Bitrate for youtube video must be 8 mbps? Why?

[hello0]

Hello, i read on the youtube video settings site the "Recommended upload encoding settings "

Recommended video bitrates for uploads
1080p should be 8 Mbps upload when its 30 f/s

Why should it be 8 mbps?

What happens when its less than 8 ?

[Skiller]

These recommendations are pretty pointless, you should ignore them.
You should chose a bitrate (or better a CRF) that is a good compromise between quality and upload time for you.

What happens when its less than 8 ?

Nothing.

[hello0]

If the bitrate isn´t ok, will my video lose quality after it´s uploaded to youtube?

[Skiller]

It'll always lose quality after it's uploaded to YouTube because YouTube re-encodes the video to it's standards. You can minimize the loss by uploading a good quality video to begin with.

[hello0]

ok. Could you tell me, what i need to make the video a good quality so that it looks after uploading perfect? I mean the settings of the video etc

[manono]

ok. Could you tell me, what i need to make the video a good quality so that it looks after uploading perfect?

It will never look 'perfect' after YouTube reencodes it, as Skiller already mentioned. All you can do is to upload a good quality version using, as Skiller also suggested, a reasonably low CRF. I use 18 when making MP4 videos for upload to YouTube.

I mean the settings of the video etc

Are we mind readers? If you need something besides a CRF suggestion, it might help if you told us the program you're using and the intended final format.

[hello0]

Where can i find the CRF ? Final format is H264/AVC. My program is Adobe premiere

[manono]

My program is Adobe premiere

I would think it would have a single-pass quality-based encoding setting. But I've never used it so someone else will come along, I'm sure, to help out.

[_Al_]

Premiere does not have it. Bitrate has to be "guessed" using it's encoders. Premiere uses 1pass VBR (do not use), 2pass VBR or CBR.

But try 15Mbit for example 2pass VBR, with average 15Mbit, if that would not help, it means higher bitrate might not help at all uploading to YouTube.
Maybe your original is interlaced on the top of problems, where deinterlacing is worsening the quality too.

[Cornucopia]

• Use Premiere to edit with and to make a master file in your super-high-quality intermediate edit format (such as losslessly compressed like Lagarith, UTvideo, MagicYUV, etc., or barely lossily compressed like ProRes, DNxHD, Cineform, Edius HQX, AVC-Intra/Ultra, etc.) - yes, it will take up a LOT of space.
• Then, use a different program (NOT Premiere) to encode a high quality (e.g. CRF-18) AVC-in-MP4, such as Handbrake, MeGUI, Hybrid, Vidcoder, etc.
• Upload that 2nd version.

Scott

[hello0]

Thanks. What exactly do i have to do with Handbrake etc..? I will put the master file in it and this program will add the CRF into the video ?

[SameSelf]

It is a whole lot easier to just frameserve out of Premiere using Debugmode Framserver to a signpost.avi and feed that to your encoder of choice (e.g. x264 cli, ffmpeg, ffmbc, and the others mentioned above) using Avisynth. That way you avoid the cumbersome step of writing out a massive lossless intermediate file which saves time and drive space.

As for Youtube encodes, I just feed Youtube ~35 Mbps for HD, iow, CBR because I hate 2 pass encodes for Youtube. This is close to the max blu-ray bit rate. As has been said, Youtube will re-encode no matter what you give it, so I just give it lots to work with.

[manono]

I will put the master file in it and this program will add the CRF into the video ?

I think you add the CRF. I don't use that either, but there are guides for the use of Handbrake at that Handbrake link. One of them:

http://www.dvd-guides.com/guides/blu-ray-rip/259-rip-blu-ray-to-mp4-using-handbrake

in Step 5 shows where and how to give it an 'RF'.

VidCoder is said to be easier to use than Handbrake. It its tutorial:

http://www.digital-digest.com/articles/VidCoder_Tutorial_page2.html

also in Step 5, it also shows how and where to set a 'Constant Quality'.

