with the Osmo Pocket 3 in HLG (without correcting the bump, so the colors are pretty good.
And now in DLogM, still on the Osmo Pocket 3. Note that the skew is bigger and on the other side now. If you look closely, you can see that the sky is less saturated, while having a slight purple hue compared to the other two.
I encountered this phenomenon several time with the Pocket 3, which annoys me quite a lot...

Every shot above was graded in ACES, with a Rec709 ODT. As I use it for all my work, I need IDTs to be able to transform the original images to ACES colorspace.


All of this to say : please DJI, fixe the HLG curve and/or use a wider colorspace (Rec2020 for instance) in your DLogM fomat !

Pegasos Posted at 1-12 05:39
I have just received from DJI Support an answer to my query about why the metadata information of the colour space used by DLOG-M is not embedded into the movie files that are being recorded in the DLOG-M colour profile, whilst it is embedded when using HLG as the colour profile set up in the software of the Mavic 2 Pro. I was happy with how fast I got an answer (from the developers), but I am unhappy with the fact that there is no solution to the problem of the missing colour space metadata inside the DLOG-M movie files. Only manually entering that information for each and every DLOG-M movie file (inside a video editing application) can be seen as a (more or less) useful procedure  to avoid as much as possible (but never totally) the shifting of millions of colours (FYI: the human eye can distinguish 16 million different colours!) from the very wide DLOG-M colour space into the much more limited Rec. 709 colour space. In other and more simple words: Using DLOG-M for the Mavic 2 Pro means using a proprietary colour profile that records the movie files using a colour space nobody outside DJI knows the exact specifications of. In even more simple words: Using DLOG-M in a video editor whilst trying to keep all the useful colour information during the import, edit, and output workflow, is impossible, and converting it to Rec. 709 (which is done by default!) means shifting millions of colours without having any control over it.

Luckily, I have discovered this (although too late to apply this knowledge to my previous movie projects) and will only be using the HLG colour profile from now on!

.   So Quicktime Inspector says the Dlog file is 10-bit, but says the type is ITU-R BT.709 ?? Very wierd

On the other hand Quicktime inspector says HLG files are 10-bit and HDR type is HLG with color of ITU-R BT.2020

I totally don't get what the inspector says it sees for DlogM

I guess this is inline with proprietary format, we're not going to tell you anything.
What's even weirder is the resolutions are different!   What!?

Gadget Freak Posted at 12-20 20:42
[view_image]

[view_image]

First of all, a remark and a question.
1. You're using an Osmo 3 device. All my comments about D-LOG M and HLG were made because I'm using a Mavic 2 Pro (drone). I see a difference between the metadata of your D-LOG M files and mine. There's no metadata concerning the colour space (Rec.709 or Rec.2020 or any other) in my D-LOG M file coming out of my Mavic 2 Pro! Apparently, the Osmo 3 does record the metadata for D-LOG M files.
2. What is it exactly that you don't understand about the metadata (as shown by the Inspector in Quicktime Player) of your D-LOG M files? In fact, if you are showing us in those two images the metadata of the original files, straight out of the camera of your Osmo 3, then there is indeed something strange going on, but something still "in line" (as you say) with the proprietary format of D-LOG M and with what I have said in my previous comments about how a major part of the (more or less) 1 billion colours are getting lost when using Rec.709 in editing (and/or rendering for final output).

