Question about streaming

LAwLz · December 25, 2015

And yet you fail to mention the 4 threads from this year and 2-4 from middle to late 2014. Still "newly implemented" right? And it doesn't matter what QS can do with video converting software, it's not possible to achieve that quality at 3500kbps to Twitch with the available streaming software. I even included screenshots this time of actual game streaming with the same frames done by a moderator on the OBS forums made in late 2014.

The most important claim of all : x264 has the best quality. Quicksync is a little better than nvenc, but no hardware encoder beats a good software encoder like x264.

Their claim : but Sandy/Ivy QSV is actually far inferior quality wise to software x264 encoding.

Their claim : The dual-PC method should still give you higher quality than QuickSync, just because x264 gives better quality video

Their claim : QuickSync with low bitrates is generally not going to give you a great picture. Hardware encoders trade quality for performance, which means you generally need to use more bitrate than with x264. (2014)

Their claim : Recording with quicksync allows you to throw enough bitrate at the quality difference to make up the quality difference vs x264. (2015)

Their claim : For quality per bitrate, x264 will pretty much always win based on the current state of nvenc and quicksync.(2015)

Their claim : QuickSync trades quality for performance. 2800 bitrate isn't even enough for a game like STALKER even with x264, a superior quality software encoder. The only thing that makes QuickSync better is more bitrate.(2014)

Their claim : Quality wise x264 is going to win outright, I can tell you that with 100% confidence. If you can manage to handle the Fast x264 preset, even better. (2014)

Their claim : Basically everything in this thread, late 2015

And the OBS guide, early 2015 : https://obsproject.com/forum/threads/high-cpu-usage-high-encoding-taking-too-long-to-encode-read-this-first.23334/

<image>

Oh okay now I get why you keep arguing against my test. You think I used "video converting software". No... The test I did shows EXACTLY the same info that would get sent to Twitch's server if you were to use Quick Sync vs x264 when streaming with OBS. The quality I showed at 3500Kbps, the quality you say it's "not possible to achieve at 3500kbps", is in fact exactly the same frames that would get sent to Twitch. I recommend you to back and read the explanation of how I did the test, because you must have missed parts of it.

There are no flaws in my comparison and it shows the quality difference you can expect between x264 veryfast and QS on a Sandy Bridge iGPU.

I don't see how any of your links disagrees with anything I have said. I have always said that QS will give you slightly lower quality. I do however see them disagreeing with you, who said that QS will never even come close to the quality x264 will have.

Please stop parroting what you have heard other people say when you don't fully understand what they are saying.

Samfisher · December 25, 2015

Oh okay now I get why you keep arguing against my test. You think I used "video converting software". No... The test I did shows EXACTLY the same info that would get sent to Twitch's server if you were to use Quick Sync vs x264 when streaming with OBS. The quality I showed at 3500Kbps, the quality you say it's "not possible to achieve at 3500kbps", is in fact exactly the same frames that would get sent to Twitch. I recommend you to back and read the explanation of how I did the test, because you must have missed parts of it.

There are no flaws in my comparison and it shows the quality difference you can expect between x264 veryfast and QS on a Sandy Bridge iGPU.

I don't see how any of your links disagrees with anything I have said. I have always said that QS will give you slightly lower quality. I do however see them disagreeing with you, who said that QS will never even come close to the quality x264 will have.

Please stop parroting what you have heard other people say when you don't fully understand what they are saying.

Only ONE quite mentions "slightly lower quality", every other post unanimously says you need to pump a lot of bitrate into QS to be close to x264, and the 4 screenshots provided prove that pretty clearly. Can you really not see the amount of macroblocking going on? And going back to my own videos, can you not see the difference in quality although it's not the same run, but recorded 2-4 minutes apart on the same map, with the same kind of creatures and detail? The QS video has a weird layer of something on it that makes every detail blurry. Just randomly pause anytime I attack and compare the roof tiles. Miles worse.

This is the same quality disparity in the source video too. Not sure if I even have those videos anymore as they were made when I had dual 7950's which I switched out for my 970. I'll do the same tests again when I'm back, currently working so don't have my desktop to do any new tests.

