S-Log workflow suggestions

Help/How To
1

I am a video editing newbie and have just started filming in S-LOG3 using a Sony ZV-1,
and wanted your opinion on what could be the correct workflow when working
with S-Log videos.

Technically, using the Sony PP8 is using S-Log3 gamma and the S-Gamut3.Cine color mode, see ZV-1 Manual: ZV-1 | Help Guide | Picture Profile

Intuitively, I think this should be the filter pipeline:

  • exposure correction
  • white balance
  • LUT 3D (to transform from S-Log stuff to rec709)
  • (…other color corrections…)

I am still confused about 8bits, 10bits and what would be the best S-Log version to use, so I will probably experiment also using S-Log2 and HLG.

I found this LUTs Software & Downloads - Sony Pro from Sony but seem to not apply to the ZV-1.

Do you have any documentation to learn or workflow you use that satisfies you?
Thanks!

2

I think your pipeline is correct, there are however a few issues you need to think about.

Firstly, S-Gamut3.Cine is very wide, so wide it has non-visible primaries. Your main problem with using the LUT is that LUTs work in R’G’B’ space but your video files are in Y’CbCr. So you need to use the correct matrix to convert between the two and I’m not sure Shotcut has it. If you use the 709 matrix, you’ll get some hue shifts across the image.

image

You should be able to white balance the input using a CbCr vectorscope, but note that bars shouldn’t line up with the 709 targets.

The other thing to remember is that the LUT will have expected levels - unlike any other format I can think of, SLog has black at 10 bit 95 rather than 64. There’ll also be a target value for diffuse white in the input video - the further you move from that, the more likely the LUT won’t work well.

3

If one is so worried about their log converting to Rec. 709 in a mathematically correct manner, then is there really any advantage to shooting in log over rec. 709? I thought the main purpose of log was to facilitate manual adjustment rather than trusting your software and knowledge to be better than the camera’s. I am not really making statements of fact; I am still learning about log too. Maybe it depends on which color gamut the log format uses.

See the other new thread where I mention numerical pixel format issues with respect to Shotcut’s internals. Given that, you need to be mindful when any filter is likely to introduce clipping and move those to the end of the pipeline. Position the filters that need the elbow room of log earlier.

4

There is no single mathematically correct method as moving from a high dynamic range to 709 requires a tone curve and gamut mapping, however you can get much better highlight retention and colour rendition than using a 709 camera which clips. However with the wrong YCbCr matrix, you’ll have hue angle dependant hue shifts, which are not possible to fix using a set of colour wheels.

It’s been a number of years since I’ve used S-Log, but I think you’ll see something like a shift towards green in skin tones and towards red in the sky.

5

With my 360 footage, I find I am often (maybe always? kinda subjective) able to get better results than the built-in color science of my camera. In fact, not liking how the default footage looked color-wise is what made me start experimenting with Log color. It has taken massive work for me to get to the point where my color grading looks better than straight out of the camera auto settings, but I think I’m there. I usually use a combination of DaVinci Resolve, Affinity Photo and Capture One to create LUTs using custom made PNGs and test on various sample Log frames from various scenes.

My workflow mainly uses various CSTs in DaVinci Resolve (and things like Hue to Hue, Hue to Sat, curves, color balance, and many others, while watching the color scopes), combined with other tweaks in Affinity Photo. But I’m not knowledgeable enough to answer the original question. I have no technical knowledge of gamuts and CSTs. I’ve just experimented with hundreds, if not thousands of combinations until something “looks right.”

TBH, if someone like me can experiment and get better results in a universal LUT than a company who’s job it is to create accurate color capture, it should be kinda embarrassing for said company. But eh, whatever, I’m not the arrogant type. I just hope they keep improving their color science so I have to do less work in the future. Every year Insta360 releases a new camera or 3, so maybe I’ll eventually upgrade to something with better clarity and color.

Oh, I forgot to mention I also use LUTCreator.js which is a cool online tool. Sometimes certain color or exposure aspects aren’t supported by native LUT export functions (In DaVinci Resolve or Affinity Photo), so if I make my own PNG and apply all filters to it, I often get better results.

The main reason I can’t answer the question though is because my camera shoots in an unknown Log profile. It’s not Sony S-Log or any other profile in DaVinci Resolve. I experimented with every possible CST until I found something kinda close, then did a bunch of subtle manual color and exposure transformations.

One thing I will say is that USUALLY (not always), I do all color and exposure transformations before applying a LUT. There seems to be less color banding this way in resulting footage, especially in the sky.

6

This topic was automatically closed after 90 days. New replies are no longer allowed.