@pmikep

Hi Mike.

The encoding is being done on my CPU in software now that I've had time to look at it in more detail. My NVIDIA card is not the latest and greatest so can't take advantage of any hardware encoding OBS may have to offer.

The settings would even allow me to record at a much higher quality as well but frankly I can't see the point.

It has retained the settings I made and I've also been able to save it as a preset.

I've configured where the files are to be stored but can't find out how to change naming the created files. They are date and time stamped though so shouldn't be a problem and as I can change the files saved position could even create individual folders that could be named.

All in all, the best bit of freeware I've found in a long time short of vst soft synths.

Well pleased.

Ray.

Settings>Advanced lets you change the file naming.

@CubeAce, @pmikep

Hi Ray, Mike

Ray - you set ame off on a tangent. 😀

I have compared OBS version 25 to my normal screen recording software, Screencast-o-Matic (paid for) to see how much effect it has just recording working in MEP both programs do not put a significant load on the CPU or GPU nor cause issues with lag.

The installed space used is significantly different 427Mb vs 73MB, however OBS does have more functionality.

However, just for completeness and just in case the need arises, I also tested them to see what the effect would be while exporting to MP4 and got some unexpected results:

Exporting a 3m 03s project using the UHD 630, no transitions, 2 titles and an outro I got the following results:

1. No recording export took 2m 05s
    
2. Screencast recording - load increase on CPU was + 1 to 1.5% - export took 2m 25s
    
3. OBS recording using Default encoder (CPU) load increase on CPU was + 1 to 1.5% - export took 3m 05s
    
4. OBS recording using NVENC (New) encoder (RTX 2060) load increase on CPU was + 1 to 1.5%, RTX load 19% however was <2% - export took 3m 36s

Result 4 was a bit of a surprise I expected the loading to be significantly lower than when using the Default encoder (CPU).

CPU loading was very consistent for both recording programs

Memory usage for OBS and Screencast was, to all intents and purposes, similar 115 vs 127 MB.

The Screencast result does not surprise me - it records to a custom intermediate format which you can then export to mp4 after finishing recording - the export is quick..

John EB

That is surprising. I'm not a gamer, but reading some threads in the OBS forum, some gamers complained about OBS taking some frames during their gaming. This was supposed to be improved in OBS v25. (Which I see you tested.) The other trick is to run OBS as Administrator. Apparently Windows 10 has some priority handling routines to give Games priority, but does better when OBS is running as Admin. Perhaps this would help with MEP encodes too?

Also, I presume that you had set OBS to use the NEVC (new) encoder, as in my screenshot below.

P.S. Somewhere I remember seeing a setting where one could "release" the GPU from a program when not needed. Perhaps that was in Resolve. (I can't remember.) If I see it again and it's pertinent, I'll post here.

@CubeAce, @pmikep

Hi Ray, Mike

Ray - you set ame off on a tangent. 😀

I have compared OBS version 25 to my normal screen recording software, Screencast-o-Matic (paid for) to see how much effect it has just recording working in MEP both programs do not put a significant load on the CPU or GPU nor cause issues with lag.

The installed space used is significantly different 427Mb vs 73MB, however OBS does have more functionality.

However, just for completeness and just in case the need arises, I also tested them to see what the effect would be while exporting to MP4 and got some unexpected results:

Exporting a 3m 03s project using the UHD 630, no transitions, 2 titles and an outro I got the following results:

1. No recording export took 2m 05s
    
2. Screencast recording - load increase on CPU was + 1 to 1.5% - export took 2m 25s
    
3. OBS recording using Default encoder (CPU) load increase on CPU was + 1 to 1.5% - export took 3m 05s
    
4. OBS recording using NVENC (New) encoder (RTX 2060) load increase on CPU was + 1 to 1.5%, RTX load 19% however was <2% - export took 3m 36s

Result 4 was a bit of a surprise I expected the loading to be significantly lower than when using the Default encoder (CPU).

CPU loading was very consistent for both recording programs

Memory usage for OBS and Screencast was, to all intents and purposes, similar 115 vs 127 MB.

The Screencast result does not surprise me - it records to a custom intermediate format which you can then export to mp4 after finishing recording - the export is quick..

John EB

Duplicate post?

@johnebaker

I'm new to all of this screen recording lark. Some information takes more time to sink in than others.

I don't understand why the exports took so long on your system or am I misreading your post?

