Add GoPro Flat, Native WB to CST?

I know that GoPro Native white balance is a wider color space than Rec.709 but I have no idea what color space it is since there is very little documentation on it anywhere. Could a GoPRo Native WB option be added so that I could convert this color space to Rec.709 or Rec.2020?

The reason I don't want to use Auto WB is I am planning on using the GoPro as a C cam for an HDR project and converting Rec.709 to Rec.2020 will just result in gaps in color info, making for a color blocking mess.

Well, it shoots in 8 bit and IMHO, that's a no-go for HDR. You'll always get some banding for smooth transitions.
But its natural setting is fine as a starting point for Rec. 709, even with simply manual grading.

Well, it shoots in 8 bit and IMHO, that's a no-go for HDR. You'll always get some banding for smooth transitions.
But its natural setting is fine as a starting point for Rec. 709, even with simply manual grading.

I know that it is technically a no-go for HDR, but I am looking to incorporate it into an HDR project where I don't have the funds for a crash cam more than a GoPro. I'd love if they added 10-bit in their cameras but it seems they never will so I am stuck with 8-bit unless you know of a 10-bit camera as small and waterproof as a GoPro.

Converting rec.709 to rec.2020 will not result in "gaps" in color info. It will rather make sure that the data that was in the rec.709 source remain the same in the 2020 container. If you flag it as 2020 whilst being 709 you are stretching it out. As to luminance you don't need to scale content to 500nits or higher just because it's HDR. But it depends on how you need to use the footage alongside other material.

What you could do is just use a LUT or base grade to get it to a good level in SDR and then in HDR conversion just don't push the highlights/specular as much as better log/raw footage.

Riksha has made a DCTL for ACES if you really want to get into it though.
https://github.com/xtremestuff/protune-transforms/

Converting rec.709 to rec.2020 will not result in "gaps" in color info. It will rather make sure that the data that was in the rec.709 source remain the same in the 2020 container. If you flag it as 2020 whilst being 709 you are stretching it out. As to luminance you don't need to scale content to 500nits or higher just because it's HDR. But it depends on how you need to use the footage alongside other material.

What you could do is just use a LUT or base grade to get it to a good level in SDR and then in HDR conversion just don't push the highlights/specular as much as better log/raw footage.

Riksha has made a DCTL for ACES if you really want to get into it though.
https://github.com/xtremestuff/protune-transforms/

Thanks. I'll go take a look! I'm just hoping I don't get too much banding though the fast action will probably distract people from that since this will be mainly used for those kinds of shots.

Now the GoPro Hero11 has 10bit color, even for 4K 120P. So, if one can get a proper transform from the wide gamut "Native" WB color of the GoPro Hero11 to REC202 it would be very useful. The "Flat" profile is a log gamma profile, so a proper transform from that will also be useful.

Now the GoPro Hero11 has 10bit color, even for 4K 120P. So, if one can get a proper transform from the wide gamut "Native" WB color of the GoPro Hero11 to REC202 it would be very useful. The "Flat" profile is a log gamma profile, so a proper transform from that will also be useful.

10 bit, up to 200mbit, with fully disable-able NR. The 11 is worlds ahead of the prior GoPro.

https://gopro.github.io/labs/control/extensions/

"Super experimental, alter the log encoding for more dynamic range, or for a closer match with other camera’s log curves. i.e design your own flat profile. Ideal for use with 10-bit, and the existing flat color setting.
e.g. oMLOGB=400 for a little push in dynamic range or oMLOGB=200,-128 for a crude Fuji F-Log emulation."

I am still trying to think of a way without a Zyla to confirm if this log is better than Flat or Natural. I made posts on reddit, no one has any number for any of the 3.

If the Flog 'emulation' is accurate in the Gopro, we already have that CST in Resolve. To my eye it still seems too flat after the CST though. Or maybe just a lack of saturation. Again I wish someone with proper test charts would post test on it.


Though the Tone Mapping algorithm may mean that such tests are meaningless as the 'Algorithm' will no doubt make different decisions for a Zyla than an beach. And we can't turn it of in the Hero 11.

"Tone-mapping controls. Tone-mapping is the in-camera contrast control, dynamically adjusting the video to look good under a range of lighting conditions. HERO10 adds LTM - Local Tone-Mapping, enabling you to see details in leaves and grass textures, way better than all previous GoPro’s. HERO9 and earlier, used GTM, Global Tone-Mapping which adjusts the contrast curve for the image automatically. If you wanted to do these in post, you could use Protune Flat, where all in-camera tone-mapping is disabled and a log curve is applied (do not use these control with Flat, can be weird results.) For a more developed Rec709 video, by shooting GoPro Color or Natural modes, but you wanted to do your own tone-mapping in post–you can now do that. Note: HERO11 is a tad odd with TONE=3, as it tends to under-expose. Always experiment before using an feature. "

Now the GoPro Hero11 has 10bit color, even for 4K 120P. So, if one can get a proper transform from the wide gamut "Native" WB color of the GoPro Hero11 to REC202 it would be very useful. The "Flat" profile is a log gamma profile, so a proper transform from that will also be useful.

