Color Management Workflow in DaVinci Resolve 16

On 5/16/2019 at 11:22 PM, Grant McNair said:

Really love this, and all of the LowePost courses.  One thing that is missing, and would make all courses better, are speed controls on your video player.  Thanks!

Thanks for your kind words Grant. Click the wheel and clock and you will see the speed controls

Thanks!

This is great and really useful knowledge.  I'm noticing an issue that I hoped this would fix but still struggling to get it to work properly.  If anyone knows how to fix then that would be really much appreciated.  So here's the issue:

• Using Davinci Resolve Studio 16 (beta 5)
• Using DVR Colour Managed (RCM)
• Input CS set to - Slog2 (as most of the footage was shot with SLog 2)
• Timeline and Output CS set to - REC709 Gamma 2.4 (as output is web)
• Create a title in Fusion Using the Text+ or just clicking the clip I need to add title VFX to.
• Create my title in Fusion, adding a png graphic
• Colours look washed and .png graphic no longer looks sharp, pixelated edges.
• Cannot find a gamut or colour space transform that gets this to look correct.
• Flicking back from RCM to Davinci YRGB and the problem goes away.
• I am not able to set the input colour space for the fusion clip to bypass the input colour option in RCM settings.

Is this a bug in the beta, I know, I know - it's still a beta but I feel this is something that I am just not doing correctly rather than being a beta issue.

Can anyone help.  I've attached screenshots to show the issue.

1 hour ago, Alex Cameron said:

This is great and really useful knowledge.  I'm noticing an issue that I hoped this would fix but still struggling to get it to work properly.  If anyone knows how to fix then that would be really much appreciated...

After doing some testing with a PNG logo, I see that the PNG transparency is interpreted poorly with RCM - even when the PNG's input color space is set to sRGB, The workaround would be to import the PNG directly into Fusion through a Loader tool, which essentially allows it to bypass RCM. This would require a bit of color grading in Fusion to match the PNG to the sLog footage, which would come in through the MediaIn tool. You could also consider pushing the PNG into sLog space within Fusion by adding a OCIO Colorpsace tool.

13 hours ago, Lee Lanier said:

After doing some testing with a PNG logo, I see that the PNG transparency is interpreted poorly with RCM - even when the PNG's input color space is set to sRGB, The workaround would be to import the PNG directly into Fusion through a Loader tool, which essentially allows it to bypass RCM. This would require a bit of color grading in Fusion to match the PNG to the sLog footage, which would come in through the MediaIn tool. You could also consider pushing the PNG into sLog space within Fusion by adding a OCIO Colorpsace tool.

Thanks so much for your speedy response Lee.  I'll give what you've suggested a go.  There's definitely something wrong with the .png interpretation but beyond that the background is still not behaving properly and this is a background created in Fusion.  I have managed to get it looking somewhat normal by applying a Gamut tool and then changing the input colour space and gamut in the actual background tab.  Is the Loader tool native to Fusion or does this require a third party plugin.  Also is the OCIO Colourspace tools one of the ones within Fusion natively?  Thank you so much for your help.

6 hours ago, Alex Cameron said:

 Is the Loader tool native to Fusion or does this require a third party plugin.  Also is the OCIO Colourspace tools one of the ones within Fusion natively?  Thank you so much for your help.

The Loader tool is native, as well as OCIO Colorspace. However, with the OCIO Colorspace tool, you'll need to download the OCIO config file and load it into the tool. The config files are here: https://opencolorio.org/index.html#

Hello thank you for this course! It's great to get some reassurance on somethings I've learned elsewhere. (Going to look at skin retouching next) 

However I'm now thinking about changing my workflow. I've been using a RCM workflow with SLog3/.Cine footage from an a7S and I've been bypassing the Input Colorspace as well as not assigning clips a profile in the Media Bin. I'd basically grade to where I want it to look and have the Output Colorspace set to sRGB / rec709 in Project Settings. So it sounds like I've basically been using a display referred system. 

I'd like to follow your lead and have the Input Colorspace match what I'm shooting (slog3) in Project Settings but when I do so my highlights are clipped even though I've exposed using a display with a r709 LUT. I'm guessing this might be from the limited data range on my camera (8-bit 4:2:0)? When I turn tone and gamut mapping on (along with the necessary drop down options) it displays about how I shot it, but like you said this is destructive and I can't grade it much down the line. 

Any recommendations?

In general if I use RCM should I just set the Input Colorspace to what I shot on in the Project Settings and grade from there? Will things like noise reduction work as good this way or should I bypass Input Colorspace and have my last node be Color Space Transformation? 

Thank you! 

1 hour ago, James Conkle said:

However I'm now thinking about changing my workflow. I've been using a RCM workflow with SLog3/.Cine footage from an a7S and I've been bypassing the Input Colorspace as well as not assigning clips a profile in the Media Bin. I'd basically grade to where I want it to look and have the Output Colorspace set to sRGB / rec709 in Project Settings. So it sounds like I've basically been using a display referred system. 

I'd like to follow your lead and have the Input Colorspace match what I'm shooting (slog3) in Project Settings but when I do so my highlights are clipped even though I've exposed using a display with a r709 LUT. I'm guessing this might be from the limited data range on my camera (8-bit 4:2:0)? When I turn tone and gamut mapping on (along with the necessary drop down options) it displays about how I shot it, but like you said this is destructive and I can't grade it much down the line. 

Any recommendations?

In general if I use RCM should I just set the Input Colorspace to what I shot on in the Project Settings and grade from there? Will things like noise reduction work as good this way or should I bypass Input Colorspace and have my last node be Color Space Transformation? 

 

If you are using RCM, and the footage is not RAW, then you'll want to try to apply a matching Input Color Space. As for the clipping, perhaps there is a mismatch between the camera exposure LUT and the default Rec 709 space in Resolve. If there is clipping, maybe skip the Tone/Gamut Mapping and instead grade the footage into an acceptable range in the Color tab - at least that pushes any destruction of color values to the end of the pipeline.

Not sure if that helps - I must admit color pipeline issues can be very convoluted.

Excellent! Would love to hear one in depth on HDR set up and workflows, and also having to deliver an SDR version. 

2 hours ago, Sean Henderson said:

Excellent! Would love to hear one in depth on HDR set up and workflows, and also having to deliver an SDR version. 

Currently planning a "working with HDR" series. It will be the next one I record.