10 bit vs 12 bit video opinions

[MiniSiets]

Now that HandBrake 1.1.1 is out which added 10 bit and 12 bit support for h.265, I got curious on seeing what the differences are in quality.

I did a comparison of the same 30-second preview clip in 8, 10, and 12 bit formats and here's what I noticed:

I definitely noted that dithering to different shades of color was much less noticeable with the jump from 8 bit to 10 bit, but I wasn't really able to discern any difference in quality from 10 to 12. So it seems to me the best choice is to encode in 10 bit.

I'm curious what are others' thoughts on 10 bit versus 12 bit? Do you notice the difference, and do you feel encoding in 12 bit is worth it?

[mduell]

Utterly worthless and stupid when you have 8 bit content, an 8 bit video pipeline in HB, and an 8 bit display.

[MiniSiets]

I would disagree, if only because I notice much smoother handling of dithering. I did a blind test on myself too just to make sure I wasn't tricking myself with placebo. I was able to pick out the 8 bit and 10 bit videos without knowing which one was which. Yeah it may not be true 10 bit color, but it still looks better against an 8 bit encode.

[KenithO]

Hi MiniSiets,

I am also curious if there are any noticeable differences between H.265 8-bit and 10-bit using an 8-bit Source.

Googled using keywords h 265 8 bit vs 10 bit got ~42,600,000 results (0.74 seconds) Redid using "..." but got nothing.

To date the best I have found is "QUICK COMPARE: AVC vs. HEVC, 8-bit vs. 10-bit Video Encoding" by Archimago http://archimago.blogspot.com/2016/12/q ... vs-10.html

Conclusions
"1. Image quality comparing AVC to HEVC - it is obviously a "win" for HEVC overall. Macroblocks are less obvious in the HEVC image and details like in the branches of the tree have been much better retained. Obviously, fine details such as in the subtleties of the clouds have been lost due to the very low bitrate... However, I trust that subjectively the deterioration is less objectionable with the HEVC encoder.

2. Comparing 8-bits to 10-bits, we see an improvement with deeper bitdepth. Although the original AVC recording was less than ideal in terms of the blue sky gradients, I think we can see that the AVC 10-bit encoding had less "blockiness" especially in the upper mid portion of the deeper blue sky. As for the HEVC recording, again, the extra bitdepth seems to help provide a smoother shading in the sky despite the loss in general resolution with such a low bitrate. Also, there appears to be better retention of details in the cloud (like the right "tip" of that cloud and the shades of white/gray)."

Was unable to copy/paste file sizes but they are given in post. What surprised me was the 10-bit was smaller than 8-bit!

Recommend others read this post and check if agree with Archimago.

[Ritsuka]

12bit is an overkill, and hardware decoders doesn't support at all.

[Deleted User 13735]

What surprised me was the 10-bit was smaller than 8-bit!

Encoding 8 bit source in a 10 bit wrapper just adds air; no dithering takes place because there is no bit resampling, just the same data in a bigger box. Your reported file sizes prove that beyond a shadow of a doubt, and so would a histogram comparison.The differences you see are due to the Pygmalion Effect, because they cannot be measured or quantified.

The math is so middle school:
1024^3 / 256^3 = 64 times larger for the true 10 bit encode!
And 12 bit? 4096 times larger files than 8 bit.

[Deleted User 13735]

Now, downsampling from 10 bit source to 8 bit delivery, Handbrake shines in reducing banding, moire, dithering and quantization noise over other popular resampling algorithms. I've proven that, and I have always used a 10 bit YUV intermediate to Handbrake for 10 bit and uncompressed 8 bit source (see updated results below).

[Carlo Macchiavello]

12 is a waste of time, space, cpu to delivery.
Is excellent for shooting to have all room to work on vfx and color grading.

a simple answer is
from a generic source 8 bit, with correct structure of picture, you not have great improvment to encode at 10bit, with the exception of large smoothed gradient on video, classic film not have it.
from a source 10bit the delivery at 10 bit is good if...

- do you have a real 10 bit player device?
- do you have a real 10 monitor or video projection system to see real 10bit output?
- most of tv are 8+2 bit, like dinamic contrast, a waste of time and dinamic gamma.
- do have video card/cable/system/monitor correctly working for 10bit, then encode to 10bit, or is a waste of time.

better to encode to high level like 5 to avoid banding and other defect of 8bit encoding then use muscle to encode at 10 bit and see decreasing you color depth random from different playing device

[KenithO]

Carlo Macchiavello,

Was very interested in your comments "...do you have a real 10 bit player..."

Do you know a website that covers which 4K products have real 10 bit hardware?

Did find "4k HDR – 8 bit, 10 bit, 12 bit,..." https://dgit.com/4k-hdr-guide-45905/ but does not mention any websites.

[Deleted User 13735]

Carlo Macchiavello,

Was very interested in your comments "...do you have a real 10 bit player..."

Do you know a website that covers which 4K products have real 10 bit hardware?

Did find "4k HDR – 8 bit, 10 bit, 12 bit,..." https://dgit.com/4k-hdr-guide-45905/ but does not mention any websites.

Kenneth, those questions become preemptory only after you have met the source requirements -- acquired 10 bit, not lipstick on a pig.

The only time a 10 bit render is indicated from 8 bit source is with a YUV digital intermediate from 4:4:4 subsampled RGB source. I know this is new, but I've actually tested this to death.

