4th cone and color vision and AMV's.

[HotaruT]

Has anyone come across to this one?
https://www.linkedin.com/pulse/25-peopl ... ana-derval

Now don't go cheating on the test. ^^

I'm just a bit curious since i have had some reviews saying that there is no difference in quality and color space in hi10bit and standard 8bit videos. And others say there is a big difference. (i have 4 cones apparently)

What say you? How many colors you see and have you noticed any difference in lower and higher quality videos? Have you done any testing with, let's say, 1920x1080 material that is near/exceeds 12mb/sec bitrates?

[BasharOfTheAges]

37, but the uniform banding kinda makes it seem like it's leading you to a higher answer.

[HotaruT]

37, but the uniform banding kinda makes it seem like it's leading you to a higher answer.

That's the cheat prevention so people don't just count bars. 37 is a very good result. And no-one go blaming monitor or tv's. They have millions of colors.

[Qyot27]

That's not what the 8-bit/10-bit quality divide in encoding is about, though. 8-bit *still* contains millions of colors. It has to do with the way the encoder works (how the color curve gets subjected to quantization), not so much the way humans see - really, the biggest difference you're likely to notice is that 10-bit preserves color gradients better than 8-bit does, since it has that higher precision in the curve. This leads to a reduction in banding, so the gradients will appear smoother...at least if it was processed correctly beforehand. Someone should torture-test it with Ergo Proxy.

But this is not to say that you can't find examples of 8-bit videos where the color gradients look fine; 10-bit (or 12-bit, 14-bit, or full-on 16-bit*) is not a magic bullet if the source was already mangled. There's cases where any additional compression in 8-bit will only make banding worse or reintroduce it, and using 10-bit allows that circumstance to be avoided or strongly mitigated. Where higher bit depths really matter is when capturing or processing video in the mastering phase, since the greater dynamic range allows for more precise adjustments on the professional side of the equation that consumers and casual viewers typically never see or even get close to.

For the record, I also got 37. And the article's 'predictions' about your wardrobe based on cone count are total nonsense.

[Qyot27]

*x264 users can't use these higher depths, though; you need the JM reference software to encode 12-bit or 14-bit H.264 sequences, and it takes forever. x265 is beginning to implement 12-bit mode, so that won't be as painful as needing the HM reference to do it, and 16-bit is do-able with some lossless formats like FFV1, but those aren't for distribution encodes

[HotaruT]

*x264 users can't use these higher depths, though; you need the JM reference software to encode 12-bit or 14-bit H.264 sequences, and it takes forever. x265 is beginning to implement 12-bit mode, so that won't be as painful as needing the HM reference to do it, and 16-bit is do-able with some lossless formats like FFV1, but those aren't for distribution encodes

You are spot on in technicality but how about commenting on the link and the fact that only 25% of people actually can see the full color range? To those people, the banding is much more visible then those who can't. And yes, hi10bit versus 8bit "widens" the color spectrum and it helps to compensate banding. It won't correct it but it adds smoother transitions to banding. Can't see it? Test explains why.

And there is another added bonus, this is not official but after lots of testing and tweaking, hi10bit needs only 2/3 of the bitrate that 8bit does and with newer codecs, there is partial hardware acceleration so it won't kill an older PC's CPU that easily. Before codec, cpu was 50-60% at worst, now it barely hit's 10-15% so Nvidia acceleration does something even it's not officially supported. Tests in PrecisionX support this because it takes GPU to much more use when in software mode it just idles. Google is your friend if you wanna find these newer codecs.

And let's get back to the main point of this post.
How many colors did you see in the picture and have you noticed anything like what they say in that link?

P.S. I sometimes get to use material that has 20MB bitrates and over. That's like 1GB for a minute of video alone in h.264 mp4 format but no-one would want to download AMV if it exceeds 1GB in size now would they? I study all the time to get that original high quality as small size i can and this is where hi10bit has shown it's good points in compressibility and quality. I never upscale, i only try to preserve as much of the original quality in a size that people actually want to download.

[Qyot27]

I did report that:

For the record, I also got 37. And the article's 'predictions' about your wardrobe based on cone count are total nonsense.

It's definitely not 25% of people. The genetics of color vision just don't hold up on that point. The article was written by a marketing firm, and doesn't quote any peer-reviewed scientific journals that are the source of its claims. In other words, and as some of the commenters on the article pointed out, I smell clickbait.

Actually finding the few human tetrachromats that exist would require genetic testing and a full battery of things similar to the Ishihara test with recalibrated color channels to find them by exclusion, not a row of paint swatches. Human vision isn't RGB with a possible tetrachromat being RGBY.