For clarity's sake, a summary of what happens with colours during the workflow of: 1. recording/capturing the images on video in a device; 2. importing and editing the images in a video editor (NLVE); 3. rendering and "exporting" them in order to view them as a movie on another device (computer screen, TV monitor, smartphone, tablet, etc.).
It is essential to understand that during each of these three stages (recording, editing, rendering), two elements play a major role, two elements that define how we perceive the colours (or shades of colours, in mixes of RedGreenBlue RGB).
First, in each stage a different device is being used (except when the final video is viewed on the same computer display as the one it has been edited on), and each device has its own peculiar way of "seeing" or interpreting colours and showing them to us. In my case, the Hasselblad camera on my Mavic 2 Pro records the images. I edit them using an iMac 5K Apple display that is set to use its default P3 colour profile (which is not HDR but shows far more colours than a "simple" sRGB display). And I view the edited videos on the same display and on several other displays as well (iPhone, MacBook Pro, Apple Thunderbolt Display...). In order to have the colours in a video displayed as accurately as possible on different displays and devices, it is necessary to use "standard" colour spaces (also called colour gamuts). This is the second element in play.
So, this second element, the "colour space" used in a video to render the colours as accurately as possible, is what is important to keep track of during the three stages (recording, editing, rendering).
What is a "colour space", e.g. sRGB, Adobe RGB (for photography) and Rec.709, Rec 2020, etc. for video? It is a three-dimensional space in which colours are placed in an ordered and precisely defined way, so that our screens, displays, TVs, and monitors can show them as we have intended them to be shown during the editing stage. In other and more simple words, you could think of a "colour space" as a box in which colours fit. It is logical that, in order to record, edit and render for viewing 1 billion colours (which necessitates the 10-bit recording you see in the metadata), you need a much larger "box" than if you would limit things to 16 million colours (as in sRGB). The nice thing about thinking of a "colour space" as being a box, a smaller or larger one, is that you can easily understand that the 1 billion colours that some cameras on DJI devices are able to capture, need a very large box, and that if you decide to edit and render the videos in a smaller "box" (like Rec.709!!!), millions of colours or shades of colours will be thrown out and lost during editing and/or rendering. The "strange" thing that I notice in your metadata of the D-LOG M file, is that it is definitely 10 bit, but it has been "converted" already inside the Osmo 3, during the recording of the images on file, to Rec.709!!! Something which doesn't seem to be the case with the Mavic 2 Pro, or at least it's not "shown" in the metadata as the information about the colour space which the camera's software used during recording, is simply not there, it hasn't been saved in the resulting file coming out of the Mavic 2 Pro...
In other words, editing a video that was recorded in a colour space like Rec.2020 (the very large box able to contain 1 billion colours) in Rec. 709 (a much, much smaller box with much less shades of colours present!), is possible but not without losing millions of shades (or nuances if you want) of colours. In still other words, Rec.709 will force the colours of Rec.2020 to be re-positioned and re-defined, and will unavoidably lose millions of shades of colours in the process.
Coming back to D-LOG M and the metadata that is being recorded (or not, as my Mavic 2 Pro, like I just said, doesn't record them for D-LOG M, whilst it correctly does so when I film in HLG), as you can see them on your files:
Although D-LOG M records in 10 bit and is therefore capable of capturing millions of colours more than Rec.709 can ever contain and display, it is necessary to edit the D-LOG M video files in Rec.709 to have more or less reliable results concerning colours and shades of colours (shades meaning different hues, saturation, and luminosity of a colour). That's why there's a Rec.709 conversion LUT for my Mavic 2 Pro, that a lot of people indeed use when editing D-LOG M movie files coming out of their Mavic 2 Pro.
Although the final results of videos edited and rendered in Rec.709 can look great, it is important to understand that D-LOG M records far more colours and that, because it is a proprietary format that is not standardised and therefore not "recognised" or used by NVLEs (Final Cut Pro, Premiere, Da Vinci, etc.), millions of colours are getting lost as it's necessary to edit D-LOG M files in Rec.709 in order to have a reliable or colour-consistent result! In other words, D-LOG M is overkill.
I have switched to HLG since I discovered everything I wrote about in my previous comments. HLG, in my opinion, only has advantages.
1. It's a standardised format, used by Sony and the BBC, two major players in the world of video.
2. It's a breeze to edit HLG movie files in a NVLE like Final Cut Pro because there's no need to use a LUT and there's way more latitude to work with concerning contrast, hues, luminosity, exposure etc.
3. H(ybrid) L(og) G(amma) has the best of both worlds, as the movies can be shown both on HDR displays (which are becoming more and more affordable and used everywhere) and on "regular" displays. Of course, the full quality of HLG only shows itself on HDR displays, but even on regular ones I see much more detail and shades of colours. (By the way, D-LOG M is NOT HDR...) So even if you use a non-HDR display to edit the HLG files, you get consistent and reliable results on colours and shades of colours (at least Final Cut Pro allows me to see them reliably on my non-HDR 5K iMac P3 Retina display).
4. If you keep your original files, today you can edit them as non-HDR, *and* in Rec.709 to accommodate every non-HDR display or monitor or screen out there. And tomorrow, you can edit them a second time in Rec.2020 if you're ready for HDR output. However, I now always edit and render in Rec.2020 as even platforms like YouTube and Vimeo are capable of dealing with the much larger colour space of Rec.2020.