[hello0]

thanks. Do you know, if i could use this editor too?: http://www.videosoftdev.com ? I have it too, and it has a Two-pass option, but when i want to save the file, it doesn´t save it. I don´t know why. It starts saving it on the drive, but i dont see it on the drive. It only saves the single pass.
Here a pic

[manono]

http://www.videosoftdev.com ?

Never heard of it. Sorry.

Your picture doesn't seem to indicate a quality-based encode is possible.

[SameSelf]

thanks. Do you know, if i could use this editor too?: http://www.videosoftdev.com ? I have it too, and it has a Two-pass option, but when i want to save the file, it doesn´t save it. I don´t know why. It starts saving it on the drive, but i dont see it on the drive. It only saves the single pass.
Here a pic

You stated above that you have Adobe Premiere. That is not a lightweight NLE. It has many built-in, powerful effects. You would have to spend a lot of money to find a better NLE. So my advice is stick with it. Learn it inside and out. Every software has its strengths and weaknesses. Don't go downloading other video editors thinking Adobe Premiere is not the right tool for the job.

The only real choice you have is deciding how you want to encode your timelines from Premiere. You can either:

1. Use the built-in Premiere encoder along with one of the many presets that you can customize as needed. Bit rate is one of the customizable options. Choosing a high bit rate like 35 Mbps for HD material is a simple and adequate approach for Youtube if quality is what you are after.

2. Frameserve your timeline from Premiere to a third party encoder of your choice. Most people use x264. People choose this route and encoder over Adobe's built-in encoder because Adobe's built-in encoder is not that good, at least not compared to x264 which is freeware.

With that said, there is nothing wrong with option #1, and option #2 is for people who obsess about encoding quality.

[Cornucopia]

It is a whole lot easier to just frameserve out of Premiere using Debugmode Framserver to a signpost.avi and feed that to your encoder of choice (e.g. x264 cli, ffmpeg, ffmbc, and the others mentioned above) using Avisynth. That way you avoid the cumbersome step of writing out a massive lossless intermediate file which saves time and drive space.

As for Youtube encodes, I just feed Youtube ~35 Mbps for HD, iow, CBR because I hate 2 pass encodes for Youtube. This is close to the max blu-ray bit rate. As has been said, Youtube will re-encode no matter what you give it, so I just give it lots to work with.

Not always. DbFs has had a rocky history with certain installations of Premiere (and other NLEs). It will be "easy" until it craps on you and you can't get it working correctly again for a while. Ultimately, frameserving isn't really any "easier", just potentially drops out one step of the traditional workflow.
Also, writing to a master file is not "cumbersome" at all, unless you are extremely strapped for space (which if you are truly interested in video & editing, you should never be that strapped). Plus, it makes much more sense to "distribute" or "delegate" your CPU load, so your rendering out of PPro, etc. is optimized and not burdened by concurrent processes of the frameserving and the encoding. And subsequently, your encode is prioritized and not burdened by Ppro & the frameserving, just ALL encoding cycles.
Add to that the possibility that you just may intend (if you are interested in distributing through multiple channels & devices) to encode to separate versions, and that means the proportion of "additional burden" of prior rendering gets lessened (to the tune of n+1/n encodes).
And most importantly, you now have a pristine master file to archive, refer back to, compile with, reuse, etc. without appreciable loss, and any other thing you want to do with it (all without having to re-run & re-render PPro). It's a safer way of working, which is why true professionals only do it this way (unless they are doing a one-off and are extremely strapped for time).

Scott

[_Al_]

Yes, that intermediate huge file is used, perhaps even even high bitrate H.264 with I-frames, but that professional comes home , edits his home video on laptop, where sources are gathered from whatever, phones, camcorders , photo cameras, uses debugmode frame server to get it out using CRF, renders only one instance, maximum 2x. Perhaps through the night, so there is just result in the morning. Only those original clips are archived, if NLE offers smart export , most likely that one is kept (DVavi, HDV), perhaps project as well.

[Cornucopia]

Good luck with that! My productions are too important to treat in such a cavalier and risky way.

One instance where a backup of your precious multi-hour, multi-cam, multi- track complex and arduous endeavor cannot be fully and properly restored and you'll change your mind.

