EPT[G] Streaming with OBS (Open Broadcaster Software) - Page 57
| Forum Index > Tech Support |
|
Gladiator6
Sweden7024 Posts
| ||
|
Malpractice.248
United States734 Posts
 Any ideas? | ||
|
Shiros
United States11 Posts
| ||
|
Cyro
United Kingdom20344 Posts
You can use higher resolutions, but you need a lot of bitrate behind them to give a decent viewing experience in high motion scenes unless you want a blurry, blocky stream Presets are more relevant for encoding efficiency at lower resolutions too, so if you post your CPU and resolution you can probably get decent quality gains there too. I'd advise not randomly messing with it though because there's far too many streams out there with bad settings that using settings too CPU intensive to the level of negatively affecting viewing experience, a few recently who think they can stream 1920x1080@60fps at veryfast or faster with a stock i5/i7 2xxx/3xxx or worse, using Superfast and having awful encoding efficiency while maxing CPU anyway because Superfast uses something like 6% less CPU than veryfast while requiring more than double the bitrate for the same quality | ||
|
z0rz
United States350 Posts
Does anyone have any suggestions on how I should run the test? Is game source vs game capture, screen region vs monitor capture, window capture vs window capture fair? I'll be recording CPU usage and FPS during the recording and using identical presets/bitrate/resolution/etc, of course. | ||
|
Rannasha
Netherlands2398 Posts
On February 08 2013 01:52 z0rz wrote: Does anyone have any suggestions on how I should run the test? Is game source vs game capture, screen region vs monitor capture, window capture vs window capture fair? I'll be recording CPU usage and FPS during the recording and using identical presets/bitrate/resolution/etc, of course. Don't think there's a need to put the specific modes against eachother for the two programs. A single table/graph/list (however you want to present the results) with all 6 options would be the most informative, so people can easily see what the effect is of using a specific program / capture mode. I'd also recommend testing a number of different resolution/bitrate settings to see how the different programs / modes scale with different loads. If possible, monitor bandwidth usage to see how far programs exceed the specified bitrate during spikes (so that streamers know how much overhead to reserve). Of course, having YT clips of all test-segments would be beneficial so people can compare quality. | ||
|
Cyro
United Kingdom20344 Posts
Id take a single replay, play it through to pre-load the whole thing and then run through it multiple times in the same game session with different programs/settings (with one of the player cams), while using fraps benchmark, graphing FPS and/or frametimes. CPU usage, all other variables aside, should be almost identical, yet is quite variable even running the same benchmark multiple times in a case like this, so probably best to focus on game FPS/frametimes and nothing else for the sake of testing, unless there is a major problem or difference somewhere else. Id reccomend test at 1920x1080 @ 30fps, 960x540 @30, 60fps, you should be able to get a good read for performance at resolutions between from those two (a good min and max end) because performance loss seems to scale suprisingly linearly for me at least with resolution and framerate, at least for several capture methods Make sure to have aero on/off correctly, and for Gamesource you need the optimization checkbox enabled Ive been interested in doing something like this myself for a long time, but i was unsure how to collect and deal with data and its also a lot of time+stress, and i couldnt get OBS to give me a consistent FPS output recently, so i put it off. You shouldnt do X vs X, Y vs Y IMO, just throw them all on a graph together or something. Ive been told by people from both sides that they aparantly have amazing performance, but i didnt see that (i7 950@3.8ghz, gtx580 then gtx260, no notable performance changes with GPU switch) For me right now, @1920x1080, 40fps, the best performing option cuts my FPS by about 35-40% (OBS game capture) which beats Xsplit screen capture by a decent margin - Gamesource at high resolutions is a wreck, but at 960x540, 30fps i lose only about 5% of my framerate with 1.2 optimization checkbox Ive heard of people getting worse or aparantly better performance, but i dont think i have actually seen solid documentation of better performance yet, so will be interesting to see Always have "reduce mouse lag" checked, and you should bench on low settings. Note your system specs, too Gamesource in 1.2.1301.2501 performs a lot worse in dx9 games for me than it does in 1.2.1301.1501 aswell, i use that version currently | ||
|
Junichi
Germany1056 Posts
| ||
|
Ekital.
