It’s far too easy to read opinions online rather than testing different camera specs for yourself. So here’s someone’s opinion online to go with.
It’s a fun test for you to look at a few different clips and see for yourself. You could cheat and look at the file names and have a decent idea, but don’t do it! If you double-click the text below the thumbnails in the “Clips” panel. Try some extreme color changes on them as well and see how the different footage handles. “D” on the edit page disables the clip so you can see the clip below, or, even simpler, go to the color page and hit the “unmix” button near the play button and then just click through the clips (cmd+F goes full screen for a better view).
Just download the media above, double click to unzip, then click the .drp file to open the project directly in Resolve. Right click the “Master” bin in the media pool and “Relink Clips For Selected Bin” if needed. If you have a snappy computer, first, set timeline resolution to 4k for best comparison.
Incidentally, much of this is more easily viewed in Fusion page using , and . memories and / to wipe with ⌘⌥ shortcut moving the wipe. Use the CST node inside Fusion to get your conversions done since there’s no color effects at the Fusion stage.
The first frame looks nearly identical (I-frame) on the temporally-compressed codecs
You CAN see a difference between Atomos and internal recording 4:2:0 vs 4:2:2 (zoom in 1:1 on the red and yellow squares and look at the ‘R’ channel alone) but what about when scaled 4K to HD?
The macroblocking of the compressor is worse than the chroma subsampling and the Atomos gives a much better result
The gamut transform gets nasty fast.
8 bit codecs can work in log if you nail exposure and don’t try mapping from a wide gamut to a narrow one on Sony cameras. I’ll design a LUT that takes Sony colors from S-log2 Rec709 gamut and normalizes only the gamma if you’d like.
It’s probably worth the Atomos if you do green screen keying or deliver in 4k. The reduced macro-blocking from the compressor is the biggest advantage, but the codec is also more edit-friendly, will media manage well, and the screen for production use is super nice.
If your exposure technique is not good, or you work in an unstable environment, raw formats can give you quite a bit more flexibitility. Other than that, probably not worth it until you’re really shooting a cinematic project where the small difference in image quality is critical.