Introduction
The idea of CTU-tree (cutree mode) is as follows:
a block is more important if it is referenced by many other blocks, i.e. has many ancestors in the graph (or in the tree).
cutree makes sense if the lookahead is enabled, otherwise we can’t know ahead how much a given block is referenced in next frames. Strongly referenced blocks are quantized finer, and poorly reference blocks are quantized coarsely.
For more details i recommend to read the paper the paper “Evaluating the CU-tree algorithm in an HEVC encoder” by Vladimir Grozman.
For evaluation of cutree mode we apply x265 (in ABR mode) on two 4K streams – Sintel (3840×1744, 24 fps) and TOS (3840×1712, 24 fps), with the lookahead buffer size 5 frames, with two different bitrates 30Mbps and 20Mbps.
Notice that x265 does not allow lookahead buffer smaller or equal to the number of consecutive B-frames. Therefore, the minimal lookahead length is equal to 3, since up to 2 successive B-frames are utilized in our evaluation. If the lookahead is disabled (‘–rc-lookahead 0’)
Sintel:
TOS
X265 Configuration
All streams are coded with the following fixed parameters:
max CU size = 64, min CU size = 8
max TU size = 32, TU depth = 1 (for both inter and intra)
ME algo = hex, ME range (–merange) = 57, subpel mode (–subme) = 2 (one half-pel iteration in 4 directions, one quarter-pel iteration in 4 directions) ,
maximal number of candidate MVs in merge/skip mode = 2 (–merge 2)
scenecut = 0 (no scene cut detection)
Gop size = 30 frames
bframes = 2 (IPBBPBB..), badapt = yes (B frame can be replaced with P frame), no b-pyramid (i.e. B frames are not used for reference)
refs = 2 (two reference frames)
signhide = yes
tmvp = yes (temporal motion vector prediction enabled, but with ‘–merge 2’ it impact is expected to be ngeligible)
strong-intra-smoothing = yes (since 32×32 TU used)
deblock, sao = yes
Results (PSNR)
Notes:
- In case of cutree=off, the lookahead buffer length does not impact on quality
- PSNR is not a good metric to compare the cutree mode, it’s better to use more HVS-correlated metric like VMAF.
Sintel, cutree = OFF
| Lookahead | Target Bitrate 30 Mbps | Target Bitrate 20 Mbps |
| 5 frames |
PSNR: 44.574 actual bitrate: 37.8 Mbps
|
PSNR: 43 actual bitrate: 25.7 Mbps |
Sintel, cutree = ON
| Lookahead | Target Bitrate 30 Mbps | Target Bitrate 20 Mbps |
| 5 frames |
PSNR: 44.894 actual bitrate: 39.3 Mbps
|
PSNR: 43.367
actual bitrate: 26.7 Mbps |
TOS , cutree = OFF
| Lookahead | Target Bitrate 30 Mbps | Target Bitrate 20 Mbps |
| 5 frames |
PSNR: 42.851 actual bitrate: 25.4 Mbps
|
PSNR: 42.333 actual bitrate: 17.6 Mbps |
TOS , cutree = ON
| Lookahead | Target Bitrate 30 Mbps | Target Bitrate 20 Mbps |
| 5 frames |
PSNR: 42.644 actual bitrate: 22.3 Mbps
|
PSNR: 42.177 actual bitrate: 15.5 Mbps |
Conclusion:
Rate control reveals strong overshoots for Sintel and undershoots for TOS. PSNR metric is not adequate to evaluate cutree mode.
Appendix
By default cu-tree is enabled.
cu-tree is automatically disabled if lookahead is set to 0, because cu-tree uses the lookahead pass. To disable the lookahead you need set '--rc-lookahead 0'
Example:
x265 --input test1080p.yuv --input-res 1920x1080 --fps 60 --b-adapt 0 -b 0 --ref 1 --no-scenecut --keyint 10 --min-keyint 10 --repeat-headers --bitrate 3000 --frames 60 --rc-lookahead 0 -o test.h265
…
x265 [info]: Lookahead / bframes / badapt : 0 / 0 / 0
x265 [info]: b-pyramid / weightp / weightb : 0 / 1 / 0
x265 [info]: References / ref-limit cu / depth : 1 / off / on
x265 [info]: AQ: mode / str / qg-size / cu-tree : 2 / 1.0 / 32 / 0
…
23+ years’ programming and theoretical experience in the computer science fields such as video compression, media streaming and artificial intelligence (co-author of several papers and patents).