Then you can explore hardware player options, currently available over HDMI 2.0 Bluray, Fire stick, etc.

[KenithO]

Carlo Macchiavello,

Guess was not clear, I would like to upgrade to a 4K Blu-ray DVD player that also supports 10 bit video. Did some searching but most reviews do not mention this. Even the recent "The Best 4K Blu-ray Player" https://thewirecutter.com/reviews/best- ... ay-player/ does not.

So checked Sony Full Specifications and Features for UBP-X700 https://www.sony.com/electronics/blu-ra ... ifications no mention.

Then checked other Sony UBP-X700 reviews:
https://www.whathifi.com/sony/ubp-x700/review
https://www.trustedreviews.com/reviews/sony-ubp-x700

Seems may not because review https://hometheaterreview.com/sony-ubp- ... -reviewed/ mentions "...You don't get the advanced setup options found in a higher-end player like the Oppo UDP-203 (like the ability to choose YCbCr 4:2:0 and 10- or 12-bit color"

Wish there was a website that would make this easier!

Ken

[KenithO]

musicvid,

"The only time a 10 bit render is indicated from 8 bit source is with a YUV digital intermediate from 4:4:4 subsampled RGB source" Thanks for the tip!

"I know this is new, but I've actually tested this to death" Have you posted your tests anywhere?

Question: As previously mentioned I am shopping for a 4K Blu-ray DVD player that supports 10 bit and possibly 12 bit video. Have been very disappointed with the reviews I have found to date.

Do you have any recommendations concerning 4K Blu-ray DVD player reviews (Any Language) that covers things like 10 bit, 12 bit video etc?

Thanks again for your comments!

[JohnAStebbins]

If you run your own tests (as MiniSiets has done), you can see the difference in both quality and file size of 10 bit vs. 8 bit encoded content even when starting with an 8 bit source. The visual differences are most pronounced in flat-ish backgrounds and darker areas where an 8 bit encode will typically show some haloing or banding. A 10 bit encode will show less haloing or banding. The reason for this it tied to how the encoder transforms and quantizes the data. See this thread:

viewtopic.php?f=6&t=36432&p=171379&hili ... cy#p171354

Going from 10 bit to 12 bit will not improve things much. I doubt anyone would be able to see any difference in quality. There may be some small advantage in file size. The only reason I can think of to use 12 bit would be for multi-generational encoding where you are trying to preserve as much quality as possible because you know you'll be losing more in a subsequent encode pass.

[KenithO]

JohnAStebbins,

Thanks for the comments and link!

"If you run your own tests (as MiniSiets has done), you can see the difference in both quality and file size of 10 bit vs. 8 bit encoded content even when starting with an 8 bit source."

Have you posted any tests of 10 bit vs 8 bit?

[KenithO]

Just read "8, 12, 14 vs 16-Bit Depth: What Do You Really Need?! by GREG BENZ https://petapixel.com/2018/09/19/8-12-1 ... ally-need/

Was especially interested in "With a clean gradient (ie, worst case conditions), I can personally detect banding in a 9-bit gradient (which is 2,048 shades of gray) on both my 2018 MacBook Pro Retina display and my 10-bit Eizo monitor....An 8-bit gradient is relatively easy to see when looking for it, though I might still potentially miss it if I weren’t paying attention. So, for my purposes, a 10-bit gradient is visually identical to 14-bits or more."

"It is also important to note that you are likely to run into false “banding” when viewing images at less than 67% zoom." https://gregbenzphotography.com/photogr ... -photoshop

Any comments concerning this article?

[Deleted User 13735]

So, for my purposes, a 10-bit gradient is visually identical to 14-bits or more."

While I agree in principle with the article, and I have advocated 10 bits for YUV digital intermediates since I ran similar tests in 2011, I find that statement a bit of a stretch. Bit depth has more to do than just reduce banding.

However, now that Magic YUV has made its way to Handbrake, repeating those same tests last week tells me that Magic YUV 8 bit 422 should be every bit the equal of 8 bit uncompressed 444 both with respect to banding and chroma subsampling, for single-generation intermediate purposes. Surprised me too.

I hope someone will take the time to run their own tests and share their impressions.

[MiniSiets]

What surprised me was the 10-bit was smaller than 8-bit!

Encoding 8 bit source in a 10 bit wrapper just adds air; no dithering takes place because there is no bit resampling, just the same data in a bigger box. Your reported file sizes prove that beyond a shadow of a doubt, and so would a histogram comparison.The differences you see are due to the Pygmalion Effect, because they cannot be measured or quantified.

What I noticed from when I did my tests was that while encoding to 10 bit from an 8 bit source obviously didn't improve the quality of gradient shading from the original source, it did however prevent it from degrading further as seen in the 8 bit to 8 bit encode. The transitioning to different shades of color were much closer to 1:1 identical to the original source instead of getting worse. As you said, it's just the same data in a bigger box, so it's not adding any data to it, but for whatever reason it does seem to do a better job at maintaining the quality of the original source than a straight 8 bit to 8 bit encode.

So I'll be encoding everything in h.265 10-bit from now on.

[BradleyS]

10-bit being smaller than 8-bit can be measured and quantified: http://x264.nl/x264/10bit_02-ateme-why_ ... dwidth.pdf (PDF).

TL;DR: More accurate compression leads to fewer truncation errors, saving space.

[mduell]

"Strangely" no one makes this argument about higher bit depths...

[BradleyS]

Diminishing returns.