Original render (24mbps, two pass encode, very slow H.264 profile, h.264 Level 4.1) with Handbrake
Youtube encode (8mbps)
Handbrake encode (8mbps, Very Slow, h.264 level 4.1)
Why is Google so shit at videos?
>>53341976
What did you expect?
>>53342026
A 2560x1440 video to not look like absolute trash.
>>53341976
Quality definitely is way worse, but Youtube does this to preserve harddrive space and to offer vod to worse connections.
However, the color is less greyed out, which I personally like more.
>>53342372
You prefer oversaturated colours?
>>53342372
>However, the color is less greyed out, which I personally like more.
YT reallys fucks with videos. If you have a video that's slightly dark whatsoever then YT just makes it pitch black.
>>53342384
I've been comparing the two for a bunch of times now and 1.png just seems to have a grey filter over it.
>>53342412
I've noticed the really dark blacks yeah, it's horrible in some videos, especially if you have even a small amount of glare on your screen.
>>53342484
comparing to the source AVI it appears you may be right
does youtube encode with cpu or gpu?
>>53341976
Try the VP9 versions
>>53342596
Pretty sure they use x264, which is CPU-based
You are NEVER EVER supposed to upload lossy video to youtube. When you convert from the raw video you MUST use lossless H264 (aka CRF 0). If your internet sucks ass then use a CRF of 10 or 16 if your internet is really that slow. Never use a bitrate as this is to target bitrates not quality. Also use the fast-medium preset if size isn't too much of an issue to upload, it sucks having to wait hours for an encode set with the very slow preset.
Your youtube video will still have lower quality but it won't be as bad.
9/10 times the reason why a youtube video looks like shit is because the uploader transcoded the source to a smaller video file because his connection was shit or he used a random bitrate instead of CRF value.
>>53342639
My internet is dogshit.
What CRF would you recommend for the video?
My upload is 1mbps
>>53342639
>crf 0
>lossless
>>53342663
How long are you willing to spend uploading the video?
Multiply that time by your connection speed and pick your CRF in order to aim for that size.
>>53342684
I can leave it overnight, its nothing too pressing.
For a 1.9GB file it took about 4 hours.
>>53342663
What is your connection speed?
How long is the video?
they most certainly use faster presets/settings and/or gpu encoding
they get a shitload of uploads daily, they need to be able to encode shit faster than people upload shit
>>53342691
Upload is 1.1Mbps
Video is 10:47
>>53342672
>"The range of the quantizer scale is 0-51: where 0 is lossless, 23 is default, and 51 is worst possible. A lower value is a higher quality and a subjectively sane range is 18-28. Consider 18 to be visually lossless or nearly so: it should look the same or nearly the same as the input but it isn't technically lossless."
https://trac.ffmpeg.org/wiki/Encode/H.264
https://www.youtube.com/watch?v=dJQX0tZsZo4
>>53342744
>how it affects gaming
pls be kill,,sir
>>53342744
that feeling when totalbiscuit will die this year ;_;
>>53342714
>Upload is 1.1Mbps
jesus
>Video is 10:47
Try a CRF of 10 with the medium preset, you should get around 1GB for the video file and upload time around 2 hours.
Though if you have the patience just go balls out with a CRF of 0 with a fast-medium preset (both will be lossless) and let the video upload overnight.
>>53342773
It's 15% of the way there and the file is 500MB already
>>53342787
Hmm, guess I underestimated the potential file size, sorry. You're gonna have to settle for a CRF of 16 if you're too much in a hurry it seems.
>>53342856
I'll try lossless but on a slower preset to see if i can save some space.
It says it'll only take 2 hours to encode, and I'm going out later anyway, so I'll trigger the upload as i leave.
thanks.
>>53342980
np. Good luck you brave, brave man.
>>53343032
At least I have sufficient cooling for this.
>>53342719
CRF 0 is ``lossless".
Fact is, H264 is a lossy codec.
Only H.265 has a lossless setting, a very specific and well defined setting at that, unrelated to CRF.
Stop misinforming people. Autistic details should propagate over wrong ones.
>>53342744
Wow, a youtuber talking about encoding technology in a way that isn't actually blatantly wrong?
Almost everything he *said* is fantastically accurate, the only thing that bothered me is the fact that he claimed downloading videos off youtube required another transcode, and the fact that his “testing” methodology is flawed as all fuck (but he did mentioned this, so he's aware of it).
>>53343215
Oh, that said, he apparently doesn't know what VP9 is or how it factors into the equation, or he probably would have mentioned it.
>>53343179
bruh you just said crf 0 was lossless...
>>53343235
how come my H.264 encode is better quality at the same bitrate?
Surely VP9 is trash
youtube has to encode thousands of videos at the same time, it's no surprise they use quick settings which look crap
>>53343319
The fact that people upload crawling_in_my_skin_amv_reencode9001.flv doesn't help either.
>>53343276
>how come my H.264 encode is better quality at the same bitrate?
Short answer: Because you did something wrong.
When comparing apples against apples, VP9 will easily achieve over twice the space efficiency with optimal settings compared to H.264.
