I want to see the benefits of 10-bit footage in software like Premiere Pro. To me it behaves exactly like 8-bit video. How can I "see" the benefits?
3 Answers
REC 709, the current standard for HD delivery (and YouTube), really only requires 5 stops of dynamic range (since the legal range is 16..235 out of a full scale 8-bit encoding 0..255). Now, in much the same way that a 2x4 piece of lumber is milled from a starting point of 2"x4" down to 1.5"x3.5", an 8-bit encoding gives one the freedom to shave down the highest and lowest bits (where noise lives in the shadows and where clipping occurs in the highlights) and still have one more bit of slop to adjust brightness. Which is not very much room to be creative.
A 10-bit source file allows for two bits more dynamic range that can be used to pull down highlights or lift shadows. That may not sound like very much, but against a delivery format that encodes a total of 5 bits, it's quite a lot!
Now, it's true that with careful processing (especially noise reduction), one can make more use of the top and bottom bits of dynamic range, which means even more opportunity to stretch the dynamic range of the source material to conform to the intended tonal range of the output format.
Another place where 10 bits becomes essential is in the world of HDR video. YouTube now supports HDR playback on devices that support HDR10 (and Dolby PQ). HDR 10 really does can use all 10 bits of dynamic range, but as Ansel Adams teaches in his books on The Zone System (The Camera, The Negative, The Print), the art of photography (drawing with light) is not merely creating a 1:1 mapping between acquisition and exposition, but in how one chooses to distribute the tonal range to best effect. Starting with a full tonal range in the source makes it that much easier to push things around to get them where you want them to be on output.
answered May 8, 2018 at 17:36
Only when an image Is pastel like or has low gamma where the depth of colour is stretched to see the visual resolution steps . These quantum affects of eight bit colour that will be four times better (smaller) with 10 bit. However DAC linearity must be equally better which are often not so great if viewing such content. In high contrast imaging , it would be hard to notice the improvement , but in shades of grey quite an improvement.
I use DPT.exe a free tool to calibrate my desktop and video and these weaknesses are obvious there. But the LowFi listener probably won’t recognize the difference.
Most of the time your video is going to be viewed in 8-bit colour, often with lower chroma resolution. If it's on the web or broadcast it's also going to be compressed to a varying amount. So it's a good question why you would bother with 10 or more bits per-channel.
The main advantage of 10+ bit footage is before the final delivery. When you are grading or adjusting the colour having 10+ bits is massively useful, because you have a lot more information available to work with. For every level in 8 bit footage there are 4 more in-between levels in 10 bit footage per channel, meaning 64 times as many available colours. So if you're pulling down the highlights of an exterior shot that's a bit overexposed, you'll see details in clouds, instead of your whites just going uniformly grey, or if you're changing the colour balance it won't start to look solarised.
10 bit footage also has four times as many steps available in a gradient. Say you have a gradient that goes from black to white across the width of a frame (1920 pixels), in 8 bit footage there are only 256 possible colours the pixels can be: [0,0,0] to [255,255,255]. That means each of those colours will have to be spread out across a few pixels: 1920/256 = 7.5 pixels wide (on average). So you'll clearly see the bands. A 10 bit image with the same gradient will have bands less than 2 pixels wide. Go to 4k and the problem is magnified by 2, and for more subtle gradients the problem is even worse. Even for display purposes 8-bits starts to show a lot of banding in 4k.
Even if you're delivering in 8-bit having the extra information available when processing the video will lead to visibly better results.
