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.

 

Go Back   Doom9's Forum > Video Encoding > MPEG-4 AVC / H.264

Reply
 
Thread Tools Search this Thread Display Modes
Old 6th May 2011, 01:59   #1  |  Link
pwnsweet
Registered User
 
Join Date: Nov 2008
Posts: 100
Should I do 720p or 1080p at these settings?

Hi,

Below are the x264 settings I'm planning on using on 2.4:1 feature films that average about 2 hours in length. I'm talking about your average movie with a few fast action scenes such as Blood Diamond or Quantum of Solace. These settings result in a roughly 10-12hr encode time on my computer for these films and I'm not prepared to spendtoo much longer than this amount of time to encode.

My question is, should these settings be sufficient for these films to not display any significant visual artificating at 1920x800 (after cropping) or should I lower the resolution?

I've done an encode of Blood Diamond at 1920x800 using these settings and personally can't see any artifacts during a very quick runthrough of the film (I skipped ahead to some dark scenes and action scenes). Even though I can't see any artifacting on this film, I'm concerned about my success with future encodes of other films because I've found that it seems to be generally accepted on these forums to do 1280 at ~4.2mbps, not 1920. Is this because advocates of this philosophy are using worse settings than what I'm using, resulting in worse picture quality at 1920 than what I'm getting or do these settings at 1920 generally always produce significant artifacting and either I just can't see them, don't know where to look or both? If this is the case, what type of scenes are the best scenes to see artifacting?


Anyway, here are the settings I used on Blood Diamond at 1920x800 using x264's film preset:

cabac=1 / ref=4 / deblock=1:-1:-1 / analyse=0x3:0x113 / me=umh / subme=10 / psy=1 / psy_rd=1.00:0.15 / mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-3 / threads=3 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / constrained_intra=0 / bframes=5 / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=40 / rc=2pass / mbtree=1 / bitrate=4110 / ratetol=1.0 / qcomp=0.60 / qpmin=10 / qpmax=51 / qpstep=4 / cplxblur=20.0 / qblur=0.5 / ip_ratio=1.40 / aq=1:1.00

Last edited by pwnsweet; 6th May 2011 at 03:48.
pwnsweet is offline   Reply With Quote
Old 6th May 2011, 04:25   #2  |  Link
Blue_MiSfit
Derek Prestegard IRL
 
Blue_MiSfit's Avatar
 
Join Date: Nov 2003
Location: Los Angeles
Posts: 5,988
Quote:
I've found that it seems to be generally accepted on these forums to do 1280 at ~4.2mbps, not 1920
Who told you that? Thinking that any bitrate is always appropriate to a given resolution is fairly silly. Source complexity is extremely important. In other words, for very simple content, 4.2mbps might be enough bitrate to make very nice looking 1080p. For other content, this will look pretty terrible (though not utterly awful, as you might think).

More importantly, why do you care about hitting a specific bitrate (aka using 2 pass vbr)? CRF mode lets you do two things

1) Do the whole encode in one pass, saving a lot of time (which you can spend on using slower settings if you want)

2) Giving you fairly consistent visual quality. Note that CRF is not really "constant quality", as this would be a very difficult thing to achieve properly. However, it's very close in my opinion.

Start at CRF values of about 20. Do the encode, and if the result doesn't look good enough, lower the CRF by a value of 1. Keep doing this until it's good enough, or the final average bitrate is too high. Once you get a sense of things, you will probably never use 2 pass again.

You might think "but 2 passes means better quality" - not so with CRF. In fact, it's been proven that given a crf encode that comes out to xxxx kbps, a 2 pass encode targeting xxxx kbps will look identical. They're not MATHEMATICALLY identical, but very very close.

Also, next time you post settings, post the x264.exe CLI argments you used. Nobody really wants to wade through that It looks like you're not using one of the x264 speed presets. Why is this? You're probably not going to do any better by manually tweaking things.

Cheers,
Derek
__________________
These are all my personal statements, not those of my employer :)
Blue_MiSfit is offline   Reply With Quote
Old 6th May 2011, 06:15   #3  |  Link
pwnsweet
Registered User
 
Join Date: Nov 2008
Posts: 100
Hi Derek,

Thanks so much for replying so quickly. I'll address each point you've made in the order they were raised.

There's no one person specifically that said all 720p encodes should be done at 4-5mbps, but from my time researching both here on Doom9 and elsewhere, it seems to be an 'unwritten rule' that people who want smaller file sizes whilst maintaining quality for most material follow. Obviously this 'rule' isn't set in stone and I know that for certain content (such as 2D animation) you could get away with 4-5mbps 1080p with few compression artifacts. A question I have concerning this unwritten rule has never been answered though. Does it apply to 1.78:1 720p (1280x720), 1.85:1 720p (where only 1280x692 pixels are not part of the border) or 2.4:1 720p (where only 1280x533 pixels are not part of the border)?