LAwLz · December 25, 2015

Only ONE quite mentions "slightly lower quality", every other post unanimously says you need to pump a lot of bitrate into QS to be close to x264, and the 4 screenshots provided prove that pretty clearly.

They say you need a higher bit rate to match the quality. You are the one adding "a lot" and "to be close", and those are the parts I disagree with.

If you had said "Quick Sync will have a lower quality" or "you will need a higher bit rate to make Quick Sync match x264 in terms of quality" then I would have agreed, but that's not at all what you said.

The QS video has a weird layer of something on it that makes every detail blurry. Just randomly pause anytime I attack and compare the roof tiles. Miles worse.

Maybe that's because you have not configured it correctly? I don't know what it might be but neither of my videos behaved like that so the only conclusion I can draw is that your configuration is the problem, not Quick Sync.

This is the same quality disparity in the source video too. Not sure if I even have those videos anymore as they were made when I had dual 7950's which I switched out for my 970. I'll do the same tests again when I'm back, currently working so don't have my desktop to do any new tests.

You don't have to do it for my sake. I don't trust that you are knowledgeable enough to make a fair and scientific test. Not after your statements in this thread.

Samfisher · December 25, 2015

They say you need a higher bit rate to match the quality. You are the one adding "a lot" and "to be close", and those are the parts I disagree with.

If you had said "Quick Sync will have a lower quality" or "you will need a higher bit rate to make Quick Sync match x264 in terms of quality" then I would have agreed, but that's not at all what you said.

Maybe that's because you have not configured it correctly? I don't know what it might be but neither of my videos behaved like that so the only conclusion I can draw is that your configuration is the problem, not Quick Sync.

You don't have to do it for my sake. I don't trust that you are knowledgeable enough to make a fair and scientific test. Not after your statements in this thread.

Only thing I changed between the two tests was the encoding option, nothing else.

And heh at your last sentence, I have more than enough knowledge to test it. There's nothing to be unfair about testing when I myself want a smoother and less glitchy stream, which is the whole point I did my own test in the first place when I started streaming. You though have not provided any evidence of QS being near x264 quality in a streaming setup, although the principles are the same as video conversion.

I've provided videos of my own and screenshots of actual stream snapshots provided by a 3rd party that shows the major quality difference. That amount of macroblocking does not go away until a huge bump in OBS bitrate.

LAwLz · December 25, 2015

You though have not provided any evidence of QS being near x264 quality in a streaming setup, although the principles are the same as video conversion.

I think my tests from page 1 is enough evidence since you know, it shows QS being near x264 quality at the same bit rate.

No, screen recording when you play a video is nothing like transcoding the video. The principles are completely different. Again, your lack of knowledge regarding the subject is shining through.

My test accurately depicts the difference between QS and x264 when recording with OBS. It is nothing like "video conversion".

Samfisher · December 25, 2015

I think my tests from page 1 is enough evidence since you know, it shows QS being near x264 quality at the same bit rate.

No, screen recording when you play a video is nothing like transcoding the video. The principles are completely different. Again, your lack of knowledge regarding the subject is shining through.

My test accurately depicts the difference between QS and x264 when recording with OBS. It is nothing like "video conversion".

Either way, your test does not use OBS or any other streaming software, there is more than sufficient proof that during streaming, QS isn't anywhere near as detailed as X264 and I have provided more than enough proof to back it up, from actually using OBS and from people actually using OBS. You however have not and therefore your posts are not only irrelevant but inaccurate when used with OBS.

If and until if you post your own OBS screenshot comparisons with the exact same bitrate and with minimal quality difference, you don't really have any say in this conversation. I've done my part, and so have the people on the OBS forum which you refuse to acknowledge. Whether or not my actual wording is "nowhere close" or "quite a bit worse", the screenshots do not lie from actual OBS users.

LAwLz · December 25, 2015

Either way, your test does not use OBS or any other streaming software, there is more than sufficient proof that during streaming, QS isn't anywhere near as detailed as X264 and I have provided more than enough proof to back it up, from actually using OBS and from people actually using OBS. You however have not and therefore your posts are not only irrelevant but inaccurate when used with OBS.

