Gopro Wide Extreme saturation after convertion

[markusfinholt]

I tried out Gopro Wide+LOGB=400 to DWG and it gave me some really pink and saturated results.

Here is the normal shot without wide enabled. Only LOGB=400

Looks fine, colors are accurate after adding contrast.

Here is WIDE=1 added. But no convertion.

Here is the WIDE=1 enabled, converted to DWG, and output to REC709, Gamma 2.4

Very dark result, and since the mid gray is so high, I added 2.5 stops of linear gain.

Extremely pink results, and super saturated.

Desaturating seems to make it very useable, but it's still noticeably more pink than it's supposed to be.

Node tree is the following:

Node 1: IDT (gopro native to DWG)
Node 2: 2.5 Stops of linear gain
Node 3: ODT (DWG/DWG to Rec709, Gamma2.4)

Care to chime in on what I'm doing wrong?

[RikshaDriver]

If you're using WIDE=1, this is equivalent to using the NATIVE values with auto white balancing. The camera is attempting to automatically perform white balancing and the tint/temperature are baked into the video.

If you're compressing the full gamut to a Rec.709 color space, you will need to perform some form of gamut mapping to bring the saturation down, otherwise everything will be oversaturated. What you're seeing is largely expected.

Also keep in mind that the transforms are using a standard LOGB value of 113, not 400, so the correction curves are not the same.

[markusfinholt]

Very happy with the GP-Tune transform so far. But I haven't found a good solution for gamut compression yet. The built in plugins in resolve are not compressing the gamut nearly enough, even with very strong compression, so I end up using the saturation knob to about 33 in Davinci Wide Gamut. I feel bad reccommending this workflow to noobs without a built in gamut compression setting in the plugin. Is this something you could develop?

[RikshaDriver]

You will find ACES has stronger gamut compression. I personally recommend the ACES workflow as the DaVinci Color Managed workflow has a lot of hidden gotchas and inconsistencies.

Also don't forget that sometimes the Labs settings don't stick. You may have a scenario where you are thinking that the footage was shot with WIDE=1 but it's actually still limited to Rec.709 gamut. If you use the WIDE or Native input transforms on this footage, the colors will be oversaturated. You can confirm that WIDE=1 is actually set by checking the metadata in your original clip:
https://gopro.github.io/labs/control/metadata/

Towards the end of the list, you should see something like: WIDE 1

[markusfinholt]

Interesting, didn't know that labs doesn't always stick to the recordings. I haven't had any issues with that yet I think.

What kind of gamut compression curve would you suggest I use for the wide/native shots? A strong gamut compression towards the more saturated parts, or a general desaturation method for the whole shot? The shot looks very evenly saturated, if that makes sense.

[RikshaDriver]

Using !M is permanent whereas oM doesn't withstand a reboot. There are also some quirks sometimes.

Generally ACES 1.3 with Reference Gamut Compress has worked well for me - sometimes a bit too well in the desaturation. If I need to desaturate a particular color (i.e. Blue), I use the Hue v Sat to bring down the saturation level for that color.

Circling back to your original post, have you checked that your original footage is in fact WIDE=1?

[markusfinholt]

My footage is Wide=1 yes, I tested it. I'll look into some more desaturation methods to see which one works. I'm not going to be leaving RCM anytime soon, as I've gotten too deep into the system to change to ACES.