There was something that had been int he back of my mind for a long time. How do you properly video capture PC games? I know we have OBS and similar programs but those programs are fairly "recent" and I am not sure if they used back then.
You buy a nVidia card, and you use shadowplay. The end. 
Explaination: Nvidia has a hardware encoder for h.264/h.265 on their cards for a while now ( https://www.quora.com/Does-Nvidia-Shadowplay-use-GPU-encoder ), that can do this for you with little to no overhead (performance lost in games).
Needs to be activated in nVidia experience (the software), nut sure if enabled by default.
Once enabled software like OBS can hook into that function/API and actually use it for captures (set OBS to use nVidia hardware encoding, or shadowplay or something similar (whatever name they chose for that option)). Thats it - after that you are basically done.
(Maybe check black levels (so if black is black and not grey in your captures, or if you are crushing near blacks, because of a full/limited mismatch) but afair shadowplay is clever enough to deduce that from the graphics settings you set on your card, or taps into full RGB signalpath internally anyhow... Thats just an important step if you do video production. Probably more important with external capture devices (over HDMI). (I cant tell you the 100s of times, when professional video game review site Giantbomb got that wrong...) )
You might want to increase bitrate (unsure if you can do it, when using shadowplay), so the videofiles are better quality and bigger.
--
Why nVidia and not AMD? AMD has the same feature (called somthing differently), but historically - it sucked (videoquality usually). So on AMD systems, you might still want to capture not using HW accelleration, but CPU only (coming to think of it, some Intel CPU (unsure about AMD), with on chip GPUs might have a similar hw acceleration option build in, thats usually not as good as Nvidias, but also will cut down on processing overhead - so thats an option also (IF OBS f.e. supports it - unsure, because CPU doesnt do the games rendering) - handbreak f.e. does.).
--
That or you invest in external encoders, and record straight from the HDMI port. For PC stuff thats fine, for Consoles stuff you might need a chinese splitter, which first will help you disable HDMI copy protection ( https://de.wikipedia.org/wiki/High-bandwidth_Digital_Content_Protection ).. (Read Amazon buyer comments, most powered ones do.)
Explaination: Nvidia has a hardware encoder for h.264/h.265 on their cards for a while now ( https://www.quora.com/Does-Nvidia-Shadowplay-use-GPU-encoder ), that can do this for you with little to no overhead (performance lost in games).
Needs to be activated in nVidia experience (the software), nut sure if enabled by default.
Once enabled software like OBS can hook into that function/API and actually use it for captures (set OBS to use nVidia hardware encoding, or shadowplay or something similar (whatever name they chose for that option)). Thats it - after that you are basically done.
(Maybe check black levels (so if black is black and not grey in your captures, or if you are crushing near blacks, because of a full/limited mismatch) but afair shadowplay is clever enough to deduce that from the graphics settings you set on your card, or taps into full RGB signalpath internally anyhow... Thats just an important step if you do video production. Probably more important with external capture devices (over HDMI). (I cant tell you the 100s of times, when professional video game review site Giantbomb got that wrong...) )
You might want to increase bitrate (unsure if you can do it, when using shadowplay), so the videofiles are better quality and bigger.
--
Why nVidia and not AMD? AMD has the same feature (called somthing differently), but historically - it sucked (videoquality usually). So on AMD systems, you might still want to capture not using HW accelleration, but CPU only (coming to think of it, some Intel CPU (unsure about AMD), with on chip GPUs might have a similar hw acceleration option build in, thats usually not as good as Nvidias, but also will cut down on processing overhead - so thats an option also (IF OBS f.e. supports it - unsure, because CPU doesnt do the games rendering) - handbreak f.e. does.).
--
That or you invest in external encoders, and record straight from the HDMI port. For PC stuff thats fine, for Consoles stuff you might need a chinese splitter, which first will help you disable HDMI copy protection ( https://de.wikipedia.org/wiki/High-bandwidth_Digital_Content_Protection )..
(Read Amazon buyer comments, most powered ones do.)
Last edited by notimp,
Three main approaches, maybe with a sub type.
1) Internal capture.
Today Open Broadcast System aka OBS seems to be the reference standard for most. Time was there was a difference between normal desktop and 3d as 3d needed something extra.
As video encoding is one of the big things along with games to be a system grinder and you would be attempting both at once then some means were provided by external cards at one point and later graphics cards (at this point non graphics-sound-network cards are basically the equivalent of consoles with external devices a la the 32x, megacd, n64dd, kinect, PS eye, PS move... on consoles which is a bit of a pity as they had some promise) to try to do it in house. These were typically inferior as hardware video encoding then and only marginally less to this day has gone for the "cheap-fast-good, pick two and as fast is necessary and gamers are cheap..." approach to a truly notable degree -- back when H264 was rising up (X264 being the codec most used to make it) then the hardware accelerated encodings from graphics cards and whatnot was for the same bitrate probably comparable in metrics (PSNR is dubious but whatever) or plain visual even to the untrained eye to MPEG4-ASP which would have been known more popularly by the likes of xvid/divx (think your average AVI downloaded video, though they pushed things almost to breaking to stick with the 700 meg CD size so maybe not the most fair comparison).
