YouTube streaming - bitrate limit

  • Thread starter Thread starter Deleted member 330497
  • Start date Start date
If you have the GPU to spare, enabling will give higher quality. Some prefer it disabled so it has less of an impact on game FPS.
 
If you have the GPU to spare, enabling will give higher quality. Some prefer it disabled so it has less of an impact on game FPS.
Ah, I see, thanks for explanation! Thankfully my current streaming settings don't impact in-game framerate in any significant way.

By the way, what is this all about with YouTube stream keys? I use the default variable stream key but I can create additional ones for specific resolution/framerate/bitrate. Is there any point in using other than the default variable one? For what reason were they introduced? I'm trying to understand if they might be of any use for me. After reading the official entry page the only reason behind their introduction that comes to my mind is to make it possible for other users (let's say some friends) to stream at (as) my YT channel. Unless there's other reason I'm not aware of.
 
How does that matter at all for the streamer? Like why bring it even up? It is not in your controll. Also you won't stream in a lower resolution because of it.

Please read it in the context of the whole thread. On a former post it was asked wether it makes sense to upscale and send 1080p material encoded as 2160p (the idea seemed to be that yt allows/reserves more bandwidth for that). I argued that the additional bandwidth merely compensates for the risen resolution of the envelope, not for better encoding of the 1080p message within that 2160p hull.

Finally i argued that (as you said) we don't have control over screen resolution and algorithm on the player/client side. So the whole question and chain of arguments was if it makes sense to upscale and downscale for the hope that the encapsulated 1080p material would benefit in any way.

I don't think so.
- every up- and downscale brings in noise and degradation of the material,
- benefit from more bandwidth that way is - at least - doubtful.

I agree to a certain degree that there may be a benefit when YT switches to VP9. But if that prevails all the downside of the artificial complicated process? Finally i don't know.
 
So it turns out that the theory about video buffering issue is more of a myth than an actual thing?

In the case of Youtube and the downlink from YT's Content Delivery Network to the viewers: YES. A myth.
But there may be buffering issues on your side regarding your upstream capabilities, of course.

Recently i managed to stream over an self-managed RTMP server with no re-encoding, just re-distributing to the viewers. Then its totally up to you, because exactly the bandwidth you choose is the bandwidth the viewer must provide, too. If you send video with 4 or 8 mbps, every of your viewers must be capable to downstream the same amount. If some viewers have weak links, they will suffer from your "great" bandwidth.

So i experimented with (additional) on-the-fly downscaling and re-encoding (from 1080p to 720p with lower bandwidth) on my server (thus providing a second stream for weak viewers). The outcome is viewable for normal purposes (church transmission), but definitely not for fast gaming (where people tend to be unhappy with framerate below 60) . ^^ **grin**
 
Please read it in the context of the whole thread. On a former post it was asked wether it makes sense to upscale and send 1080p material encoded as 2160p (the idea seemed to be that yt allows/reserves more bandwidth for that). I argued that the additional bandwidth merely compensates for the risen resolution of the envelope, not for better encoding of the 1080p message within that 2160p hull.

Finally i argued that (as you said) we don't have control over screen resolution and algorithm on the player/client side. So the whole question and chain of arguments was if it makes sense to upscale and downscale for the hope that the encapsulated 1080p material would benefit in any way.

I don't think so.
- every up- and downscale brings in noise and degradation of the material,
- benefit from more bandwidth that way is - at least - doubtful.

I agree to a certain degree that there may be a benefit when YT switches to VP9. But if that prevails all the downside of the artificial complicated process? Finally i don't know.
I don't think you understand how youtube works. With increasing the resolution to atleast 1440p you get vp9 for all resolution that youtube delivers + that the stream that youtube delivers at 1440p and 2160 gives you not only more bitrate in total but higher bits per pixel ratio.
So your thinking is wrong because of lack of information.
 
By the way, what is this all about with YouTube stream keys? I use the default variable stream key but I can create additional ones for specific resolution/framerate/bitrate. Is there any point in using other than the default variable one? For what reason were they introduced?

They were mainly introduced due to some (former) faulty encoders. There is a world outside OBS. ^^ For instance i choosed to stream from a chinese drone a while ago. And some of the exotic encoders make false assumptions about resolution, bitrate aso. YT ingestion servers then may have problems to detect the right stream settings from the stream itself. With the specific streamkeys you could/can bind specific settings. In practice it only worked for the resolution for me, not even the framerate. Well then...
 
I don't think you understand how youtube works.
That is exactly what i admitted in former posts regarding inner secrets of VP9 (that i don't know). All other said should be right.
Thank you anyway. Do you have web resources/links regarding these background information? I would be pleased to...
 
For high popularity channels youtube enables vp9 codec for all resolutions. I am at 4.5k subs and still have to upscale from 1080p to 1440p to get good video quality, otherwise high motion + detail scenes become "pixelated".
While upscaling makes video blurry (i use lanczos 32 samples), when downscaled in web video player image becomes very sharp, while native 1080p videos become blurry due to low bitrate avc1 codec.
 
While upscaling makes video blurry (i use lanczos 32 samples)
I don't know what content you stream but is it some gaming streams? If so, wouldn't Bicubic filter be more appropriate for it? I have already mentioned the matter of the downscale/upscale filters on the previous page. Since that time i have done some more reading across the forum (and outside of it) and I've found out that Bicubic filter is more suitable for gaming streams than Lanczos. I'm curious what are opinions about it. I've done some comparison of Bicubic and Lanczos on my own stream but I can't make a decision which one is 'better' as I don't see much difference.
 
I just compared image quality of bicubic 16 samples vs lanczos 32 samples and didn't notice a difference. My screen is not 1440p though, so i used a video player and set image zoom at 100%.
 
Back
Top