United States15 Posts
| ||
|
R1CH
Netherlands10342 Posts
| ||
|
scottyyy
United Kingdom796 Posts
In TF2 I maintain 120 fps with both gamesource and screen region, but it just feels way worse with gamesource for whatever. More microstutter and mouse lag. | ||
|
z0rz
United States350 Posts
I do agree that different resolutions and FPS should be tested as well (yikes.. this is going to be more recording than I initially thought). Should I use the same bitrate for the different resolutions? I'm usually limited to about 2000kbps but I might have to lower it to really be sure there are no dropped frames in the tests. Should I use, say, 1750 for all resolutions or 1750 for high resolution and 1250 for low resolution? I don't think it really matters, but maybe someone is interested in those results. | ||
|
slytown
Korea (South)1411 Posts
Specs: i7 @ 3.4 ghz 4gb RAM Network: 33 down/2.3 up Settings expected: 1280x720, 30FPS, Quality of 8, 1800bits, medium x264, AAC 192 In-Game I tried max settings, medium, and low. All still get jitter. Any suggestions? | ||
|
Junichi
Germany1056 Posts
On February 08 2013 08:09 R1CH wrote: Yes, laptops are a big problem because games / OBS / etc can all be running on separate GPUs. Laptops aren't very good for this reason. Yeah thought as much, but I wanted to have confirmation from someone who knows more about it than I do. Thank you!  Gotta get money and buy a desktop pc then. | ||
|
Cyro
United Kingdom20344 Posts
On February 08 2013 08:07 Ekital. wrote: I'm actually able to stream with OBS at sub-par upload speed of only 500 kbps with a decent quality. It's so much better than xsplit. Same encoder I think CPU load measurements might be relevant here since Jim has said OBS is coded to make use of the GPU while XSplit might not Thats mostly relevant for game performance though AFAIK and not being able to reach higher encoding settings. Its hard to measure because CPU load varies so much (ive ran the same benchmark twice and got as much as 20-30% load differences, as well as different utilization of cores) and in terms of quality output, you would really have to use command line for at least crf because neither program has a transparent way of setting ratecontrol Settings expected: 1280x720, 30FPS, Quality of 8, 1800bits, medium x264, AAC 192 In-Game I tried max settings, medium, and low. All still get jitter. Shouldnt be using Medium @1280x720, 30fps on a stock 2600k for best results unless you are doing some pretty in depth tests and recording info like individual core load % every second and simulating max CPU load while in game and encoding, what do you mean by jitter though? | ||
|
felisconcolori
United States6168 Posts
| ||
|
R1CH
Netherlands10342 Posts
| ||
|
JakeBake
Canada27 Posts
After about 20-30 minutes, my stream on Twitch will become extremely choppy (audio and video) and will remain this way, or eventually freeze and/or go to a black screen. During all of this though, OBS is still displaying everything perfectly in the OBS client. When I stop/ restart stream, it comes back instantly, but crashes on Twitch.tv again after about 30 minutes; sometimes more or less. If anyone else has encountered this and/ or can help me out, please let me know. Thanks! | ||
|
RiceAgainst
United States1849 Posts
Build: + Show Spoiler + i5-2500 6GB DDR3 RAM HD7850 Internet: 5 Down/.9 Up VOD: http://www.twitch.tv/thericeagainst/b/365497156 Settings are: 750 kb/s and 750 kb/s AAC @ 128 USA West, Los Angeles (Secondary) Downscale to 1280x720 @ 60 FPS (native is 1080p) Faster preset Just wondering if I can improve the stream quality a bit, despite my limited internet. I'm playing on max settings and I'd like to keep it that way, as I plan on streaming HotS campaign/MP on max settings too. Problems might arise in ~200/200 armies (yet to test), so I might lower in-game settings but I prefer at least High graphics. Also, I can't seem to find a clear answer, but how do I alt-tab but let the game continue to stream? Can I use the screen switcher, and set the game to windowed? | ||
|
EleanorRIgby
Canada3923 Posts
| ||
| ||