the author is looking for new job, my resume
An impressive share, I just given this onto a colleague who was doing a little analysis on this. And he in fact bought me breakfast because I found it for him.. smile. So let me reword that: Thnx for the treat! But yeah Thnkx for spending the time to discuss this, I feel strongly about it and love reading more on this topic. If possible, as you become expertise, would you mind updating your blog with more details? It is highly helpful for me. Big thumb up for this blog post!
Dead pent content, thanks for entropy. “He who establishes his argument by noise and command shows that his reason is weak.” by Michel de Montaigne.
Oh my goodness! an incredible article dude. Thank you However I am experiencing problem with ur rss . Don’t know why Unable to subscribe to it. Is there anybody getting an identical rss drawback? Anybody who knows kindly respond. Thnkx
Howdy very nice web site!! Man .. Excellent .. Amazing .. I’ll bookmark your site and take the feeds additionally?KI am happy to search out numerous helpful information here within the submit, we want develop extra techniques in this regard, thank you for sharing. . . . . .
You are my aspiration, I possess few blogs and often run out from to brand : (.
I would like to thank you for the efforts youve put in penning this site. I really hope to view the same high-grade content by you later on as well. In fact, your creative writing abilities has inspired me to get my own website now 😉
Dead written subject matter, Really enjoyed looking at.
I went over this web site and I think you have a lot of wonderful info, saved to favorites (:.
I’ve been surfing online greater than three hours nowadays, yet I by no means discovered any interesting article like yours. It?¦s lovely price sufficient for me. In my view, if all site owners and bloggers made just right content as you did, the web can be much more helpful than ever before.
Hey, you used to write fantastic, but the last several posts have been kinda boring… I miss your tremendous writings. Past several posts are just a bit out of track! come on!
After research a number of of the blog posts in your web site now, and I truly like your manner of blogging. I bookmarked it to my bookmark web site record and will be checking back soon. Pls try my website as properly and let me know what you think.
I like this post, enjoyed this one appreciate it for posting.
I believe you have noted some very interesting details , thankyou for the post.
I’ve been absent for a while, but now I remember why I used to love this website. Thanks , I¦ll try and check back more frequently. How frequently you update your site?
I have been exploring for a little for any high quality articles or blog posts on this kind of space . Exploring in Yahoo I ultimately stumbled upon this web site. Reading this information So i am satisfied to express that I’ve an incredibly just right uncanny feeling I discovered just what I needed. I most certainly will make certain to don’t omit this site and give it a look regularly.
You got a very great website, Gladiola I observed it through yahoo.
obviously like your web-site but you have to test the spelling on several of your posts. A number of them are rife with spelling issues and I to find it very troublesome to inform the truth then again I will certainly come back again.
Throughout the grand scheme of things you receive a B+ with regard to hard work. Where you actually misplaced me ended up being on all the specifics. You know, people say, the devil is in the details… And that couldn’t be more accurate here. Having said that, allow me reveal to you what did work. The writing can be very engaging which is possibly the reason why I am taking an effort to comment. I do not really make it a regular habit of doing that. Next, even though I can notice a leaps in logic you make, I am definitely not confident of just how you appear to unite the ideas that help to make the actual conclusion. For the moment I will subscribe to your issue however wish in the near future you actually link the facts much better.
most of posts require some background, otherwise they look chaotic.
i recommend you to read the paper “Evaluating the CU-tree algorithm in an HEVC encoder” by VLADIMIR GROZMAN
I got what you intend, regards for putting up.Woh I am happy to find this website through google.
Can I simply say what a aid to search out somebody who actually knows what theyre talking about on the internet. You undoubtedly know find out how to bring a problem to light and make it important. More people need to read this and understand this aspect of the story. I cant believe youre no more common since you positively have the gift.