If and until if you post your own OBS screenshot comparisons with the exact same bitrate and with minimal quality difference, you don't really have any say in this conversation. I've done my part, and so have the people on the OBS forum which you refuse to acknowledge. Whether or not my actual wording is "nowhere close" or "quite a bit worse", the screenshots do not lie from actual OBS users.

Okay now I am 100% sure you don't actually read my posts. You just assume a bunch of things. MY TEST USED OBS!

Just check the meta data of the video files.

Writing application : Open Broadcaster Software v0.657b

You have provided jack shit in terms of proof to support your claim. You have posted 0 credible evidence to support it. All you have done is post one horribly flawed test you did a year ago, and then a bunch of posts from the OBS forum which does not even agree with you.

Posting my own OBS screenshot comparison with the exact same bit rate IS EXACTLY WHAT I HAVE FUCKING DONE THIS ENTIRE THREAD. How dense are you?

I do acknowledge the people on the OBS forum. They are saying the same thing I am saying (there is a quality penalty). They are not however, saying what you are saying (QS does not even get near the same quality as x264).

Samfisher · December 25, 2015

Okay now I am 100% sure you don't actually read my posts. You just assume a bunch of things. MY TEST USED OBS!

Just check the meta data of the video files.

You have provided jack shit in terms of proof to support your claim. You have posted 0 credible evidence to support it. All you have done is post one horribly flawed test you did a year ago, and then a bunch of posts from the OBS forum which does not even agree with you.

Posting my own OBS screenshot comparison with the exact same bit rate IS EXACTLY WHAT I HAVE FUCKING DONE THIS ENTIRE THREAD. How dense are you?

I do acknowledge the people on the OBS forum. They are saying the same thing I am saying (there is a quality penalty). They are not however, saying what you are saying (QS does not even get near the same quality as x264).

So you're saying the 4 screenshots provided above is not a great difference in quality? How blind are you? All your screenshots are low movement with nothing much going on screen, not indicative of many games, and definitely not of GTA V which the OP mentions. Static shots sure, QS may be comparable but with any sort of moderate to quick movement and/or any complicated effects like spell effects, explosions etc you get fed macroblocking up the wazoo.

LAwLz · December 25, 2015

So you're saying the 4 screenshots provided above is not a great difference in quality? How blind are you? All your screenshots are low movement with nothing much going on screen, not indicative of many games, and definitely not of GTA V which the OP mentions. Static shots sure, QS may be comparable but with any sort of moderate to quick movement and/or any complicated effects like spell effects, explosions etc you get fed macroblocking up the wazoo.

So you did not watch the videos I uploaded. Got it. I think we are done here.

No point in arguing with someone who doesn't even look at your posts.

Nice cherry picking with those 2 screenshots by the way. 6 screenshots and you pick the two that look the worst, and ignore for example the explosion on the house.

Samfisher · December 25, 2015

So you did not watch the videos I uploaded. Got it. I think we are done here.

No point in arguing with someone who doesn't even look at your posts.

Nice cherry picking with those 2 screenshots by the way. 6 screenshots and you pick the two that look the worst, and ignore for example the explosion on the house.

The explosion was a fairly small part of the screen and you CAN see the difference in quality especially on the dashboard etc, and sure, I give you that in that 1 screenshot it look fairly similar, and then there's 2 more totally crap ones.

And I DID watch your videos, and this is what I saw :

QS 1 : https://dl.dropboxusercontent.com/u/40347480/QS%201.png

x264 1 :https://dl.dropboxusercontent.com/u/40347480/x2641.png

QS 2 : https://dl.dropboxusercontent.com/u/40347480/QS2.png

x264 2 : https://dl.dropboxusercontent.com/u/40347480/x2642.png

QS 3 : https://dl.dropboxusercontent.com/u/40347480/QS%203.png

x264 3 : https://dl.dropboxusercontent.com/u/40347480/x2643.png

And you say that QS is nearly as good? Seriously now? Your own videos, same frames on both.