@marcinzm, u r totally wrong!
1, human eye can distinguish billions color but human tech can till now only capture maximal ca. 80% of that area.
2, capture is one thing and transport/save is another thing. Thus the subsampling is developed, goal: minimizing the techical lost of color details and maximizing the mobility of data, e.g. your professional camcorder keeps 10bit422 subsampling so that this source material can bring high-quality endproduct for your TV or something like that.
3, the endproduct refers to industrial standard such as Rec709/2020, because A: it must be a standard/definition there for hardware, B: technical limitations or production costs.
4, in the subsampling illustrations you can see only decreases occur in color sampling because human eye acts somewhat "duller" compared to light. By the way, the last sampling means 420-402-420-402...
By the way #2 (as recommended by JB, not JFK, John-Dennis),
when you watch TV you know it's a TV, but wehn you open a window and look out, you know it's the nature. That's the ability of your eyes.
The new type HDR10 video is currently the maximal approach to the nature. And its color sampling is still 420 but the light goes to till 10,000CD/m².
By the way #3,
there's a test that 6 people sit in an absolut dark room, during the time a 1/3000sec light swept the room. 3 of them affirmed they "saw" the light; 1 did feel the light and 2 didn't know what happened.
I contributed a small post here, take a sight if you have interest.
@RogerS I cleared up my doubts regarding to the answers that I should export video to 8-bit 4:2:0. The reason to do it so is that human eye is able to see no more than 10 milions colors which even doesn't cover the full 8-bit range (16.7 milion of colors). That's why humans doesn't see any significant changes between 8-bit and 10-bit video. Can you add it to it something more? Do you agree with my explanation?
The fact is that 95%+ of all video delivery is still 8 bit 4:2:0. What good is your high bit masterpiece if no one can view it?
human eye can distinguish billions color
Where did you get that figure from? Nothing I can find out there on the internet gets that high. Some say one million, others 8 - 10 million perceived colours. Apart from "tetrachromats", people who possess apparent superhuman vision according to one researcher. Obviously it varies from person to person and one source suggested 300,000 was a more realistic figure.
I agree with the explanation. The extra data is useful for editing not viewing.
@RogerS Ok. I am happy that my information was proved by you. Thank you.
But please explain it to me the following thing. While moving/changing brightness bars, contrast bars etc while editing 10bits, the editing video which human will see just after edit is much natural, fit to real and accurate than if we would edit 8 bits. Am I right? But in both cases we will still see 10 milion colors as we wrote ealier.
The reason to shoot 10-bit is that there aren't enough steps in 8-bit to make dramatic gamma and color adjustments. You only have 256 steps in the 3 color channels and subtle tones the human eye is sensitive to, like blue skies and skin, will start to show artifacts from discontinuous jumps in luminance (and color).
Don't take my word for it, shoot a flat profile in 8 bit with lots of dynamic range compressed into it with a log curve like Sony S-log 3. Then add a LUT or correction curve to bring it back to Rec 709 (so it has normal contrast to the eye). You'll likely see dancing bands in the sky and potentially event odd bits of purple and magenta amidst the blue. 10-bit gives you more data precision to avoid these artifacts. After you make the corrections using the 10-bit file in a high-bit spaces (32 bit in Vegas) you can output an 8-bit file which is more than enough for human vision.
Here's an example I made a while ago with my Sony showing the limits of 8-bit files. Look at the sky and look at the scopes.
4:4:4 vs 4:2:0 etc. gives greater chroma precision which can be helpful when it comes to the edges of objects, for example getting a clean key if doing composites, etc.
human eye can distinguish billions color
Where did you get that figure from? Nothing I can find out there on the internet gets that high. Some say one million, others 8 - 10 million perceived colours. Apart from "tetrachromats", people who possess apparent superhuman vision according to one researcher. Obviously it varies from person to person and one source suggested 300,000 was a more realistic figure.
Pls don't disarrange system of knowledge with experience of individual.
