unless you’re hitting a power limit, nvenc will basically not compete with the cuda cores at all, so it shouldn’t be too much different if you’re encoding something while playing a game
That kinda was my thinking when I started doing tests with Sekiro, and makes sense (from what I’ve read).
But then I started getting framedrop on video (not in game).
Maybe the card’s ram was overwhelmed?
Can’t really remember but I don’t think I was always recording to the HDD, and did some 1080p tests on the SDD.
Yeah, recording to my HDD kinda sucks, recording just gets framedrop unless I am doing it with a ridiculously low bitrate.
yup, I guess memory or disk bottlenecks could happen too in addition to power!
If your laptop is an Intel CPU, it should have Intel graphics with Quicksync. You’ll have to install the Intel graphics driver and then there may be some trickery to get it to cap your screen----while you are using the Nvidia card to actually play the game. *Although, I assume OBS has this figured out as a one click setting. I dunno.
Nop, I really don’t think I do.
Only the NVidia card:
BTW, I’ve already been able to do a couple of recording with MSI Afterburner, and I do admit is fast.
However, it’s giving me a few frame skips on the capture so far.
Also, I can’t import .mkv to my editor, but I guess there should be a MKV to MP4 re-container out there. I’ll get more into that eventually because .avi (only other container that MSI records into) can’t really deal with H.264 (or at least MSI says so).
So far I was only able to import MJPEG .avi on the editor.
I didn’t record under the best possible conditions, but I was getting about 40% frame skips when recording DS2 with MJPEG @ 95%.
I’ll have to play with this further. I do take my time but I will be thorough with it, and figure if I’m able to record with MSI properly.
I also do not have quicksync but it’s because I can’t get my computer to post with the iGPU enabled with my overclock
You have Intel HD Graphics 630
try setting MSI Afterburner to cap your framerate while you record. That setting is on the same page as the recording settings.
You can also give the recording 8 threads, if you have hyperthreading turned on.
Oh and you can use MKV2Vob to very easily change an MKV to an MP4. Its was the gold standard for getting TV show rips to play on PS3. Its still a great program.
I’ve been playing with Bandicam. As I said, you have to pay to remove their logo from videos. I didn’t pay. But, it looks like they give you most of the features for free.
As I said before about my past experience with bandicam: the captured video is very smooth. Also, gamma correction works with all of the codecs, unlike MSI afterburner. In fact, bandicam does it automatically.
So I’ve been comparing Bandicam’s MJPEG @ 90% quality Vs. Quicksync H.264, at constant bitrate of 40,000kbps (40mbps) (I compared constant bitrate 40,000 vs. quality based at 90% and the constant bitrate won fairly clearly)
For about 30 seconds of video capture:
Quicksync H.264 30fps/40,000kbp = a bit over 100mb of video. If you capture at 60fps, I would go ahead and double the bitrate to 80,000. You might be able to get away with 60,000. You’d have to test.
MJPEG 30fps @ 90% quality is indeed a tiny bit better quality, than 40,000kbps H.264 @ 30ps. However, the file size varies a lot, depending upon the scene (bright and colorful Vs. Dark scene with less colors). You can expect half a gig per minute, at 30fps and over a gig per minute, at 60fps. Keep in mind 2 things.
I couldn’t tell it was better quality than 40,000kbps H.264, without carefully comparing the two videos side by side. Darker scenes were slightly easier to notice the difference. For bright scenes, it was really tough.
this is still a “lossy” format. So, if you alt-tab between an MJPEG video file and the actual game-----the actual game is still going to look a little bit sharper.
Something else to understand here: I’m currently doing these tests with an Intel Ivy Bridge processor. Its from roughly 2013 and has the 2nd version of quicksync. Newer Intel processors have updates to quicksync, which should make it better quality. Possibly bridging the small gap on 90% MJPEG. There are also Nvidia and AMD’s own H.264 capturing, to consider as well.
I mention all this about quality vs. filesize because; if you were capping a couple of hours of gameplay, you are going to have to think about hard drive space. If you have a bunch of extra space, doing the MJPEG is probably worth it for the great quality and also because it loads into video editing software very easy and may also render out more quickly, after edits.
If you don’t have a bunch of hard drive space; H.264 40,000mbps @30fps and 80,000mbps at 60fps, are a very good compromise.