10 bit, up to 200mbit, with fully disable-able NR. The 11 is worlds ahead of the prior GoPro.

https://gopro.github.io/labs/control/extensions/

"Super experimental, alter the log encoding for more dynamic range, or for a closer match with other camera’s log curves. i.e design your own flat profile. Ideal for use with 10-bit, and the existing flat color setting.
e.g. oMLOGB=400 for a little push in dynamic range or oMLOGB=200,-128 for a crude Fuji F-Log emulation."

I am still trying to think of a way without a Zyla to confirm if this log is better than Flat or Natural. I made posts on reddit, no one has any number for any of the 3.

If the Flog 'emulation' is accurate in the Gopro, we already have that CST in Resolve. To my eye it still seems too flat after the CST though. Or maybe just a lack of saturation. Again I wish someone with proper test charts would post test on it.


Though the Tone Mapping algorithm may mean that such tests are meaningless as the 'Algorithm' will no doubt make different decisions for a Zyla than an beach. And we can't turn it of in the Hero 11.

"Tone-mapping controls. Tone-mapping is the in-camera contrast control, dynamically adjusting the video to look good under a range of lighting conditions. HERO10 adds LTM - Local Tone-Mapping, enabling you to see details in leaves and grass textures, way better than all previous GoPro’s. HERO9 and earlier, used GTM, Global Tone-Mapping which adjusts the contrast curve for the image automatically. If you wanted to do these in post, you could use Protune Flat, where all in-camera tone-mapping is disabled and a log curve is applied (do not use these control with Flat, can be weird results.) For a more developed Rec709 video, by shooting GoPro Color or Natural modes, but you wanted to do your own tone-mapping in post–you can now do that. Note: HERO11 is a tad odd with TONE=3, as it tends to under-expose. Always experiment before using an feature. "

We are getting closer: the latest GoPro Labs firmware update for the Mini appears to allow setting the primaries of the color gamut to REC2020:

"Added WIDE (flat color wider gamut, like wN with white balance controls), 2020 (color primaries), MTRX (color matrix) extensions."

Ok, 5.3K 60P Linear and REC2020 color gamut, 10bit color, Fuji Flog in camera, DaVinci Resolve Studio transform from REC2020 gamut and Flog gamma to HDR PQ:

Zero adjustment from the transform in post except to increase color saturation.

So no need to figure out what Native WB is, and the Fuji Flog emulation seems to work OK.

We are getting closer: the latest GoPro Labs firmware update for the Mini appears to allow setting the primaries of the color gamut to REC2020:

"Added WIDE (flat color wider gamut, like wN with white balance controls), 2020 (color primaries), MTRX (color matrix) extensions."

Note from David Newman on Github:

oMWIDE=1 enables a wide color gamut, sensor native, while still supporting white balance.

oM2020=1 it is such a crude hack, I was not intending to list it. It does use 2020 color targets, but it still has limited gamut. So don't use this, as there is zero advantage resulting.

We are getting closer: the latest GoPro Labs firmware update for the Mini appears to allow setting the primaries of the color gamut to REC2020:

"Added WIDE (flat color wider gamut, like wN with white balance controls), 2020 (color primaries), MTRX (color matrix) extensions."

Note from David Newman on Github:

oMWIDE=1 enables a wide color gamut, sensor native, while still supporting white balance.

oM2020=1 it is such a crude hack, I was not intending to list it. It does use 2020 color targets, but it still has limited gamut. So don't use this, as there is zero advantage resulting.

That is disappointing, but I do not think there is "zero" advantage. First, of course the REC2020 "hack" still has a limited gamut - I do not know of almost any viewing device or camera that can reproduce the full REC2020 gamut. Second, using the wider GoPro "Native" color gamut with no known properties, we have to in post "crudely hack" a transformation of that gamut to some standard, like REC709 or REC2020, ourselves. It would seem that we would be better off using the GoPro REC2020 approximation specified as such as the input gamut in Resolve for the transformation.

Finally, my experiment with the REC2020 GoPro "hack" gamut yields good-looking colors translated in DaVinci Resolve to both REC709 and REC2020.

I do agree that first best is starting with the full Hero 11 sensor color gamut, but for that GoPro has to provide the IDT, or a LUT (which is just a crude approximation).

Technical gamut transform is linear by nature so should be representable with LUT without issues as LUT interpolation should produce correct values within the precision of LUT data itself. Whether it is practical to do so is another story.

Note from David Newman on Github:

oMWIDE=1 enables a wide color gamut, sensor native, while still supporting white balance.

oM2020=1 it is such a crude hack, I was not intending to list it. It does use 2020 color targets, but it still has limited gamut. So don't use this, as there is zero advantage resulting.

This is pulled from the release notes. Not yet on the 'man page'. Hopefully they add a bit of documentation when they post there

https://gopro.github.io/labs/control/extensions/

What does 'wide gamut' mean? How are we supposed to transform it back to Rec709?

I tried to use the following on my 11

Code: Select all

CBAR=1 - enable a small 75% saturated color bars for video tools evaluation (HERO10/11 limitation: only works 4Kp30 or lower res/fps.)
LBAR=1 - enable a small luma sweep for video tools evaluation (HERO10/11 limitation: only works 4Kp30 or lower res/fps.)

Hoping I could use it do decode the color space they are using. But it doesn't do anything, just beeps that it read the QR and nothing. Yes, using 4k/24.

Wide Gamut is the same thing as Native, except that it includes a white balancing component whereas Native is purely just the sensor data.