I lost count at 500,000. 😏😏
@RogerS You mentioned that 10bit footage is better in editing than 8bit footage, because of the greater steps than exists in 8bits. So in summary, the 8bit output rendered video file, edited previously in 10bit raw footage, will be much better and in better quality than if we would edit 8 bit raw footage and rendered it to 8 bits. Can you agree with me? Does everyone see a significant difference in quality between such two 8 bits output files (1st previously edited in 10bits and 2 previously edited in 8bits)? Is it a much strong quality difference which let viewer tell "WOW. WHAT A HIGH QUALITY VIDEO" when he/she would see 1st output video?
It would be much better quality if the original file attempted to compress a wide dynamic range into an 8-bit file. Personally I shoot 8-bit using Cine 2 on Sony cameras and there aren't any visible artifacts. I sacrifice some dynamic range, but there's frankly enough for what I do. 8-bit is not appropriate for S-log 3 and arguably S-log 2. I gave you an example of how bad it can look.
If you did a nice job with S-log 3 (or with Cine 2 and the dynamic range of the subject fit into that of the profile) they would both look excellent as the final 8-bit file. The main difference with S-log 3 will be more detail in highlights and/or shadows.
I'm a photographer experienced in color correction so I take image quality seriously. You can do excellent work from 8-bit footage if you respect its limitations.
So what are you shooting, what camera are you using and what settings are you using with the camera? Based on that you can choose a 32-bit or 8-bit workflow in Vegas and get excellent results.
So in summary, the 8bit output rendered video file, edited previously in 10bit raw footage, will be much better and in better quality than if we would edit 8 bit raw footage and rendered it to 8 bits. Can you agree with me? Does everyone see a significant difference in quality between such two 8 bits output files (1st previously edited in 10bits and 2 previously edited in 8bits)? Is it a much strong quality difference which let viewer tell "WOW. WHAT A HIGH QUALITY VIDEO" when he/she would see 1st output video?
All good questions, and really good sounding theories, but the only definitive answers you will ever get is if you run your own objective tests in a controlled environment and decide for yourself whether it is worth all your extra time and angst. Remember, "looks better to me" is only true for you, and for the people who love you so much they always agree with you
Remember:
Eagerly awaiting your side by side visual examples. @marcinzm
@marcinzm 10bit vs. 8bit analogy: Cloth cut on table. (Source footage inside editing environment)
Everything 8bit, source, editing, delivery:
Say you were going do a small trim - to cut a 31" square shape out of a 32" square piece of patterned cloth. You have a 32" square table - actually the table is large cube shape. (This would be analogous to 8bit output (finished piece) from 8bit editing source/format (cloth and table).This is going to work well and you can get your clean cut easily. Just set the cloth squarely on the table (no pun) and cut the cloth 1/2" in from the edge of the table.
That would analogous to editing with very little or no fx, just trimming unwanted footage.
10bit source, 10 bit editing, 8 or ten bit delivery.
Now, say you'd want that same 31" square shape, but rotated 45 degrees relative to the table, and you'd want to pick up certain parts of the pattern in the cloth, but still desire the same outcome, you'd need a bigger table (10bits),- without that larger table you'd have some of your cutting run down the side of the cube, though still could be done, but you'd be introducing some potential rough/not straight edges (artifacts). Starting with a larger cloth (10bit source) also gives flexibility. The desired END result would be the same, but your cutting options are cleaner with a larger (10 bit) table and larger cloth.
10-bit avc isn't supported by any hardware decoders so it'll make your edit process a bear if you shoot it. In the ffmetrics testing I've done, 10-bit hevc renders score higher than 8-bit renders, but not by as much as one might expect. But I still shoot 10-bit hevc, set Vegas to edit 32-bit float, and render 10-bit hevc for delivery whenever possible. But I don't lose any sleep delivering 8-bit. High end amd video boards may display 10-bit and higher to a monitor but can only render 8-bit. Intel and Nvidia support 10-bit hevc rendering.
Regarding the need to dither high-res video clips... I would doubt that. That's a low-res audio thing that isn't even appropriate for higher res audio. I would expect any applicability to video would be similarly relegated to it's lower resolutions.