F-log2 and "X H2s -> Video Assist BRAW" in RCM workflow

Good day!

I have the new Fujifilm X H2-s camera and trying to figure out the best workflow for it in Resolve color managed environment.

1) When shooting internal ProRes I would like to take advantage of the new F-log2 profile.

I have the RCM set up like this:
Color science: DaVinci YRGB Color Managed
Color processing mode: HDR DaVinci Wide Gamut Intermediate
Output color space: Rec.709 Gamma 2.4

If I understand correctly I need to assign Input Color Space for each of the Prores clips separately (or from the Media tab in bulk).

The problem is that that in the list of color spaces there is only "Fujifilm F-log" and no F-log2 option.

Does this mean that the Resolve does not have the F-log2 option available (yet)? Or is there a workaround for this that I am not getting?

Could the workaround be that I grade the F-log2 image using a color checker, grab a astill and apply it to all my future F-log2 ProRes clips. Or am I being too simplistic here (highly possible).


2) When shooting BRAW externally from X H2-s to Video Assist 12G I get BRAW files, but under the "Camera raw" tab it states it is in Rec.2020 color space and gamma as BMD Extended Video.

Is this correct and RCM managing it correctly? From the "right click on clip" list I can only "Bypass Color Management", no input option available so I guess I can trust it is dealing correctly with it.


3) And a bonus question

If I were to use the Resolve Film Looks LUT for such RCM workflow would the correct compound node include something like this:

CST: DaVinci Wide Gammut Intermediate -> Rec.709 Cineon Film Log
LUT: Rec.709 Kodak 2383 D60
CST: Rec. 709 Cineon Film Log -> DaVinci Wide Gammut Intermediate

Or is my logic meesed up here? I.e converting from DaVinci Wide Gammut Intermediate to Rec.709 and back again be sort of destructive?

You can look at third-party transforms to bridge the gap for F-Log2 footage in the immediate term.

I would personally recommend using ACES over RCMv2 because of all the issues with forced Input & Output DRT settings as part of the default presets.


With BRAW you shouldn't be able to define an input color space for the clip.

I also got the X-H2S but haven’t had time for a lot of testing. For your #1: I use a CST from Rec.2020, Fujifilm F-log to Rec.709. Images look good! I currently prefer the CST approach over Davinci color managed workflow because of the color changes with Fusion clips.

Does this mean that the Resolve does not have the F-log2 option available (yet)? Or is there a workaround for this that I am not getting?

There is no F-Log2 yet.

When shooting BRAW externally from X H2-s to Video Assist 12G I get BRAW files, but under the "Camera raw" tab it states it is in Rec.2020 color space and gamma as BMD Extended Video.

If the BM VA has official support for that camera it will (should?) use the data correctly to store in BRAW meaning Resolve can decode to Blackmagic's transfer functions or any other you pick from the available list entries. It also means that it is automatically managed if using RCM or ACES.

Rec.2020 is the gamut Fuji records the camera primaries in. They call it F-Log-Gamut but the coördinates are the same so it's not added as a unique name in software for convenience probably, only on the white paper.
https://dl.fujifilm-x.com/support/lut/F ... er.1.0.pdf

If I were to use the Resolve Film Looks LUT for such RCM workflow would the correct compound node include something like this:

CST: DaVinci Wide Gammut Intermediate -> Rec.709 Cineon Film Log
LUT: Rec.709 Kodak 2383 D60
CST: Rec. 709 Cineon Film Log -> DaVinci Wide Gammut Intermediate

Or is my logic meesed up here? I.e converting from DaVinci Wide Gammut Intermediate to Rec.709 and back again be sort of destructive?

This is not correct. The film print emulation LUTs shipping with Resolve both emulate the film and convert from Cineon to Rec.709 Gamma 2.4. So the second CST should go from Gamma 2.4 to Intermediate.

And yes this is destructive in the sense that because the LUT targets a specific display you can't upscale reliably to wider gamut or HDR. You could use the P3 versions instead to mitigate this but that's still not optimal.

If you know you'll never output for a different display but just want to have the Intermediate space to grade in underneath there will be no issues at all I'd say. But ultimately creating or finding a LUT or PowerGrade that is designed in scene referred state and outputs as that same state is the better choice.
https://freelut.cullenkellycolor.com/
https://juanmelara.com.au/products/koda ... powergrade

I also got the X-H2S but haven’t had time for a lot of testing. For your #1: I use a CST from Rec.2020, Fujifilm F-log to Rec.709. Images look good! I currently prefer the CST approach over Davinci color managed workflow because of the color changes with Fusion clips.