When I record MEP, putting MEP on my monitor 2 (NVIDIA) and have OBS on monitor 1 (Intel), I also have the folder of the destination files open alongside OBS. Then when the recording stops the fie appears in the folder pretty much immediately. That was on a 3 minute recording and I've bumped up the quality settings a bit so the end MP4 file is now 990MBs.

I wonder if it's a number of cores difference thing? For the encoder I'm using Hardware (QSV) whatever that is.

You can see all my settings below.

This is my complete set of settings within OBS.

I don't really know what I'm doing at this point so left a lot of the settings I didn't understand alone.

The end file is MP4 and ready to go.

@pmikep

Hi Mike.

No, I kept the profile to simple. See my screenshots above.

It's working brilliantly so I'm not likely to go poking around further 😟

Ray.

@CubeAce

Hi Ray

. . . . It's working brilliantly so I'm not likely to go poking around further . . .

To clarify:

It does work brilliantly if you are recording you working on a project as I stated in my first paragraph, it is not a significant load.

The 4 recording tests were carried out to see what effect OBS and Screencast had on the export times of MEP while MEP was exporting a project to MP4.

John EB

@johnebaker

Ah!

OK. Confused as to why you would want to, but understood.

One thing I have noticed is in the difference in file export sizes from OBS. Quality doesn't suffer that much when files are smaller for a given length of clip but there seems a huge difference in size per minute depending on what is being recorded. I know I left the recording quality up to OBS and the CPU loads etc vary depending on capture material but was surprised at the variation of final file sizes and bit rates.

Ray.

FWIW, I am rerecording my Noir movies with OBS as they rerun on the web. Previously I had recorded them at 720x480 with CL's ScreenRecorder. (Which, like Magix's Screen Recorder, came with CL's PD.) Now I record at 1080p using OBS's NVENC to h.264. Even tho picture size is larger and BW's about the same, the file size is approx the same for both! Amazing.

@CubeAce

Hi Ray

. . . . there seems a huge difference in size per minute depending on what is being recorded . . .

That is correct, for a relatively static scene the compression will be high - less data is needed to be able to reconstruct the image, for a scene with a more action (change) the compression will be lower - and needs more data to reconstruct the image.

The actual maths and physics for this is quite complicated - I would not advise you go go there it is a very big complex rabbit hole. 😱

HTH

John EB

@CubeAce

"For the encoder I'm using Hardware (QSV) whatever that is."

QSV = Quick Sync Video which is the hardware encoder on the Intel iGPU.

@johnebaker @Scenestealer@Scenestealer

I will try the same test later today after I've queued for my wife's prescription. I may be some time.

Peter's explanation may be the cause if your NVIDIA card is doing nothing which it shouldn't be as it should not be select-able using the screen you say you are using. I wonder if having more cores will make a difference as some of my cores show no use on export in MEP.

Ray.

@CubeAce

Ray - I am not quite sure what you are getting at but just note that you will not see the choice for a different hardware encoder as @pmikep has selected because your GT1030 does not have an NVENC chip on it.

@johnebaker

Ah!

OK. Confused as to why you would want to, but understood

I think John is interested to see if the recording software could upset the various loads on the hardware that could influence a performance comparison, if we posted that, showing our systems when encoding or during playing back.

Peter

@Scenestealer

Hi Peter. I looked at my last reply and it didn't even make sense to me 😕.

I was referring to Johns NVIDIA card but my CPU.

So, John is running his setup with MEP on the Intel powered monitor which on my system doesn't allow the use of my NVIDIA card at all. I can select my NVIDIA card on the Intel powered monitor but it doesn't take. It does if I put MEP on my NVIDIA powered monitor. Whether it just renders cross fades and doesn't do any HEVC encoding, it's doing something.

My second reference was to CPUs. My CPU has more cores than Johns I think and is split into more virtual cores but if the encoding from OBS is happening on the in the Intel GPU for an MP4 output then I should see the same increase in export time.

Sorry for the confusion. I was in a hurry to leave the house before the queues got too long at the chemist which at present is taking up long amounts of time.

Ray.

@CubeAce

Hi Ray

. . . . I wonder if having more cores will make a difference as some of my cores show no use on export in MEP . . . .

MEP like many programs will use the number of cores/threads it needs or is restricted to.

There is an interesting article here which has been echo'd many times on the Internet (as recently as last month).

I also read somewhere, cant find it at the moment, that there is a non linear curve where performance increase tails off as the number of cores increases.

John EB