You know how everyone says 10bit is better than 8bit when it comes to heavy grading, I am finally testing that.

You know how everyone says 10bit is better than 8bit when it comes to heavy grading, today I am finally testing that. It makes sense but I wanted to see it. Anytime I’m curious usually something unexpected happens.

Tests not about detail/sharpness, not about the codecs or 4:2:2 vs 4:2:0. Not talking about rolling shutter, etc.

I compared the Sony a7Sii to the Canon C100, Sony FS5 and the Red Scarlet Dragon.

Conclusion

  • I guess the most surprising take away from this is: while the a7sii is only 8 bit camera it can hold it’s own when shot at 4K when you really have to stretch the image. The advantages you get from 10 bit on the FS5 or the 12 bit from the Scarlet when stretching the image kind of goes away because when using slog or other formats that require high ISO’s makes the images so noisy that the cleaner 8 bit image holds up better when you stretch the crap out of it.

Like I have said many times in this video I am not an expert at color, I am just sharing my experiences so others can learn from me or my mistakes.

Special thanks to Wade Yamaguchi and James Drake.

 

12 comments

  • Hey Dave, let me just say that first off, despite that you’re no Roger Deakins, I have definitely watched a bunch of your videos over the years and truly appreciate what it is you do. Coming off a two month shoot where our A camera was the FS5 and the B camera was the A7Sii, I have to say that on set, I appreciated both cameras for different reasons but we’re now coming to the post production and I’m learning how well the A7sii actually holds up in colouring (we’re also finishing the project in 1080). At times the two cameras are indistinguishable and identifiable only by the depth of field in certain scenarios. I think my biggest pet peeve is how much the codecs fall apart at higher frame rates (on both cameras)with the slog’s and in the future I may stick to shooting a more standard or neutral PP when ramping the frame rate up on these cams.

  • Thanks Dave. One of the most interesting videos you have done. This is exactly what I was wondering. And i thought it was a massive difference in 8 vs 10 bit.

  • Dave,
    Terrific video. Very helpful, practical results.

    How you wade into the details of this stuff without….A: Having your head explode or B: Punching yourself in the face…is beyond me. But thank you for doing so!

  • This is a very helpful video, indeed! I’ve been hearing the claim that 4k 4:2:0 can be transcoded to 4:2:2 10 bit, but I’ve never tested myself, and this video somewhat proves it. Thanks for your extensive work.

  • You’re more of an expert than you realise Dave. Another superb video and put across in such a clear unbiased way. Respect. 10/10

  • Hey Dave, thanks for sharing these tests, and all of the work thats required in doing them.

    I’m not sure that I agree with your conclusion though.

    Comparing bit depth at different resolutions, yields inconclusive results in my opinion. The very first test was apple to apples, and showed that bit depth does matter in having a chance at recovering a usable image from underexposure.

    Most of the following tests though, didn’t compare on even ground. The internal codec on the FS5 records 1080 10 bit at 50mbps, and the A7s/r records internal 4k 8bit at 100mbps. That shows that even internally you have twice the amount of image information to work with. Even going straight to prores on the shogun, you’re looking at more data in the 4k vs 1080.

    It may not be a true 10bit 1080, when downscaled from 4k, but there is absolutely more data there than true 8 bit.

    I totally see where you’re going with the comparison, and wanting to find out limits/latitude, but I think you may have missed on this one.

    I enjoy watching your videos, and I’ve definitely learned from you over the years. Thanks for producing great content.

  • Thanks for the video Dave, you always provide great info.

    As a new owner of the FS5, shooting in SLOG3, it is crucial to shoot with your highlights just under over exposure to reduce the noise when color corrected. It’s kind of the opposite way you are used to shooting as you are always trying to preserve the highlights, but not with the FS5 in SLOG3.

    So I think this is also a big take away for viewers and maybe a future test for you – shoot in SLOG3, expose your highlights just before they clip, color correct, and then take a look at the noise. It should be significantly reduced shooting this way.