I personally also prefer manual approaches. What I read from the paper middle gray sits lower in FLog2 and the total curve covers an extra stop below and above, likely to account for an improved sensor. It shouldn't give any severe issues using FLog for simple projects, but for VFX and higher level grading and technical work I would probably look into having someone creating a LUT to convert FLog2 to Flog or create an ACES input transform for FLog2 based on the white paper data.

I also got the X-H2S but haven’t had time for a lot of testing. For your #1: I use a CST from Rec.2020, Fujifilm F-log to Rec.709. Images look good! I currently prefer the CST approach over Davinci color managed workflow because of the color changes with Fusion clips.

Now trying the CST workflow out I agree with you. Takes a little more time to set up, but at least it seems more solid and I know exactly what the settings are doing.

Rec.2020 is the gamut Fuji records the camera primaries in. They call it F-Log-Gamut but the coördinates are the same so it's not added as a unique name in software for convenience probably, only on the white paper.

Ah now I got it, this "F-gamut" got me confused.

This is not correct. The film print emulation LUTs shipping with Resolve both emulate the film and convert from Cineon to Rec.709 Gamma 2.4. So the second CST should go from Gamma 2.4 to Intermediate.

Ah of course, it even states that in the name of the LUT

And yes this is destructive in the sense that because the LUT targets a specific display you can't upscale reliably to wider gamut or HDR. You could use the P3 versions instead to mitigate this but that's still not optimal.

If you know you'll never output for a different display but just want to have the Intermediate space to grade in underneath there will be no issues at all I'd say. But ultimately creating or finding a LUT or PowerGrade that is designed in scene referred state and outputs as that same state is the better choice.

I will try to stay on the flexible side of things and use conversion to rec.709 only as a last CST.
I will give CKC Kodak 2383 a go.


Can anyone comment on the Elements LUT bundle Cullen kelly has there VS getting film print emulation powergrades from Juan Melara for DWG?

Any third party F-log1->Flog-2 conversions out there?

Can anyone comment on the Elements LUT bundle Cullen kelly has there VS getting film print emulation powergrades from Juan Melara for DWG?

I don't have the bundle but it looks like a bunch of creative grades. That's not the same as film print emulation. Film print emulations are a solid way to build a look from where creative grades from someone else, nor the look for a client and are just a bit random if you ask me. Never really fits the taste of someone else and doesn't provide any logical basis because it's heavily dependent on what has been captured to look interesting or not. So if you ask me what to get as a beginner or a pro wanting to expand it's library it's definitely FPE. The rest of the grading should be in your own hands.

If you have all XH2S in your timeline you can apply the CST to the timeline. Just a little extra effort. But don’t wonder if the GoPro Clips you drop on this timeline look weird

Not exactly your topic (ProRes Raw) but he shows something interesting about film grades. Min. 10…

Rather than start a new thread, I am reviving this one.

As a new owner of the Fujifilm X-H2S I have been experimenting with color management and workflow to best ingest and transform F-Log2. So far, the best results I have found have been through using XtremeStuff's ACES IDT (linked below) within an ACES managed project.

However, I am interested in trying the Davinci RCM Wide Gamut workflow as well to see how the results differ. However, without an F-Log2 input transform, this makes it quite difficult. I have tried the ARRI workaround and doesn't seem close enough for me to want to move away from ACES.

Any hope of F-Log2 coming to Resolve any time soon? When the new version dropped today, I was hoping . . . alas, no luck yet.

https://xtremestuff.net/store/fujifilm-f-log-transform/

So far, the best results I have found have been through using XtremeStuff's ACES IDT (linked below) within an ACES managed project.

A while ago a similar thread was going on on ACESCentral. For starters there is a site where you can generate IDTs which supports FLog. However it doesn't support FLog2. In that ACESCentral thread however I piped the numbers from the FLog2 paper into the IDT. Haven't done any proper testing on it myself but this should just work. In other words, there was no need to pay for some glorified DCTL to support FLog2.

However, I am interested in trying the Davinci RCM Wide Gamut workflow as well to see how the results differ. However, without an F-Log2 input transform, this makes it quite difficult.

You can use the ACES transform in conjunction with RCM if you do manual color management, so it's easy to grade in DaVinci Wide Gamut/Intermediate and use DaVinci as the DRT via CSTs.

Perhaps you are able to check the numbers in my IDT against the one you bought to see if they match.

Links:
FLog2 IDT: https://www.dropbox.com/s/fsvwqj6tdxqyy ... .dctl?dl=1
IDT DCTL Generator: http://acesidtdctl.tcolorscience.com/
ACESCentral thread: https://community.acescentral.com/t/fuj ... /24?page=2

So far, the best results I have found have been through using XtremeStuff's ACES IDT (linked below) within an ACES managed project.

