Do you mean filesize as in bitrate (because filesize = bitrate x running time) ? or resolution? 720x304 ?
If this isn't for a specific device (they might have other limitations besides resolution , like encoding settings, levels, b-frames, reference frames) , then you can play with higher resolutions as well (it doesn't have to be full HD, something in between). Resolution is part of quality.
Given that you have a bitrate limitation for your flash drive, it's going to depend on the specifics of the movie. This movie has quite a bit of action so it won't compress very well - you might not be able to go much higher in resolution before it starts to degrade too much. I'd be willing to be it will look better overall at a higher resolution given the 2111kbps limitation
+ Reply to Thread
Results 31 to 49 of 49
-
-
try this:
subme 9 or more
bframes=6 or more
ref frames = 6
Macroblocks: uncheck P4x4 (because it is useless almost)
B-Pyramid = Normal
M.E.Range = 16 (24 is too high)
B-frame bias = 20 (max 25) ; this will make more consecutive bframes and will improve compression -
-
-
-
4 reference frames and 3 b-frames are usually safe for high profile, unless it's a device that only supports baseline profile (e.g. early generation ipods) , (then no b-frames for baseline profile, no cabac)
When using high amounts of reference frames and b-frames, often people randomly enter the highest values - that's the wrong way to do it. You should look at the log files and see how many are actually used for that particular source, with those particular settings (e.g. if you enter 16, it might only use 4 consec max) - it will be exponentially slower with no added benefit in quality or compression. Usually content like simple animation can benefit more than live action footage. When you encode more you will get a feel or rough idea of what settings to use -
Yes, I'd stick with 3 or 4 b frames and reference frames. If you use more you're just asking for playback problems on many devices -- just to save a few percent on the file size (CRF encoding) or quality (bitrate encoding).
Try it yourself. Perform a CRF encode with 16 b/ref frames. Then perform the same conversion again using 3 or 4 b/ref frames and see the change in file size. You'll see the difference is only a few percent. And 16 b/ref frames takes much longer to encode.Last edited by jagabo; 25th Jul 2012 at 23:28.
-
Thank you, jagabo and poisondeathray.
I still can't figure out what settings i should use. I don't really wanna use CRF as the filesize is so random, but i can't really tell a difference between the settings somespirit gave me to try (changed B-frames to 3 and R-frames to 5 this was before you told me to just use 4) and the original settings i posted while back.
Can any of you?
http://screenshotcomparison.com/comparison/137548
New is somespirits settings and old is my original settings.
I guess you can see a little more detail in somespirits settings, but when actually playing the video. It kind of doesn't look as good.
What's a good M.E Algorithm to use? I've been using Multi Hex. -
By the way, it's pretty much useless to compare 1 screenshot (e.g. they might be different frametypes, e.g. P vs. IDR, vs b) . Some frames in some sections might be better, some worse. A valid comparison would put equal frametypes at the same average bitrate
All the settings have tradeoffs (pros vs. cons - usually enocoding time is the tradeoff). Some of them are exponentially slow for minimal quality or compression gain
There are no "best" settings, because you might use custom settings for different types of sources.
Start with this to get a rough idea on what the settings do
http://mewiki.project357.com/wiki/X264_Settings
If you don't know or don't have time to experiment or learn, just stick with the presets, that's what they are there for.
Filtering can also play a large role in the final quality (often it's more important than encoding settings) -
The difference between using 3 b/ref frames and 16 b/ref frames may give you 2 percent difference in file size with constant quality (CRF) encoding. That means, using bitrate based encoding, you could use a 2 percent lower bitrate on the 16 b/ref video to get the same visual quality as the 3 b/ref frame video. Or, at the same bitrate, the 16 b/ref video is "2 percent" better quality. You can't see differences that small.
You're also working at very high bitrates for the video you're encoding. You'll barely see a difference between your 2900 kbps encoding and an encoding with half that bitrate. So the difference between 2900 and 2840 kbps (about 2 percent lower) isn't visible. -
I guess i'll just switch to CRF then since the little bit of an extra bitrate from using a larger file size isn't really worth it. Thank you guys for the help. I'll have to read up on that link poisondeathray gave me even though i did see that before was just to lazy to read it all.
