Let's try this test again but with less variables.

Got a lot of down votes than I normally do on the last 8 bit vs 10 bit video test you can watch here. I think the main complaint in the comments is I had too many variables in all the tests I did. I am learning, so let me try again and share my results, however this time I have a twist, not only will I share my conclusions, but I will have my friend Matt Scott also provide his conclusions.

You can download Matt’s files here if you want to play with the files.

You might get two different conclusions, one from a person that does some color grading and doesn’t push the image that hard, and another conclusion from someone that has a lot more experience and pushes the images more. While I record my conclusions, I have no idea what Matt’s conclusions are, and he will not know mine, I have a feeling we might have two different conclusions.

Already proven to myself that 10bit is superior to 8 bit when shooting at 1080.

Why not use a GH4 where I can use the same camera to have less variables? I don’t own one anymore and I did those tests when I had it and I couldn’t see any difference.

I have done all the blue sky gradient tests you can watch in the other video, pretty much my conclusion from the other video was yes I can see a difference between 8bit and 10bit when the a7rii was shot at 1080 but at 4k I could not see a difference.

I not sure if my monitor is a true 10 bit, I am also using a GTX card which I am not even sure if it now outputs 10 bit as well, so I might not even be seeing the correct results.
Thanks to Erik Stenbakken

5 comments

  • dude, you are an AMATEUR. you don’t know anything about coloring, you don’t know about bit depth, you don’t have any training about any of them, stop making stupid waste on internet.

  • Hi Dave- Matt’s demo was really interesting – totally agree with Matt on most of his test – especially the codec and camera types being part or solution to a lot of the problems
    but – obviously he doesn’t like reading advice on forums or facebook groups – I think that if you dust off all the cobwebs or regurgatated information there is commonality in the problems that normal shooters and future colorists have with 8-bit cameras.
    Most of that advice/problems discussed are with slog-3 and the advice is to not recommend slog-3 on an 8-bit camera.

    Slog-3 is a popular choice because the mid to highlights ( above middle grey ) is a linear curve and reacts similar to rec709 in grading. Long story short – it is easier to grade than slog-2, however-
    The part that creates all the problems is that the bit information that gets to the shadows has very little data per stop. So reading the forums might not be as good as testing but what it does tell you is – what and where people are having problems and what to test.

    I have to admit though that a lot of problems with noise and images breaking is due to rasing gain – not lowering. In Matt’s tests – he is mainly pushing down on the grade. Pulling up under exposed in slog-3 in 8-bit is near impossible on some cameras – v-log on GH4 – same thing – in fact the curve for v-log and s-log 3 as to middle grey and white points are almost identical and the same problems arise for each.

  • Dave, I always love your videos! It’s too bad there are so many people out there who are ready to put others down for sharing their thoughts and opinions. I think you’ve made an extremely good case for 8 bit vs 10 bit. If others disagree, they need to make a video to SHOW why they believe you’re wrong; otherwise it’s just words. Keep up the excellent work, Dave!

  • especially the green-mangenta artifacts in your video at 19:29-22:30 look very intersting to me. i found similar issues while evaluating GH4 v-log. but what’s so interesting in this particular demonstration, is the fact, that it appears only in 4:2:2 high qulaity intermediates, not int in the 4:2:0 footage straight from the camera. this reminded my of a horrible bug in resolves color science discussed recently: https://forum.blackmagicdesign.com/viewtopic.php?f=21&t=52038

    but concerning the main topic of your video i’m not really satisfied by your demonstrations. 8bit vs. 10bit comparisons, that do not reflect the use of dithering, will IMHO never grasp reality!