Pegasos Posted at 2023-4-13 01:02
Dear Patrick,

You're welcome.

If you plan to display your final video online this is standard practice to convert to rec 709. Now you problalby already know that all video is to be converted to rec 709 upon export after working  in wide gamut working space within the the editing software but what you seem to be missing is that thats how all log profiles (not just DJI’s) from ARRI, RED is intended to be used.  Shooting color space —> working color space —-> display color space.

D-log and most logs are larger  color space than sdr (rec709) shooting formats from consumer cameras. You use that larger color space in your editor and then after eding and grading, THEN you color space trasnform for the display.

You can do that with rec709 lut or if you need hdr you can do that with a few additional  steps in resolve by first using HDR specific (color space aware) tools. I cant speak for final cut or premiere but I don't think you can do this in those applications.

Pegasos Posted at 2023-4-13 01:02
Dear Patrick,

You're welcome.

If you plan to display your final video online this is standard practice to convert to rec 709. Now you problalby already know that all video is to be converted to rec 709 upon export after working  in wide gamut working space within the the editing software but what you seem to be missing is that thats how all log profiles (not just DJI’s) from ARRI, RED is intended to be used.  Shooting color space —> working color space —-> display color space.

D-log and most logs are larger  color space than sdr (rec709) shooting formats from consumer cameras. You use that larger color space in your editor and then after eding and grading, THEN you color space trasnform for the display.

You can do that with rec709 lut or if you need hdr you can do that with a few additional  steps in resolve by first using HDR specific (color space aware) tools. I cant speak for final cut or premiere but I don't think you can do this in those applications.

Chevon Petgrave Posted at 1-12 21:35
If you plan to display your final video online this is standard practice to convert to rec 709. Now you problalby already know that all video is to be converted to rec 709 upon export after working  in wide gamut working space within the the editing software but what you seem to be missing is that thats how all log profiles (not just DJI’s) from ARRI, RED is intended to be used.  Shooting color space —> working color space —-> display color space.

D-log and most logs are larger  color space than sdr (rec709) shooting formats from consumer cameras. You use that larger color space in your editor and then after eding and grading, THEN you color space trasnform for the display.

You're mistaken in thinking it's a necessary and a "standard" practice to always convert to Rec.709 to output video and show it online. I will explain below. And you don't seem to understand or at least you don't mind what is happening to millions of colours when HLG or DLOG-M video clips are "converted" from a Rec.2020 colour space/gamut to a Rec.709 one - whether this is done during editing or during exporting. Those millions of colours are completely lost and have to be shifted into grades of colours that are present in Rec.709. My original post was and still is about what the actual use is of DLOG-M (and HLG) in a Mavic 2 Pro (and other DJI drones) if you can't use the full potential of those wider colour gamuts by having to convert to Rec.709 in your video editor software... DLOG-M is a proprietary format and Mavic 2 Pro drones don't even write the necessary metadata into the recorded video files, which then of course necessitates the use of a proprietary LUT to have DLOG-M files display properly in a Rec.709 colour space. DLOG-M is therefore nonsense to use because millions of colours are lost when outputting.   
There is absolutely no more need to follow the "old" standard practice and thus convert every video you want to output and publish online to Rec.709, at least not in my experience. Vimeo and YouTube fully support my HLG videos and whatever they do to them in their kitchen behind closed doors between uploading and publishing, gives correct results on several displays and screens I've used to look at them and verify things.
Recording in HLG makes total sense. Recording in DLOG-M hardly makes any sense anymore. HLG has the advantage it is not a proprietary format, the advantage that my Mavic 2 Pro drone writes the correct metadata into the video clips, the advantage that in Final Cut Pro it's a breeze to edit the clips (almost no correction needed compared to DLOG-M), the advantage that it uses a colour gamut that is far wider than the one DLOG-M uses, and the advantage that the final results - without any need to make two different versions (Rec.2020 and Rec.709) - look great on any display, whether that display is HDR or SDR.
Of course, the "old" standard practice is to convert to Rec.709. And if people want to stick to something they are used to and don't mind losing millions of colours, having to use LUTs, and/or spend a long time getting the colours look right, adjust luminosity and contrast etc. in their editors, that's fine with me. But it is totally counter-productive and I think they would understand what I mean when they would simply try out HLG on the Mavic drones that support it, and see how good the results look, in their editors and on their displays after output.
To your sequence "recording (shooting) colour space" > "editing (working) colour space" > "output (display) colour space" should be added: "1 billion colours" (HLG)> "1 billion colours" (if editing in Rec.2020 HLG)" > "16 million colours" (if outputting in Rec.709)...
Now I know that the human eye can't even see so many colours, but there is definitely a good reason to keep the 1 billion colours or colour grades in the final output, and a very good reason to shoot in HLG and not in DLOG-M when HDR displays are conquering the market (even my old iPhone X has an HDR display...).
My original post started out with a question: What happens to all those colours when recording in DLOG-M and editing/outputting in Rec.709 (the former having a wider colour space than the former)? The answer (partially thanks to what DJI engineering told me) is that it makes hardly any more sense to record in DLOG-M as it is a proprietary format that is not supported by any video editing software, thus has to be converted to Rec.709 whilst throwing away millions of colours (!), and doesn't look good without using a proprietary LUT or without the need to spend a lot of time editing colours, contrasts, luminosity etc.
If you can record in HLG, I recommend everyone to do so. It's HDR, it's easy to edit, and the final output looks great on any display out there.
  

