Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
|
23rd November 2008, 22:50 | #1 | Link |
Pinhead
Join Date: Sep 2002
Location: Twin Cities, MN, USA
Posts: 158
|
What is the benefit of trellis 2, subme 8/9, merange?
I've been doing some x264 test encodes to see if I can notice visually the benefit of using a certain feature. For example, the use of psy-rd is quite obvious even to my untrained eye but others are not so clear to me. If any of you with sharp eyes can explain what effect or benefit one should be able to expect from either of these features it may help me find them also.
trellis 2: After some testing of a 10000 frame clip of a Band of Brothers bluray disc, I can't say that I notice any difference at all except a slower encode. I've heard on this forum that in conjunction with psy-rd it's supposed to help retain grain. I've done side by side comparisons of two clips but I can't see it. Any pointers or examples in which it would help? Band of Brothers is very motion intense and grainy so I thought it would be a good test sample. subme 8 or 9: I don't see a difference from subme 7 to these either. Are there certain elements of a frame that I should be looking at? These don't seem to help compression, but numbers wise I do see that they sometimes lower or raise certain quantizers. merange: In what type of footage does a high merange help? Last edited by Guest; 23rd November 2008 at 22:58. Reason: rule 9 |
23rd November 2008, 22:53 | #2 | Link |
RipBot264 author
Join Date: May 2006
Location: Poland
Posts: 7,812
|
I would like to see example where --b-adapt 2 beats old method...
__________________
Windows 7 Image Updater - SkyLake\KabyLake\CoffeLake\Ryzen Threadripper |
23rd November 2008, 23:04 | #3 | Link |
x264 developer
Join Date: Sep 2005
Posts: 8,666
|
Generally, higher settings give progressively less return quality-wise with given speed.
In general, the basic idea is that by combining them, you can eke that last bit out of your encoding. For example, let's say that subme7 -> subme9 increases encoding efficiency by 3%. Well, you probably won't notice that really, but that means you can get the same quality at a 3% lower bitrate. These things also tend to add up: 10 cases of 1% higher quality is 10% higher quality. Note also the higher your bitrate, the less you will probably notice the gain from higher options. Higher merange is generally not that useful--the cases in which it is are very complex and fast, sudden motion. Generally, higher merange is only particularly useful on HD footage. Subme8/9 can be more useful than one would expect with psy-RD because they give psy-RD yet another place to act, and the more it can act throughout the encoding process, theoretically, the better the image quality.
__________________
Follow x264 development progress | akupenguin quotes | x264 git status ffmpeg and x264-related consulting/coding contracts | Doom10 Last edited by Dark Shikari; 23rd November 2008 at 23:07. |
24th November 2008, 00:57 | #5 | Link | |
x264 developer
Join Date: Sep 2005
Posts: 8,666
|
Quote:
"I encoded a clip from a DVD at 4 megabits with x264 and xvid, and the x264 one doesn't look any better! x264 must not be any better than xvid." The problem with this is that at 4 megabits, odds are both of them look basically perfect, so its not like they'd notice anything anyways. |
|
24th November 2008, 02:57 | #6 | Link |
The Crazy Idahoan
Join Date: Jun 2007
Location: Idaho
Posts: 249
|
My biggest mistake when first encoding was using a bitrate that was too low and judging quality by random small segments. I have some pretty bad looking video as a result.
If absolute file size isn't that important (which it doesn't seem to be given that you are willing to shift your bitrate) then I would say that CRF encoding is the way to go. |
23rd November 2008, 23:06 | #7 | Link |
Software Developer
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,248
|
Lots of fast motion, high resolution or both.
__________________
Go to https://standforukraine.com/ to find legitimate Ukrainian Charities 🇺🇦✊ |
24th November 2008, 03:05 | #8 | Link |
x264aholic
Join Date: Jul 2007
Location: New York
Posts: 1,752
|
CRF also gives a more direct result of any settings you changed. Say at CRF 18 (roughly transparent quality) using subme 7 and trellis 0 gives a stream with an average bitrate of ~1500 kbps. If you throw on subme 9 and trellis 2, you'll notice the bitrate will drop down some amount. So that 1500 kbps stream is now only (for arguments sake, don't take this as word for the exact amount it'll drop) 1300 - 1400 kbps.
With 2-pass you can either get higher quality at a bitrate or lower your bitrate to get the same quality, but that's a bit difficult.
__________________
You can't call your encoding speed slow until you start measuring in seconds per frame. |
24th November 2008, 04:42 | #9 | Link |
x264 developer
Join Date: Sep 2004
Posts: 2,392
|
Um, what? With 2-pass you get exactly the bitrate you ask for, and any settings difference shows up only in quality. With CRF, various settings affect bitrate or quality or some mixture thereof; if you see bitrate was reduced, you can't even be sure whether that was an improvement without comparing quality too.
|
24th November 2008, 05:22 | #10 | Link |
x264aholic
Join Date: Jul 2007
Location: New York
Posts: 1,752
|
I meant that you can get higher quality at the same bitrate when using slower settings or, if you reduce the bitrate you can get the same quality at slower settings. Except, it would be difficult to gauge how much you'd have to decrease bitrate to get the same quality as before.
__________________
You can't call your encoding speed slow until you start measuring in seconds per frame. |
24th November 2008, 05:47 | #11 | Link |
x264 developer
Join Date: Sep 2004
Posts: 2,392
|
And likewise it's hard to gauge how much you'd have to increase or decrease the CRF to get the same quality as before.
Repeat after me: CRF is quality-based, but not constant quality. It's good enough that you can generally get away with picking one CRF value that's transparent to you, but it's not constant enough to evaluate other options. Last edited by akupenguin; 24th November 2008 at 05:49. |
Thread Tools | Search this Thread |
Display Modes | |
|
|