YouTube uploading with best quality

In the past, I used the integrated feature "Share to Youtube" for uploading my videos. Since few months I started to render to disk with native ProRes 422 format, and then manually upload to Youtube web. But I am not sure that it is the best way. I look for quality over any other consideration.... am I doing it right? anyone here uses different procedure?

Maybe this is a previously discused subject? though I can't find any info. I guess the only way is trial and error... Right now I am using Apple Compressor's Youtube setting, i will upload that output to test posible qualty diferences with former ProRes upload..

alvarog01 wrote: … I look for quality over any other consideration....…

in theory, a less compressed file (prores) should result in a 'better' YT experience, than a h.264 encoded file -but that's what YT itself recommends:
support.google.com/youtube/answer/1722171?hl=en

But nobody knows for sure, what or how YT handles uploads...

Plus, prores is by far larger than h264... bandwidth/time is probably wasted without 'effect'.

Speaking of quality:
The most damage is done way before any compression - on recording: too much gain due to low light conditions, 'noise', hectic pans, too high shutter speeds, too much/wrong compression in camera, wrong encoding on processing the video etc etc....

I noticed many years ago in some test, that a jpg created terrible banding in a video just because itself was too much compressed - using another jpg, which was 10x bigger = no banding. ...

And the 'pixelpeepers' do epic discussions, what effect it has, to NOT output within the rec709 specifications....

I would follow YT recommendations - and focus more on 'quality' before upload....

Karsten Schlüter wrote: in theory, a less compressed file (prores) should result in a 'better' YT experience, than a h.264 encoded file -but that's what YT itself recommends:
support.google.com/youtube/answer/1722171?hl=en

.

I have just done a test: I uploaded ProRes 422, and later a Compressor YT version. The first one was slightly more detailed / less blurred, especially concerning text / titles.

But even with this intermediate codec, artifacts pop up, pixelations and so in certain scenes with a lot of 'movement' and FX. (my hard disk render hasn't any of these artifacts). It is annoying. Maybe Vimeo has more quality player but I haven't tried.


Edit: I record my videos with the FaceTime HD Macbook Pro webcam :lol: but video seems ok on FCPX, and again my renders do not show any of the artifacts that pop up in Youtube uploaded version.

The data rate of the file you upload also makes a difference. The artifacts, pixellations, etc. - especially in scenes with lot of movement - might be better with a higher data rate. In fact, I do use the Compressor presets for YT/Vimeo, but I make copies of them and increase the data rate.

It may be overkill, but for 720p I use 10,000 kbits/sec and for 1080p, 15,000 kbits/sec.

The files are larger, but not as large as prores.

I'll try to play with data rate settings then, thanks for the comment.

alvarog01 wrote: .. I record my videos with the FaceTime HD Macbook Pro webcam :lol: but video seems ok on FCPX, and again my renders do not show any of the artifacts that pop up in Youtube uploaded version.

please, don't take the following as a personal offense, but ... :
This is nonsense!

A webcam delivers super-high compressed, (be)low-standards material. On first sight probably 'nice' (=artificially enhanced, esp. over saturated colors and processed 'sharpen') , but by far no base for any meaningful 'test'. sorry to say...

compression artifacts 'multiply' - although the 'naked eye' doesn't spot them, they are there, for sure, any compression is lossy! on next compression level (h264 => h264, from FCPX export to YT conversion) they get (more) visible.

Just make a simple test:
select any of your 'raw' clips in FCPX, and what does the scopes tell, hmm? I'm pretty sure far from 'filling' 0-100%, perhaps outside legal ranges etc etc

........ no, a webcam doen't deliver any 'quality' discussion worthy material ........

Hi Karsten, I take no offense at all. I am here to learn from more skilled people.


The material imported from FaceTime to FCPX seems ok, also after editing, applying FX and rendering the video shows fine. I upload this whole ProRes 422 file to Youtube, it takes ages but I start the task on night and go to sleep, so doesn't bother me. Maybe I am too picky, but I notice that the YT uploaded version clearly shows some degraded zones. I know that YT recompress everything to some H.264 for the HD stream version, but... having the original ProRres file, shouldn't YT encoding process keep the original quality? By curiosity I took the ProRes in my laptop and compressed it to H.264, and the result shows fine.

Sorry if I misunderstand your criticism.

you make (unconsciously) a few assumptions, which are probably wrong.
for a start, watch this extreme example:
gizmodo.com/5555359/the-weirdness-of-a-y...-uploaded-1000-times


What's happening there?
h264 video gets uploaded to YT; YT converts (again!) to h264; on the first run: no noticeable quality loss.
Then, that guy downloaded the YTvideo, and uploaded it again. YT converted again: ....
... a 1000x ...

assumption #1:
when input is h264, and output is h264, this is a 'native' process, therefor lossless.
That is wrong!
That hype of 'native processing of source material' is based upon the assumption, any encoder would 'hand thru' the video. Unfortunately this is impossible with h264. With dv it was, that was an intraframe codec. h264 is an interframe codec. wiki for the difference......

asumption #2:
You can 'see' compression failures.
uhm, well ... for sure, a 720p file with 10mbps vs. a 720p with 400kbps IS indeed noticeable better in pic quality. But a twice, with 10mbps encoded h264 ...? No human eye can notice the difference - but there IS one! That's what I wanted to express with my remark about 'multiplying artifacts' ...

assumption #3:
Quality is in the eye of the beholder.
Sure
No, there are technically measurable criteria too; what LOOKS like a good picture, is probably just terrible.
Spend two evenings watching these award winning documentaries:
www.zacuto.com/great-camera-shootout-2010
www.zacuto.com/shootout-revenge-2012

A handful of super-pro cameras were set-up for recording the same scene again and again; then, under perfect conditions, they were shown highly professional movie-makers (Coppola) in Hollywood, London, Paris - they should tell, which cam delivered what. Quality-wise....

To spoil here: one device was an iPhone...............

Yepp, 3h, lots of text, many above my head, but I learned tons of worthy info about pic-quality.
______________

To discuss seriously 'quality', it is plane nonsense to use a webcam as source. And to judge, 'prores is slightly better' ... perhaps you notice a difference. OK, then it's up to you to decide, this grain of difference justifies a by factor 50-100x higher bandwidth in upload.

Most probably, your audience doesn't.
Or, notices instantly "Done with a cheap webcam - what 'quality' at all?? LOL" .........

_____________

I'm following attentively the discussions on the net about consumer-cams in 4k - such as the Lumix FZ1000, which is (close, ... uhm: not that close) to my price range... and there are tons of 'demoes' avail at YT, which are performed under so low circumstances, shaky handheld , low light, compressed stuff ... I'm alerted to say: What quality??? On the other hand, that Zacuto Shoot Out demoes, even a tiny cheap cam in the hands of a pro delivers superb pictures - but, when you know where and what to look at, an iPhone never replaces a RED or Arri .. . even when the Secretary of Finance in my household doesn't see a difference..........

QUALITY is an endless topic; and from "I can afford it and I like it!" to "You're a pixelpeeper!" it is a very short way.

When your webcam fulfills your needs, and you think a prores upload justifies to burn the bandwidth - just do it!