I will try to explain this, as I understand it.

Dlog-M is useless.

As you mentioned

Q. What colour space is DLOG-M recording its movie files in?

A. Dlog-M, which is a secret, unknowable color space and transfer function.

Q. Why does DLOG-M not write the correct information about its colour space to the metadate of the movie files?

A. It wouldn't make a difference, as no other program on earth would know how to handle it. You could overwrite the meta data and make it say whatever you want, but since they are not using a published, standard color space, it will not provide any additional information. You would be right back where you started, with an undefined color gamut and transfer function. It is "mislabeled" as Rec709, but that is just so that it can actually be displayed at all by a display. I will come back to this point in a little bit.

Q. Why does DLOG-M appear to be using the much more limited colour space Rec. 709 whilst HLG is using the far wider and far more up-to-date Rec. 2020 HLG?

A. This misunderstands to some degree the relationship between the color gamut and the metadata. The camera is recording numbers like 0.0005, 0.12, 0.037, based on the signal it reads off its sensor. This doesn't necessarily "mean" anything. You can read these values and interpret them by using whatever color gamut you like, but that won't be a translation of what the camera saw or did. It has some mapping of voltage level to decimal number, and that is using Dlog-M, which is what determines what numbers relate to what voltage levels.

OK, so basically, we can see that adding a color profile that no one knows how to translate is basically useless, as if i had written this comment in a language I made up, that no one knows how to read. That is what DLog-M is.

Given that, in my opinion, DLog-M is completely useless, what is it ACTUALLY? It is tempting, as you have done above, to think, well it must be a variant of DLog...they did publish DLog, right? Maybe this is some sort of "light" version of that? No, it is not. From what I can see, DLog-M is just actually Rec709 using a 10 bit depth. If you look at the footage, it is not actually flat like a Log profile, it doesn't actually add dynamic range (that I have been able to test), and it is not easy to grade as we have no clue what the starting point is. You can actually see this when you look at what their provided LUT does...it just increases contrast a bit, and lowers exposure, resulting in a slightly darker, image with a tiny bit more contrast. I am pretty sure I could do the same thing to a Rec709 file, and get the exact same results.

One quick aside, there may be some confusion as to the "billions of colors", I can create a color profile that has 12 bit color, and only has shades of blue. I can create a color profile that has a smaller COLOR GAMUT than Rec709, but in 12 bit depth has MORE colors. If I told you there were trillions of numbers between 0 and 1, would that surprise you? Of course not. So BIT DEPTH means "How many pieces can I cut this thing into?", while COLOR GAMUT means "How big is the thing I am going to cut up?" So being able to cut up Rec709 into more pieces has very little usefulness, although it may avoid banding in gradations like skies. So technically, that is more colors, but it is not expanding the range of colors that are available.

So, let's do a quick breakdown between DLog-M and HLG.

Color Gamut:                         unknown vs Rec2020 clear winner is HLG
Transfer Function:                 unknown vs Rec2100 clear winner is HLG
Requires post processing:   DLog-M, yes using their LUT vs HLG, no as it is backwards compatible with SDR displays, it could be shared with no additional processing and look "correct".
Flexiblity:                                DLog-M is basically impossible to convert to an intermediate color space like ACEScct, HLG is easy to convert to ACEScct, so I can match footage shot from multiple cameras using HLG
Ease of Grading:                    DLog-M requires using a LUT, or manually guessing, HLG can use automatic color management, to easily move into whatever color space is desired
Accuracy:                                DLog-M is impossible to accurately work with, as you don't know how it was encoded, HLG is easy to accurately work with, because it is published and defined.