A while ago a similar thread was going on on ACESCentral. For starters there is a site where you can generate IDTs which supports FLog. However it doesn't support FLog2. In that ACESCentral thread however I piped the numbers from the FLog2 paper into the IDT. Haven't done any proper testing on it myself but this should just work. In other words, there was no need to pay for some glorified DCTL to support FLog2.

However, I am interested in trying the Davinci RCM Wide Gamut workflow as well to see how the results differ. However, without an F-Log2 input transform, this makes it quite difficult.

You can use the ACES transform in conjunction with RCM if you do manual color management, so it's easy to grade in DaVinci Wide Gamut/Intermediate and use DaVinci as the DRT via CSTs.

Perhaps you are able to check the numbers in my IDT against the one you bought to see if they match.

Links:
FLog2 IDT: https://www.dropbox.com/s/fsvwqj6tdxqyy ... .dctl?dl=1
IDT DCTL Generator: http://acesidtdctl.tcolorscience.com/
ACESCentral thread: https://community.acescentral.com/t/fuj ... /24?page=2

Shebbe, thank you so much for your reply. I am afraid I am too much of a colorizing neophyte to understand all of that completely. I wouldn't know what to do with the ACES numbers. All I know is that when I use the ACES workflow my stuff seems turn out nicer!

That makes sense that I could transform the ACES mapping of F-Log2 in DRCM. But I am having trouble figuring out what I should transform to.

In a project set up for Davinci Wide Gamut I have an F-Log2 clip. Should the "Input Color Space" on the clip be "Same as Timeline" or "Project - Rec709 (Scene)"? Or should my project be set up some other way?

Screenshot 2023-03-08 113602.jpg (101.77 KiB) Viewed 4548 times

In my node tree, the first thing I have is an ACES transform going from F-log2 to Rec709. That doesn't seem to work right. Also, it seems the point is not to be working in Rec709, right?

Screenshot 2023-03-08 113720.jpg (20.97 KiB) Viewed 4548 times

Clearly I don't understand my color spaces well enough to puzzle out what I am coming from and going to. Which is why just right clicking on the clip and assigning a transform in ACES has been my preferred workflow until now. But I am trying to learn new ways of doing this, so any help would be appreciated. Thank you!

Chris

EDIT: Shebbe, I think you can disregard the above. I found your answer on ACES central forum that you linked too (screenshot posted below for anyone else looking for the same info). Just took me a while to recognize it. Thank you!

Screenshot 2023-03-08 133728.jpg (116.81 KiB) Viewed 4505 times

FLog2 IDT: https://www.dropbox.com/s/fsvwqj6tdxqyy ... .dctl?dl=1
IDT DCTL Generator: http://acesidtdctl.tcolorscience.com/
ACESCentral thread: https://community.acescentral.com/t/fuj ... /24?page=2

Oh, and this DCTL works perfectly. I don't know about the numbers, but it is visually identical to XtremeStuff. Coulda saved me some money to ask you first. Ah well.

I mentioned you could use it in RCM manually meaning no project wide color management. You could get it to work like that but the way RCM works makes it very jank to setup and even more janky to explain a beginner. The manual setup would be as follows.

Set project: DaVinci YRGB, timeline: DWG/Intermediate, output: Rec. 709/Gamma 2.4
First node: ACES Transform -> FLog2 to No Output Transform (meaning AP0/Linear, aka ACES 2065-1)
Second node: CST -> AP0/Linear to DaVinci Wide Gamut/Intermediate, nothing enabled except white point adaptation
// Grading //
Output node: CST -> DWG/Intermediate to Rec.709/2.4 OOTF enabled, tonemapping enabled

The double node input setup is required because DWG/Intermediate isn't available in the ACES transform at the moment. Maybe in the future.

The best way to set the input nodes up would be to group all your FLog2 footage and place them at the group pre clip level node tree so you set it just once for all the clips. The output node would go on the timeline node tree unless you are finishing projects inside Resolve with graphics and other elements on top of your footage then you'd place the output nodes on the group post clip level of your footage so the rest is excluded from that transform.

[edit]
I see that you found a similar message on the ACESCentral thread forgot it I put it there as well hehe.

I mentioned you could use it in RCM manually meaning no project wide color management. You could get it to work like that but the way RCM works makes it very jank to setup and even more janky to explain a beginner. The manual setup would be as follows.

Set project: DaVinci YRGB, timeline: DWG/Intermediate, output: Rec. 709/Gamma 2.4
First node: ACES Transform -> FLog2 to No Output Transform (meaning AP0/Linear, aka ACES 2065-1)
Second node: CST -> AP0/Linear to DaVinci Wide Gamut/Intermediate, nothing enabled except white point adaptation
// Grading //
Output node: CST -> DWG/Intermediate to Rec.709/2.4 OOTF enabled, tonemapping enabled

The double node input setup is required because DWG/Intermediate isn't available in the ACES transform at the moment. Maybe in the future.