Being a fairly generalized 'rule' I would think it doesn't apply to any one specific case but the reason I questioned it is because the difference in pixels between the two possible interpretations (1280x720 and 1280x533) is significant - 921600 vs 682667. That's a full 35% more pixels for the full frame image that are not part of the border. Everytime I see someone say "I encoded my movie to 720p at 4-5mbps and it looked great!" I always wonder what the aspect ratio of the source is and if they cropped the borders. If they did, was the aspect ratio 1.78:1, 1.85:1 or 2.4:1? And finally, was their source animation, film or something else?".

The reason I'm interested in knowing which resolution to use for the 4-5mbps bitrate range is because I'm backing up these movies to single layer DVD discs and this is the average bitrate that a full length movie has when constraining the output size to 4480mb (DVD5/BD5).

Finally, I'm fairly new to x264 encoding so I'm using MeGUI to encode so I don't know how to post the x264.exe CLI arguments. I'll have a poke around in the options and settings and see if I can find out for next time.

Cheers!

Last edited by pwnsweet; 6th May 2011 at 07:32.
pwnsweet is offline   Reply With Quote
Old 6th May 2011, 12:19   #4  |  Link
asdfen
Registered User
 
Join Date: Oct 2009
Posts: 10
I just made 720p h264 1.2 mbps video which looks very close to original. So I am not sure where you are getting these rules/values from.

IMo just try it with different settings till you see the one you like best.
asdfen is offline   Reply With Quote
Old 6th May 2011, 13:47   #5  |  Link
me7
Registered User
 
Join Date: Mar 2007
Posts: 217
You should have said that you intend to save the movies on a DVD5 in the first place. CRF is out of the question then.

Forget custom cli parameters. For live action movies and CGI use "--tune film", for other animation use "--tune animation".
Start with "--preset veryslow" and let the encode run for some time to get a reasonable estimation of left encoding time (past the company logos at the beginning) or, if you use Avisynth, start the encode somewhere in the middle of the movie (Avisynth's trim function). If it takes too long lower the preset value to the next lower value and test it. Once you're satisfied with the speed stick with the preset for future encodes. That's it, don't set any parameters you don't understand, just use "tune" and "preset".

As far as the resolution go, some movies can look good enough even with 4MBit/sec at 1080p, others will blur and expose ugly macroblocks. These is no universal answer to be found here.
me7 is offline   Reply With Quote
Old 7th May 2011, 04:41   #6  |  Link
Blue_MiSfit
Derek Prestegard IRL
 
Blue_MiSfit's Avatar
 
Join Date: Nov 2003
Location: Los Angeles
Posts: 5,988
Indeed.

I'd suggest you rethink your intention of backing up your movies to DVD5... recordable DVDs are notoriously unreliable, and expensive. Hard drives are much more reliable and affordable in my opinion.

Regardless, you decision is your decision.

What me7 said regarding speed presets is also quite valid.
__________________
These are all my personal statements, not those of my employer :)
Blue_MiSfit is offline   Reply With Quote
Old 7th May 2011, 16:44   #7  |  Link
pwnsweet
Registered User
 
Join Date: Nov 2008
Posts: 100
Thanks for the replies.

In that Blood Diamond encode I actually did use the --tune film preset but according to this x264 settings page the only settings --tune film actually alters is deblock and psy-rd. As you'll see in my settings the --values for deblock and psy-rd are the same as that of the --tune film preset.

I did, however, change some x264 settings in MeGUI's advanced settings menu, namely: trellis, bframes, ref, b-adapt, direct and subme. The reason I made these changes is because, as I understand it, the value for each of these settings has a fairly significant outcome on the quality of the resulting image. I looked into increasing me_range to 24 as it is in the 'very slow' preset, but Dark Shikari says that doing so yields little to no benefit. But it seems like you are suggesting I should use all default x264 settings for every parameter and just use --tune film. If this is the case, I'm confused. Wouldn't doing that result in a significant reduction in image quality compared to my current settings, with the only benefit being a shorter encode time? I don't mind spending the 10-12hrs per encode (longer than that would start to become a problem) if I know that the extra time is being spent to significantly improve the image quality and, as far as I can tell, that's what those settings that I maximized do.

Last edited by pwnsweet; 7th May 2011 at 17:03.
pwnsweet is offline   Reply With Quote
Old 8th May 2011, 07:06   #8  |  Link
Blue_MiSfit
Derek Prestegard IRL
 
Blue_MiSfit's Avatar
 
Join Date: Nov 2003
Location: Los Angeles
Posts: 5,988
I wouldn't suggest using the default --preset medium. I'd go to maybe --preset slower, or veryslow if you can stomach the speed. --preset placebo is just that, a placebo.

Feel free to tweak things as well, but the speed presets work very well.

Derek
__________________
These are all my personal statements, not those of my employer :)
Blue_MiSfit is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 08:10.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.