Today a trained eye can still tell them apart but it is a lot better than it was, only took 15 years.
2) External capture.
Today this will mostly mean you build PC the second to take a split HDMI feed or whatever out from your main machine, do all its work there and spit it back at you/whatever streaming service you are rolling with whilst allowing your gaming machine to be busy playing games with no lag (give or take the basically zero of the splitter if you have even a half decent one, see also the nice perks of such splitters also maybe stripping HDCP protections if you are suffering such a device) and as much quality as it can muster.
You might however do something the console capture types do and have an external device (see various things from elgato) do the encoding for you. Can be a bit cheaper than a PC capable of doing such things in real time, though will probably take a quality hit vs what can be done by having a second PC.
3) The game itself has an internal capture format. You see this more in speedrun communities with various efforts from ID software (so mostly Doom and Quake) probably being the earlier and more noted examples. File sizes are tiny, and as it is a recreation of inputs and enemy places which is potentially a non random and systematic event then you can scale it as high as you like, and can do other things as far as speedrun analysis, pausing and whanot. Naturally this requires a game to be coded with support for it, or a game to have no random elements such that external capture/replay is possible (or maybe the game mode or external scripts to account for random elements). If there is random and choices made by the player scripts then we are into tool assisted speedrun territory which is somewhat a different matter.
1) Internal capture.
Today Open Broadcast System aka OBS seems to be the reference standard for most. Time was there was a difference between normal desktop and 3d as 3d needed something extra.
As video encoding is one of the big things along with games to be a system grinder and you would be attempting both at once then some means were provided by external cards at one point and later graphics cards (at this point non graphics-sound-network cards are basically the equivalent of consoles with external devices a la the 32x, megacd, n64dd, kinect, PS eye, PS move... on consoles which is a bit of a pity as they had some promise) to try to do it in house. These were typically inferior as hardware video encoding then and only marginally less to this day has gone for the "cheap-fast-good, pick two and as fast is necessary and gamers are cheap..." approach to a truly notable degree -- back when H264 was rising up (X264 being the codec most used to make it) then the hardware accelerated encodings from graphics cards and whatnot was for the same bitrate probably comparable in metrics (PSNR is dubious but whatever) or plain visual even to the untrained eye to MPEG4-ASP which would have been known more popularly by the likes of xvid/divx (think your average AVI downloaded video, though they pushed things almost to breaking to stick with the 700 meg CD size so maybe not the most fair comparison).
Today a trained eye can still tell them apart but it is a lot better than it was, only took 15 years.
2) External capture.
Today this will mostly mean you build PC the second to take a split HDMI feed or whatever out from your main machine, do all its work there and spit it back at you/whatever streaming service you are rolling with whilst allowing your gaming machine to be busy playing games with no lag (give or take the basically zero of the splitter if you have even a half decent one, see also the nice perks of such splitters also maybe stripping HDCP protections if you are suffering such a device) and as much quality as it can muster.
You might however do something the console capture types do and have an external device (see various things from elgato) do the encoding for you. Can be a bit cheaper than a PC capable of doing such things in real time, though will probably take a quality hit vs what can be done by having a second PC.
3) The game itself has an internal capture format. You see this more in speedrun communities with various efforts from ID software (so mostly Doom and Quake) probably being the earlier and more noted examples. File sizes are tiny, and as it is a recreation of inputs and enemy places which is potentially a non random and systematic event then you can scale it as high as you like, and can do other things as far as speedrun analysis, pausing and whanot. Naturally this requires a game to be coded with support for it, or a game to have no random elements such that external capture/replay is possible (or maybe the game mode or external scripts to account for random elements). If there is random and choices made by the player scripts then we are into tool assisted speedrun territory which is somewhat a different matter.
Also Windows 10 has its own capture capability nowadays, but its CPU bound. You can access it via the gamebar (thing that comes up, when you press the Xbox button on a connected Xbox controller.
The only thing you need to do their is find its options and remove "also capture game cursor" - for most games. Afair you have to raise quality from default (unsure, but I'm sure I modified it), and then its good to go, to be activated from the controller.
If you only need occasional captures, and dont mind the CPU overhead, this is a good and easy option as well. (For the beginning. Not if you want to do more with it other than to upload or keep it.)
The only thing you need to do their is find its options and remove "also capture game cursor" - for most games. Afair you have to raise quality from default (unsure, but I'm sure I modified it), and then its good to go, to be activated from the controller.
If you only need occasional captures, and dont mind the CPU overhead, this is a good and easy option as well. (For the beginning. Not if you want to do more with it other than to upload or keep it.)
Back in the day people used FRAPS or HyperCam (ew)
FRAPS probably still works fine but it was made for simpler times when realtime h264/h265 encoding was not really possible so the video was stored uncompressed and the files were huge. You also didn't really have any options to overlay graphics on the image or mix together multiple sound sources and it didn't support capture cards. It's much nicer with an all-in-one program that does everything like OBS and Xsplit.