The best way to set the input nodes up would be to group all your FLog2 footage and place them at the group pre clip level node tree so you set it just once for all the clips. The output node would go on the timeline node tree unless you are finishing projects inside Resolve with graphics and other elements on top of your footage then you'd place the output nodes on the group post clip level of your footage so the rest is excluded from that transform.

[edit]
I see that you found a similar message on the ACESCentral thread forgot it I put it there as well hehe.

Seems to be working. Thanks so much for your help!

I am the author of the glorified DCTL

As long as the BT.2020 to AP0 matrix has been calculated accurately with the appropriate Chromatic adaptation, there shouldn't be any difference between my version and whatever else is out there... at least as far as an ACES IDT is concerned.

Cristopher, if you are using an RCM workflow, you can use the F-Log Transform DCTL included in the Transform bundle you got...

As with anything RCMv2, make sure you set Input DRT to none... (Also there's no tangible benefit of using DWG if all your source footage is using Rec.2020 color space)

yrgb managed

YRGB_managed.png (59.55 KiB) Viewed 4436 times

then set input color space for the clips to same as timeline:

input color

input_color.png (129.24 KiB) Viewed 4436 times

And you can then select the DCTL from the DCTL list:

flog transform

flog-transform.png (57.12 KiB) Viewed 4436 times

I am the author of the glorified DCTL

As long as the BT.2020 to AP0 matrix has been calculated accurately with the appropriate Chromatic adaptation, there shouldn't be any difference between my version and whatever else is out there... at least as far as an ACES IDT is concerned.

Cristopher, if you are using an RCM workflow, you can use the F-Log Transform DCTL included in the Transform bundle you got...

As with anything RCMv2, make sure you set Input DRT to none... (Also there's no tangible benefit of using DWG if all your source footage is using Rec.2020 color space)

Riksha, thank you for chiming in! Your DCTL works great. Thanks for detailing that! I was wondering what that extra DCTL file in the LUT folder was for! This makes the purchase worth it.

May I ask why you set the timeline color space to Rec.2020 Intermediate rather than Davinci Wide Gamut? Thanks!

May I ask why you set the timeline color space to Rec.2020 Intermediate rather than Davinci Wide Gamut? Thanks!

There's no tangible benefit to using Wide Gamut since your source footage is limited to Rec.2020 regardless.

I am the author of the glorified DCTL

Hope I didn't offend you too much. The DCTL is definitely more feature rich than just being an IDT and has it's uses but for anyone just wanting an FLog2 IDT it's a little overkill hehe.

Also thanks for explaining an RCM setup. Not using Input DRT disables your ability to use display referred elements like sRGB images because it disables the inverse. So if your project isn't purely grading camera footage it's still not a preferable choice right?

I've been shooting F-Log2 in my XH2-s and when shooting to Proress have found that in the workflow you suggest after setting up for WideGamut in settings that you can go to the color page, right click a clip in the clip viewer and "input color space" option is there for Flog... It works.
Braw into my Video Assist 12G is a whole new ball game, and have found in Wide Gamut I grade as usual.
Braw is beautiful straight off the camera; after grading.

I am the author of the glorified DCTL

As long as the BT.2020 to AP0 matrix has been calculated accurately with the appropriate Chromatic adaptation, there shouldn't be any difference between my version and whatever else is out there... at least as far as an ACES IDT is concerned.

Cristopher, if you are using an RCM workflow, you can use the F-Log Transform DCTL included in the Transform bundle you got...

As with anything RCMv2, make sure you set Input DRT to none... (Also there's no tangible benefit of using DWG if all your source footage is using Rec.2020 color space)

YRGB_managed.png

then set input color space for the clips to same as timeline:

input_color.png

And you can then select the DCTL from the DCTL list:

flog-transform.png

For some reason, the IDT does not work for me when I add it as a DCTL, or when I put it in the ACES IDT folder. I don't know if it's just my system or the latest version of Resolve, but I can't get it to work. Anyone have any ideas?

For some reason, the IDT does not work for me when I add it as a DCTL, or when I put it in the ACES IDT folder. I don't know if it's just my system or the latest version of Resolve, but I can't get it to work. Anyone have any ideas?

Hmmm. Not sure what the issue is. It is working for me and is my go-to quick transform for ACES or DWG. You did put it in the folder in the %APPDATA% directory, yeah? IIRC, there are various directories that look similar and it can be confusing.