It look's like there is no easy way to encode video though so i'll have to get over it and read up on it. -
The problem with bitrate based encoding is you never know what bitrate is required to deliver good quality for a particular video. Every video requires a different bitrate to maintain quality. With constant quality encoding (CRF in x264) you always get the quality you specify. The bitrate (and hence file size) varies depending on the nature of the particular video. The only reason to use bitrate based encoding is to get a specific file size (file size = bitrate * running time) -- like 700 MB to fit on a CD, or 4300 MB to fit on a DVD.
-
Is there anyway to predict what the filesize is going to be with the CRF settings? Like a rough estimate? Like i know you can't get pinpoint accuracy, but is there anyway to keep it in the lower 1gb area or does it depend on the movie itself like length etc on the compression?
-
It depends on the movie itself. Its length, the frame size, frame rate, and nature of the picture. Noise (film grain, static, etc.), and anything else that causes the picture to change from frame to frame (motion, wavy water, fog, smoke, flames, explosions, flickering lights, fade-ins, fade-outs, etc.), increases the bitrate requirement.
Last edited by jagabo; 27th Jul 2012 at 11:33.
-
The best you can predict is a "guess-timate". There are so-called bitrate calculators around, but they're often worst than guessing. With a little experience doing this, you'll get the idea. You can always try just converting or encoding a small part of the project rather than the whole thing, to get some idea of where it's going.
Last edited by sanlyn; 23rd Mar 2014 at 06:41.
-
I'll have to just play with it and learn from my mistakes i guess. I understand nothing is easy. I do hope to learn and understand all this a bit better with time.
Thank's for all the help, jagabo. -
It's quite simple:
With bitrate based encoding you know what the file size will be but you don't know how the quality will turn out. If you use very high bitrates all the time most videos will look good but you'll be wasting bitrate on many videos that don't need it.
With quality based encoding you know what the quality will be* but you don't know how big the file will turn out.
When the two methods produce a file of the same size the quality will also be equal. The easiest way to verify this is to perform a CRF encoding, see what the average bitrate turns out to be, then perform a 2-pass bitrate based encoding with that bitrate. The files will be the same size and if you examine them you'll see the look pretty much alike.
With x264 in CRF mode try settings between 15 and 22. The full range is 0 (highest quality) to 51 (lowest quality). I usually use 18. From what I've seen most people use slightly higher settings (lower quality). The video in your earlier post would probably correspond to a setting around 15. Start a short, varied, video and experiment with a few different CRF values until you find what you're comfortable with. Then you can use that for all your encodes. Or you may want to use a little lower for "important" videos, a little lower for stuff you don't care about.
* Using a high quality setting won't improve a low quality video, just keep it from degrading more. -
What's wrong with that though? Does it corrupt the picture or something? If it will look the same using CRF or 2pass, why not just use 2pass so you know what the filesize is going to be then just make the bitrate lower on movies that aren't as long? Could i just do that? I just don't understand how you can tell the bitrate is to high just by looking at a video. I wanna learn how to do that so i can do that as well. What kind of things should i look for to calculate the bitrate besides length.Last edited by kkiller23; 28th Jul 2012 at 07:14.
Similar Threads
-
x264 settings/AVCHD
By soneca in forum Video ConversionReplies: 6Last Post: 13th Jan 2012, 05:51 -
x264 Mediainfo to MeGUI x264 Settings
By shagratt71 in forum Video ConversionReplies: 0Last Post: 1st Jan 2012, 04:59 -
Question about x264's Tuning settings
By CubDukat in forum Video ConversionReplies: 0Last Post: 4th Dec 2011, 23:17 -
Optimum x264 encoding settings
By zammil in forum Video ConversionReplies: 27Last Post: 18th Dec 2010, 00:32 -
FfmpegX x264 settings
By bvdw in forum ffmpegX general discussionReplies: 4Last Post: 20th Jan 2009, 13:23