OK, so given that DLog-M is useless, why would DJI include it, and why would they not publish it, and why would they make it confusing by calling it Dlog-M (or DCinelike in another example)...

Hmmm, why would they try to name something similar to a well regarded, and respected Color Profile in a consumer drone, but not actually include any of the benefits of that professional color profile? Maybe to sell more drones that are not really meant to compete with their actual professional drones like the Inspire 3? It also seems like they, or at least their support, are not familiar with modern color management workflows, where DLog is WAY easier to work with than DLog-M, DLog-M is basically impossible to work with professionally. They say that the reason behind it is to make it "easier", and if you were randomly, manually spinning some dials, you would have to spin the dials less with DLog-M because its not actually flat and doesn't actually include a lot of color information. In fact, you could probably post DLog-M directly to YouTube and people might not notice, whereas you could definitely not do that with a real log profile.

So, to summarize, DLog-M is useless, I think its just a marketing ploy to trade on the useful professional color profile "DLog", and it provides no benefits. There is no scenario where it makes sense to use DLog...especially because you can just pick HLG, which addresses most of the issues with DLog-M. So, when would I pick DLog-M over HLG? Never.

If there was an option to remove it from the choices in the menus, I would, as I would never shoot anything on that drone that wasn't in HLG.

Hope that answers your questions @pegasos

By the way, the LUT's provided are all identical, I tested DJI Mini 4 Pro, OSMO 3, and Mavic, the files have the exact same contents, they didn't even change the comments. The LUT with the name
DJI Mini 4 Pro DJI D-Log M to Rec.709 has the following comment:

# DaVinci Resolve Cube (3D LUT).
#
#Generated by LUT Editor.
#Not-Clipped.
#
# Mavic 3 Pro, D-Log M, 2023-03-24
#

Pegasos Posted at 2023-9-11 06:50
You're welcome.
D-LOG-M and D-LOG will never be standards. They are proprietary format created by a drone factory, not a camera factory... HLG, however, is a standard, and since quite some time. Why would you not simply shoot everything in HLG, which has a far wider colour gamut/space than D-LOG (D-LOG-M has an even smaller one). I find HLG far easier to edit and my iMac 5K 27inch from 2019, which isn't even the latest and greatest computer out there, has no problems with it using Final Cut Pro.
If you shoot HLG, you have a very wide colour gamut/space, and you can easily convert the video clips from rec.2020 to rec.709 if ever needed. Rec.709 will become obsolete one day, perhaps sooner than we think.

DLog is published, included in Davinci Resolve, and has a whitepaper. DLog-M is (from what I have seen) just Rec709 with weird mappings, and has no future, and is useless.

Pegasos Posted at 2023-1-9 11:27
A video about what DLOG-M is and how you can use a LUT to make the clips look decent, does not give any answer to any of my questions about the COLOUR SPACE that DLOG-M is using...

In fact, the LUT being advertised in that video proves that, in order to be able to get any decent result from a DLOG-M movie file, you have to edit it in the Rec. 709 colour space! A colour space that is much smaller than the Rec. 2020 HLG one, which is being used by the HLG colour profile of the Mavic 2 Pro. Where are the 1 billion colours of the 10bit DLOG-M when you have to use Rec. 709?

Have you found the answer? I'm confused because I don't know how to grade dlogm because there is no color space transform for it

jojooost Posted at 3-16 08:29
Have you found the answer? I'm confused because I don't know how to grade dlogm because there is no color space transform for it

My question was not about finding a way to best colour-grade DLOG-M footage but more about why use DLOG-M at all when all the advantages of the vastly higher number of colours present in its colour space disappear when you have to edit and colour-grade  it in the vastly more limited Rec.709 colour space. I personally won't use DLOG-M any longer. Instead HLG is the way to go.
If you keep on using DLOG-M and want to have it "easy" to colour-grade it, use the DJI LUT inside Rec.709. It makes colour-grading easier, but you loose millions of shades of colour that were recorded.
DLOG-M was never, is not and will never be supported by editing software without having to use it within Rec.709.