Low quality when uploaded to youtube

[Refiu]

Youtube encodes 2 versions. A low quality version and a high quality version. Are you sure you weren't just looking at the low quality one?

I'm a lilttle bit late with the reply but anyway... Of course I am certain. I upload 1280x720 resolution at 23.976 and youtube gives an option of watching it in 360p, 480p, and 720p and they are all terrible; 720p is somewhat decent but not as good as in other people's videos. No matter what I encode the lossless version to before uploading, it looks terrible once being published on youtube which totally baffles me off. I know bunch of editors who also work with Lagarith and Huffyuv and then convert their videos to mp4 in various resolutions and it just looks excellent on youtube.

I have no idea where I'm making a mistake... Hmm, may I try this way: say you want to make an AMV with 1280x720 MKV file using Sony Vegas... so how would you do it? Just in a few general steps (yes, I've read all tutorials concerning Avisynth and VirtualDub and all that).

[Mister Hatt]

This might sound a little odd, but to get better quality on YouTube, blur your video before uploading it. The reason things look so bad is because of how their quantizer works. If there is less sharp detail to suck up the bitrate, it won't overquantize and generally comes out a lot better.

[Mally]

Well the best way of ensuring the best quality on youtube . Is Making sure you have a High Quality Blue ray , dvd rip or equilivant rip . Making sure that when editing the source that no resampling methods have been used . Then you can simply use Xvid settings found on 'AbsoluteDestiny & ErMaC's Technical Guides to All Things Audio and Video mk3' Guide . And you should be fine .

With Myself i use Sony Vegas , so i just simply right click the appropriate clips when brought onto the timeline . Go onto properties and disable the resample . Then Export the final edit as a Lagarith Compressed file or Lossless Equilivent with uncompressed wav source . Finally Compress video and audio on Virtual Dub Mod .

[post-it]

.. there are too many things going-on with videos these days that the basics
are all but forgotten.

A) There are filters that stabilize the Luminance balancing "all" the frames against
each other for a "pure-motion" and "easy-viewing" video -- but no one uses them!
.. a codec would have no problems if everything were encoded with only 256
colors because if a COLOR changes from Blue to Grey or Red ... well, we can see
that happening but; just how many shades of colors and shades if Brightness
are taking place between every frame of an AMV???? -- we don't see it .. the CODEC "does" see it.

..Virtual Dub is the only editor which has a filter available to handle this, "Auto-Level"
a stabilizer for Brightness. a filter for Avidemux failed. AviSynth never had one.
No Video Editor out today even offers this as an option.
Thin-out your Video's before you encode them; any codec can shrink it from that point.

B) Resize the images X4 Bi-Cubic, Soften/Blur it, Resize the images ÷4 Bi-Linear.

if your video is not clean, the codec will reward you with "shimmers" "micro-burst's" and distortions later.
The codec is not the issue, even the .RMvb files are cleaner than those found on uTube.

Thin-out your Video's before you encode them.

[post-it]

Thanks ^_^

[Mister Hatt]

Apparently being blunt is bad. To be more polite: anyone seeking advice in this thread should ignore post-it as he has no clue what he is talking about. He has no idea how YouTube's encoder works nor what it compresses better or worse. The best advice is still to add minor blur to videos; line thinning makes it worse as it makes macroblocks more complicated and that in turn makes your video get overquantized. The encoder is very much the issue and messing around with colourimetry isn't going to do anything about it. Knowing even a tiny bit about DCT quantization and entropy encoding should make that obvious enough.

[post-it]

. The encoder is very much the issue and messing around with colourimetry isn't going to do anything about it. Knowing even a tiny bit about DCT quantization and entropy encoding should make that obvious enough.

you really have no clue what your refering to -- do ya fool!