Wanted to test some heavy games, move-full or not with Crysis 3 or Witcher 3 included.
Encoder turned on, use OBS studio AMD plugin by preference(On settings, put them from "simple" to "advanced", then use "AMD advanced media encoder" "high quality" or "VBR with 90000 average bitrate" on recording, on hotkeys put two unassinged combinations like CTRL+* and CTRL+/ to start recording and stop recording options on the first TAB). Record on games and do a FRAPS bench on it. Check the video quality if it matched the desired result. FPS hit must be at 10% max versus normal FPS, plus video quality must be great as selected.
2016-12-09 16:22:11 - crysis3 no recording
Frames: 8909 - Time: 124031ms - Avg: 71.829 - Min: 45 - Max: 86
2016-12-09 16:25:24 - crysis3 obs-vbr-quality-90mbit
Frames: 8594 - Time: 126406ms - Avg: 67.987 - Min: 48 - Max: 91
2016-12-09 16:31:52 - crysis3 obs-cbr-balanced-50mbit
Frames: 8800 - Time: 123687ms - Avg: 71.147 - Min: 48 - Max: 87
2016-12-09 16:52:58 - crysis3 obs-cbr-balanced-30mbit
Frames: 8809 - Time: 128391ms - Avg: 68.611 - Min: 49 - Max: 84
2016-12-09 17:03:00 - crysis3 amd-relive
Frames: 8901 - Time: 122171ms - Avg: 72.857 - Min: 46 - Max: 93
I was surprised to see such small performance hit. Relive does not seem to affect it at all. Dafuq?
One of the key advantages AMD claims it has over all the others is that being a GPU manufacturer means they can interact with the hardware in ways the others cannot, leading to a much lower overhead and better experience than the rest.
Hello.
So I did some tests. High settings, 1080p, 7950@1.1Ghz, Crysis 3, the first couple of minutes of the "Root of all Evil" mission. You can see the path at the below videos.
Here are my benchmark results.
I was surprised to see such small performance hit. Relive does not seem to affect it at all. Dafuq?
As for the videos, the latest version of OBS Studio did not work out of the box. I had to download the latest version of the AMD plugin labeled for Relive usage.
crysis 3-7950-obs-vbr-quality-90mbit
crysis 3-7950-obs-cbr-balanced-30mbit
crysis 3-7950-amd-relive recording
The obs-cbr-balanced-50mbit has no video because nothing was recorded.
The obs-vbr-quality-90mbit only managed 44.41fps recording speed. It did jump up to 75mbits from what I saw in pot player however.
The obs-cbr-balanced-30mbit does report 59.24fps speed, but the video is very jerky. I am not sure what went wrong here.
Out of all, only the AMD Relive managed a decent recording with no performance hit too. This one reports 59.94fps recording speed and 30mbit steady bitrate.
I tried to record with GameDVR too, but it does not come up when I press winkey+G. It seems like the screen flashes momentarily when I press winkey+G. Maybe not compatible with Crysis 3?
Oh and I was thinking of converting the vids for my channel to HEVC to save some uploading time, lol.
1mbit upload here, I feel you brother. Soon to be upgraded to VDSL2 however, thank God.
Thank you very much, bro! Performance hit with GCN1 is little as expected, even with OBS VCE! As my testing did it in the past, too!
And yeah, the eng was really right. No other recorder could make captures with Polaris VCE. OBS gives more options(especially on bitrate), AMD should help the Plugin maker.
Uploaded the capture on HEVC yesterday, youtube rejected it(by being HEVC) after 100%(1.34GB) done. This thing bothers very much when you have a 600kbps upload speed. This night i give the new video link.
Why did YouTube reject ? Bad codec? or wrong bit-rate ?
also Xyamer(?) updated OBS.Try again.
aaaaaand mouse corruption bug is back, so enable pointer trail at minimum.
You have any idea of what could be causing this?I uploaded a video comparison sometime ago for BF4 scene and for a scene from a movie to check quality difference. I used 4Mbit for recording waiting Youtube H264 to reencode at this same bitrate, problem than youtube dont always reencode at 4Mbit, if the video complexity is low it will use lower bitrate. I used slow motion to my video to avoid Youtube h264 decrease quality but it caused the low complexity aswell. Problem than Youtube sometimes reencode VP9 versions to display to users and VP9 versions has lower bitrate and different type of image quality smoothing/bluring video image making it different from original source.
The good point is than VP9 smoothed blocky, making the video playback look better. The bad is than my video was done to show the blocky and artifacting for comparison purpose then VP9 version make it totally worthless. Depending on device/browser Youtube will display H264 or VP9 version for viewer, but there is an addon for Chrome/Firefox which force Youtube play H264 version. The addon name is h264ify. For check which codec you seeing when playaing a video in Youtube just right click video->stats for nerds->Mimetype will show which codec is being played, avc1=H264 or vp9. Youtube only add VP9 version to videos which reach a specific amount of views in most cases, but sometimes happen to new videos, there is a randomly not well known.
Youtube support 48 and 50fps videos too, OBS and even Shadowplay/Win10DVR has problem producing a solid 60fps for specific heavy games. Shadowplay/Win10dVR in this scenarios has a 10% duplicating frames so the 60fps video has average of 6 frames duplicated while OBS is bit higher with 10-12 frames duplicated for 60fps. When i record with OBS at 48fps the duplication is very rare. Yet would be sure to ingame minimum fps be always higher than recording fps to avoid duplicating. Unfortunately Shadowplay/Win10DVR/Relive only support 30 or 60fps with only OBS supporting any type of FPS you want to do.
You have any idea of what could be causing this?
https://www.dropbox.com/s/9kab4dh3y0bgi0c/Witcher_3_lag.mkv?dl=0