There are of course lossless formats. But those dump huge amounts of data and you need not only a lot of space, but a lot of throughput on your storage, to keep up. Based on past experiences, I would recommend platter drives in raid. Heck, I would probably even recommend raid for SSD, because many budget SSD aren’t great on sustained writes. Lossless usually also result in bigger hits to framerate, while playing a game.
New dark scene for Dark Souls 2. Capture with free logo version of Bandicam.
The game was running at 30fps, but I captured it at 60fps. 720p. Any frame drops are the game itself, dropping a couple frames below 30fps. Bandicam’s capture is very smooth.
720/60. Youtube totally chewed this one up. Holy Moly.
Same video, upscaled to 2K/1440p
As I mentioned in the past, Afterburner won’t let me gamma correct MJPEG, on my laptop. So, All of the videos I uploaded previous to today, were actually darker than they should be, with a slightly squashed colorspace, which gives more contrast to the color.
Bandicam has working gamma correction for me. However, this has highlighted another problem with Youtube: apparently they automatically brighten some videos? If you look at the dark scene video I posted above, the 720p/60 upload was brightened by youtube. and it looks really washed out, in addition to the really poor compression quality.
Well, I made a new video of Majula, with Bandicam. Its now gamme correct----and youtube also brightened this one:
I went back and looked at the video from Decinoge’s “friend” and it looks like his was also brightened.
But when I do the upscale and then upload, it comes out fine:
And I’m not sure why this is happening. The clips I did for Big Buck Bunny did not get brightened. But those were not made from video capturing. Those were just clipped from the official files provided by the website for Big Buck Bunny. There must be some sort of colorspace information which is causing this to happen. I mean youtube must be picky and flagging certain info to be brightened.
*I might actually ask youtube about this
**I think this may be due to a quirk with Bandicam’s version of MJPEG. As the upscale is converted to x.264. Due to this, it may be advisable to stick with H.264 for now. And just crank up the bitrates to 80,000kbps or more, to retain quality. Or just make sure you do an upscale in H.264, before uploading to youtube.
***ok I uploaded an 720p H.264 straight from bandicam and it is darker. But…maybe still a little bit off. Something is up with how youtube accepts video direct from bandicam. But if you process it through handbrake or an upscaler first, its fine. Hmmmmmmm!
New 4K/60 upload
Dark scene 4K/60
Now some screenshots to compare how favorably the 4K upscale on youtube, compares to the original 720p file I used to make these uploads. The darkest parts are still a little murky. But everthing else is pretty much as good or better.
screenshots from the original 720p/60 file
compare my right elbow in this pic
Screenshots from the 2K/60 upscale on youtube
screenshots from the 4K/60 upscale on youtube
That actually happened to me as well.
All those tests that I did, the H.264 were lighter on file, same for the MPEG4 one.
But the MJPEG were all darker. I then found out that is because of a render option on my editing program. That choice is available in every codec except MJPEG. So I have no other option with MJPEG than to render darker video.
However… once they are uploaded to youtube, there was no difference in color. For my case youtube actually “fixed” the issue.
Could it be that youtube itself does it’s own gamma correction when you upload the video?
Like…if the video is within YT standard values, it passes. Otherwise they apply the correction when processing the various resolution files?
Looking again into your videos/screen-shots actually makes me believe that even more.
You talked a lot about light changes, but everything we both upload that are about the same clips… the colors are just suspiciously to constant.
So yeah, I would kinda bet they do their own color and gamma correction just to keep a certain standard to their videos. If you are already within those, you will see no change.
They certainly compress colors a lot. There’s a big blue push, for one thing. And it looks like they change red to be more orange.
Blue and orange, heheh.
That would also do it.
Even a cut buy a bit at the bottom of the bit range would make a huge difference to the blacks.
Our eyes are very sensible to darks on screen, just pic up a photoshop and paint 10 squares with 10 different levels of light between 0 and 10 (up to 255).
You will notice them all.
Also I feel like this is a good and cool thread and should maybe be moved out of the Axe?
Yeah, already said that once.
@Toptube you should really do that change to the top topic.
I don’t really know where it should go, but @Felix was kinda involved itt, so he might have an opinion to share.
oh I didn’t know I had the power to do that.
I put it in “output”. That seemed to make sense to me. But anyone feel free to dispute that.