In OBS you would use "Game Capture" for fullscreen games and "Window Capture" for windowed mode games or non-games. "Game Capture" also works for windowed mode games I think but it injects code into the game so requires running OBS as admin and might cause issues with certain anti-cheat software or DRM.
Hardware encoding using the GPU is an option these days which wasn't really possible back then, but with modern CPUs having 8 cores or more usually you can go with software encoding and get a higher quality image at the cost of higher latency (which does not matter for broadcasting/recording) since games rarely use that many cores.
Intel's QuickSync hardware encoding is supposedly really good, it's higher quality than AMD or Nvidia's though Nvidia's has improved a lot too so they are closer now. The drawback of QuickSync is higher latency, doesn't matter much for recording or broadcasting but does matter for game streaming with Steam, Parsec etc.You buy a nVidia card, and you use shadowplay. The end.
Explaination: Nvidia has a hardware encoder for h.264/h.265 on their cards for a while now ( https://www.quora.com/Does-Nvidia-Shadowplay-use-GPU-encoder ), that can do this for you with little to no overhead (performance lost in games).
Needs to be activated in nVidia experience (the software), nut sure if enabled by default.
Once enabled software like OBS can hook into that function/API and actually use it for captures (set OBS to use nVidia hardware encoding, or shadowplay or something similar (whatever name they chose for that option)). Thats it - after that you are basically done.
(Maybe check black levels (so if black is black and not grey in your captures, or if you are crushing near blacks, because of a full/limited mismatch) but afair shadowplay is clever enough to deduce that from the graphics settings you set on your card, or taps into full RGB signalpath internally anyhow... Thats just an important step if you do video production. Probably more important with external capture devices (over HDMI). (I cant tell you the 100s of times, when professional video game review site Giantbomb got that wrong...) )
You might want to increase bitrate (unsure if you can do it, when using shadowplay), so the videofiles are better quality and bigger.
--
Why nVidia and not AMD? AMD has the same feature (called somthing differently), but historically - it sucked (videoquality usually). So on AMD systems, you might still want to capture not using HW accelleration, but CPU only (coming to think of it, some Intel CPU (unsure about AMD), with on chip GPUs might have a similar hw acceleration option build in, thats usually not as good as Nvidias, but also will cut down on processing overhead - so thats an option also (IF OBS f.e. supports it - unsure, because CPU doesnt do the games rendering) - handbreak f.e. does.).
--
That or you invest in external encoders, and record straight from the HDMI port. For PC stuff thats fine, for Consoles stuff you might need a chinese splitter, which first will help you disable HDMI copy protection ( https://de.wikipedia.org/wiki/High-bandwidth_Digital_Content_Protection )..
OBS can take advantage of it, Xsplit probably can too, and it's a good option since the integrated graphics are otherwise not being used, so it can be dedicated to hardware encoding and none of the primary GPU's power is being used for it.
Last edited by The Real Jdbye,
- Joined
- Feb 6, 2019
- Messages
- 3,170
- Trophies
- 1
- Age
- 25
- Location
- Ecuador
- Website
- saulfabreg-wiivc.blogspot.com
- XP
- 7,734
- Country
Xbox Game Bar's integrated DVR function (for record games of your PC) is the best selection if you have a Windows 10 PC.
Another option is Bandicam (https://www.bandicam.com/en), Camtasia (https://www.techsmith.com/video-editor.html) or Fraps (https://fraps.com/).
Another option is Bandicam (https://www.bandicam.com/en), Camtasia (https://www.techsmith.com/video-editor.html) or Fraps (https://fraps.com/).
- Joined
- Jun 21, 2018
- Messages
- 552
- Trophies
- 1
- Location
- my villa
- Website
- vilagamer999.github.io
- XP
- 1,720
- Country
ew no, please don't use those, the windows 10 recorder is honestly fine but bandicam and camtasia suck big time, they are both shareware and also terrible in terms of functionality and features, even though it's not very user friendly at the beginning OBS is the best screen recorder by far, like it is kilometers ahead of the competition and it is also free, if it wheren't I would honestly pay full price for it plus extra.
- Joined
- Feb 6, 2019
- Messages
- 3,170
- Trophies
- 1
- Age
- 25
- Location
- Ecuador
- Website
- saulfabreg-wiivc.blogspot.com
- XP
- 7,734
- Country
I agree @Vilagamer999. Since Bandicam and Fraps (and Camtasia) are too bad for recording my Dolphin Emulator gaming i decided to clearly use the integrated Xbox Game Bar DVR function (for record screen and games) on my Windows 10 PC.
Similar threads
-
sp3off
-
K3Nv2
-
DinohScene
That cat suit twink -
Veho
-
-
-
@
RedColoredStars:
Right. But they could create a ticket that gets forwarded to the proper team. I think Nintendo would care that their videos aren't working on a major browser -
-
-
@
RedColoredStars:
Dont know why FF is slower on my pc either. Tried all the "speed up firefox" tricks but notta.
-
-
-
-
@
sp3off:
Chrome is like "BRRRRRRRR" but I dislike the fact it's a powerhouse, and the fact that Manifest v3 exists
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
