I'm uploading gameplay videos to youtube, my raw source is x264 baseline 4.0 crf=18, 60fps, avi container
convert to x264 high profile 5.1, crf=26 or better, 30fps mp4 container then upload to youtube
upload the avi directly to youtube, let them process the video
the raw video looks good and really smooth, but If I encode it to fps 30 (by using MeGUI, ChangeFPS(30)), the result video looks choppy like most of the videos on youtube,
I do know that youtube only supports 30fps videos, I tried both convert to 30fps then upload, or upload the 60fps raw source directly to youtube, the results are the same.
(yes I know lower framerate means less smoothness, but...)
I saw there are some ppl on youtube, their gameplay videos still look really smooth even encoded in 30fps
Just like the videos captured from Digital Video Cameras,
is there anyway to encode videos from 60fps source to 30fps and still keep maximum smoothness?
or do I need some sort of filter to make smoother motion in lower framerates like 30fps?
here are my video examples on youtube, they are encoded from the same source.
fake 60fps (plays at 2.0x speed become 60fps, need at least 3Mbps connection for 360p without buffering)
required HTML5 player, enable it on http://www.youtube.com/html5 and play it at 2.0x speed.
I hope I can keep maximum possible smoothness on 30fps video just like the fake 60fps video
+ Reply to Thread
Results 1 to 23 of 23
Last edited by squall0833; 5th Apr 2012 at 09:11. Reason: typo
in general you could by encoding interlaced, but that wouldn't help with youtube
I'll try using mg262's motion.dll blendfps(), but the outcome may look odd, blend from 60 to 30 is about half, it may look funny
convertfps() does the good job but target fps must not lesser than 2/3 fps of original input
I'll see how then
it would be best if youtube allows 60fps video views someday
yes that would help to push 60p usage forward and also help to kill off interlace footage for home videos
BTW if you'd have web site, jwplayer in it, no problem with 60p , it is one of those false presumptions that somehow 60p has to be doubled in size , bitrate doubled and that your computer will choke and such., YouTube could step in towards 60p, it would be nice, why not, encode the same frame rate as original, same bit rate for 60p as for 30p if this is an issue for them, something to start with.
it is one of those false presumptions that somehow 60p has to be doubled in size , bitrate doubled and that your computer will choke and such.
Judging from this, 60p footage should at least have 50% more bitrate, but you suggest "same bit rate for 60p as for 30p".
Ps.: I'm also pro 60p.
Last edited by Selur; 5th Apr 2012 at 16:54.
I'll read that link later, ...., uncompressed video is doubled in size, that's clear, just to start there, so everybody understands that it cannot be double in size encoded,
It is being underestimated that smoothness of 60p has a profound effect , you watch it next to 30p (shot interlace with shutter up in the sky somewhere, very short) and it is nice, it is shown as was recorded, with particular shutter speed.
uncompressed video is doubled in size, that's clear, just to start there, so everybody understands that it cannot be double in size encoded,
Last edited by jagabo; 5th Apr 2012 at 22:19.
this script you load into x264 encoder
I do not use CQ but CRT, not sure why you insist to test it with it, it is going to be poeple to watch it, but anyway, one guy is talking sense there that you have to compare visually. You tried CQ with your video and was trying to figure out why is that that bob took almost double bit rate. Poisondeathray got some 2.6 ratio for some bob there, this is insane. Your question was there why it is happening with your video, academically stated, studied, you wanna know something but real world example by comparison is more handy I'd say.
Saying that, I get completely different results, source clean DV avi, 60i, and this keeps pattern for bitrate as I state below, pretty much for all videos 15% on the top of 30p for 60p:
83.1MB - 60p QTGMC CRF18
71.9MB - 30p QTGMC, CRF 18, selecteven()
and then I encode 60p and 30p same way with 2000kbps , 2pass, SAME bitrate and 60p is better by any means, that smoothness just beats 30p to the ground. No visual differences in quality either. Why do I have those results I do not know. Type of video, noise , as always everything turns around what concrete video we talk about, one cannot generalize here ? I always talk about camcorder videos here , no movies, TV rips etc.
As I said before somewhere I discovered that long time ago, encoded 60p with same bitrate by mistake as for 30p and could not believe that , before with 2pass I always got it too much bitrate because of that global "you need tons of more bitrate for 60p" scare crow
x264 30i: 22440 kbps
x264 30p (QTGMC, SelectEven): 16439 kbps
x264 60p (QTGMC): 18781 kbps
x264 60p (Bob): 25836 kbps
Those were all encoded at --crf=18, --preset=slow, --bff for the 30i video
And we compare 30p vs. 60p , if we start 30i vs 60p it is going to be even more interesting, I have also much more bit rate demand for 30i when tested then for 60p. Always was curious about that but it did not bother me too much because I do not encode H.264 interlace at all.
I just ran the same test on some shaky DV video, similar conditions:
x264 30i: 3861 kbps
x264 30p (QTGMC, SelectEven): 2488 kbps
x264 60p (QTGMC): 2867 kbps
x264 60p (Bob): 3545 kbps
Well, it makes sense in a weird logic sort of way...
Mpeg tracks "differences".
The differences between frames is less with 60p than with 30p or 30i, so those smaller differences are easier to track. More efficient.
Works similarly with HD vs. SD, or high Fs audio v. Std Fs audio, Surround vs. Stereo vs. Mono, Stereo3d vs 2d...
Scott"When will the rhetorical questions end?!" - George Carlin
No, use SelectOdd()
I do not use CQ but CRT, not sure why you insist to test it with it,..
Thought I might chime in on this. I had to make a similar decision somewhat recently, insofar as what to do with a video whose framerate exceeds Youtube's spec. In my estimation, there were only two options. Either blend every pair of frames, or upload it as 60fps despite what Youtube would ultimately render it as. I chose the latter.
This is because Youtube will eventually find the need to begin supporting higher framerates. I expect this transition to occur more or less in sync with the first proper trailer(s) for the Hobbit movie. Said movie was filmed at 48fps and will be shown in theaters at 48fps. In order to get theaters on board, the idea will need to be pushed into public awareness. Meanwhile, most people looking for new trailers go to Youtube to find them. I fully expect Youtube to be approached in order to ensure that they enable a 48fps presentation for the Hobbit trailer(s). The alternative prospect is that the trailer is made exclusive to some lesser video service which already does 48+fps... and this is quickly re-posted a hundred times to Youtube where most people will still be searching for it. So, with Youtube already prospectively exceeding 30fps for the first time, there is every incentive for them to take the plunge and support higher framerates universally.
And when that happens, I think it is a reasonable bet that past videos which exceeded Youtube's spec will be re-encoded, one by one, until the entire library is up to speed, just as had been done when Youtube went HD. This would include my video, of course. ;p