Scott

[hello0]

Thanks for your infos. I downloaded MAGICYUV but I dont know how to use it. I would like to test it in my other editor, that i mentioned here, too. https://www.videohelp.com/software/VSDC-Free-Video-Editor
Could you tell me how i have to use it?

[Cornucopia]

It's a vfw and QuickTime codec (maybe directshow too?). You just install it and as long as your editor supports 3rd party installable codecs in those platforms/subsystems/api's, it'll add them to the list of available codecs to choose from.

Choose it and specify which color subsampling format you need (yv12 is a likely common one) and save.

Beware: your files will be huge/the bitrate will be high, particularly for full hd and especially for 4k, so you need to make sure your system can keep up and handle it. But the quality should be optimal.

Scott

[hello0]

Hi, unfortuntely this program does not allow third party codecs. I installed Magicyuv but i dont see it in the codecs list.
The ones on the picture are all they have. Shall I take the Lagarith one?




also for lagarith, here i can set the yv12


If i choose H264 then the settings are:

[Cornucopia]

Lossless codecs are all equivalent quality-wise (assuming same bitdepth & colorspace), their differences are mainly as regards their performance (CPU load, decoding speed, etc.), their compression ratio (varies, but none really go beyond 2.5-3.0:1), their robustness (non-buggyness) and the choice of supported platform.
So, in general you can switch out one for another, as needed (of course, don't be under the illusion that they can decode each others' stuff - they can't).

For many/most things, Lagarith is fine.
Note: your problem with MagicYUV showing up may have to do with 32bit vs. 64bit compatibility (they all have to be the same).

Scott

[hello0]

Magicyuv has been saved in system 32 and SysWOW64, but it does not appear in the codecs list in that editor.

So I save the video with the Lagarith codec and after that, I encode a high quality with for ex. Handbrake or VidCoder, right?

[Cornucopia]

That's what I would do. Then, if you don't quite like the final encode (which you should have saved the settings), you just tweak the encode settings and run it again.

BTW, good habit: take a 1-5 minute segment and do a few test encodes to see which looks best, then use that best setting on the whole program.

Scott

[_Al_]

encoding for youtube he is fine with CRF and 1pass, no need to wait to encode the whole thing to see how it is going to turn out

so he might encode only segments of interest quickly testing some CRF values, if satisfactory, encode the whole thing with satisfactory CRF,
after couple of encodings he does not need to test those samples anymore, he will know his CRF and starts to encode right away

[SameSelf]

Not always. DbFs has had a rocky history with certain installations of Premiere (and other NLEs). It will be "easy" until it craps on you and you can't get it working correctly again for a while. Ultimately, frameserving isn't really any "easier", just potentially drops out one step of the traditional workflow.

I can't speak for every NLE since I only use Adobe Premiere Pro. And, I can only speak for CS3 and CS6. Both versions work flawlessly with DebugmodeFrameserver although the version of DMFS needed is different. So if you have an older version of PPro, you may need to roll back to an earlier version of DMFS to get it to work. I don't recall specifically which version is needed for CS3, but I seem to recall it is a very early version. As for drop outs, I haven't experienced any. But then again, I have a fairly powerful 'puter so it could be that YMMV.

I have a rock solid workflow right now that involves grading my footage in AE, dynamically linking that to PPro to edit my footage, then rendering with x264 using DMFS. I don't envision this changing any time soon. Having to insert a lossless render at some point into this workflow just complicates it and takes extra time and space. Lossless encodes of HD content are ginormous. And for what reason? I have the masters and the final render. All I need.

[_Al_]

To have that intermediate is good if one wants to keep master of edited footage, but for home purposes or even semi-pro purposes (web) workflows hardisks fill out really quickly.

Other reason to have intermediate (and then even deleting it) is if your sequence is time consuming to render and there is 2 or 3 delivery formats, so AE > PPro > dmfs might be. If Premiere has to compute frames every time it is served it might add some time as oppose to render that intermediate once and then just encode different versions.

[SameSelf]

Ahh I see now. Good tip if I ever find myself needing to scale.

[hello0]

ok thanks i will do it like that then.
What´s with Enable null frames, Multithreading and prevent upsampling? Do i need it?