Author Topic: Fraps performance hit?  (Read 3681 times)

Offline Krusty

  • Radioactive Member
  • *******
  • Posts: 26745
Re: Fraps performance hit?
« Reply #30 on: December 07, 2011, 02:30:55 PM »
That's not correct. I've recompressed FRAPs files by STREAMING the video (no changes in compression) or by choosing even very generous compression settings, but setting it to use MP3 audio. I've turned 3GB files into 300MB. I kid you not, recompressing to use MP3 CODEC on my FRAPS holding folder saved me something on the order of 100GB (it's a 1TB drive, I store a lot).


EDIT: You chose a 2 minute clip, though... I think you'll notice more savings on the longer clips, naturally, when you compress in MP3.
« Last Edit: December 07, 2011, 02:33:06 PM by Krusty »

Offline MrRiplEy[H]

  • Persona Non Grata
  • Plutonium Member
  • *******
  • Posts: 11633
Re: Fraps performance hit?
« Reply #31 on: December 07, 2011, 03:54:36 PM »
That's not correct. I've recompressed FRAPs files by STREAMING the video (no changes in compression) or by choosing even very generous compression settings, but setting it to use MP3 audio. I've turned 3GB files into 300MB. I kid you not, recompressing to use MP3 CODEC on my FRAPS holding folder saved me something on the order of 100GB (it's a 1TB drive, I store a lot).


EDIT: You chose a 2 minute clip, though... I think you'll notice more savings on the longer clips, naturally, when you compress in MP3.

Yep Krusty is right - uncompressed audio stream is a killer. I once broke my nerves when I edited a home video of my friends bachelor party. No matter what I tried the file size was gigantic and I only had a limited space on the web host. Only after deploying the video I stumbled into the audio setting only to see that no compression was done on audio - and the file size went down 80% when I switched it to mp3.

I assumed that any video encoding tool would compress also audio by default - obviously I was dead wrong.
Definiteness of purpose is the starting point of all achievement. –W. Clement Stone

Offline BoilerDown

  • Silver Member
  • ****
  • Posts: 1926
Re: Fraps performance hit?
« Reply #32 on: December 10, 2011, 09:12:30 PM »
That's not correct. I've recompressed FRAPs files by STREAMING the video (no changes in compression) or by choosing even very generous compression settings, but setting it to use MP3 audio. I've turned 3GB files into 300MB. I kid you not, recompressing to use MP3 CODEC on my FRAPS holding folder saved me something on the order of 100GB (it's a 1TB drive, I store a lot).


EDIT: You chose a 2 minute clip, though... I think you'll notice more savings on the longer clips, naturally, when you compress in MP3.

Then 99% of those size savings came from compressing the video.  If you look at the properties of any video file in Media Player Classic, click the MediaInfo tab, then you can see how much of the file is taken up by video, and how much is audio.  I can't find any video files on my hard drive where the audio is more than 6% of the total size.  For Fraps files they are all in the 1% range, often rounded down to 0%.  The audio component is literally insignificant compared to video, and this is an indisputable fact to anyone looking at the actual numbers and thinking mathematically.

In response to your edit, the length of the video file doesn't matter because its a ratio.  The only thing that could matter is if you're recording video at really low quality settings, and audio at really high quality settings.  That doesn't seem to make much sense though, and the amount of audio still wouldn't be significant.

Compressing raw Fraps video files to save space is fine, but you lose quality.  Those savings are mostly not coming from getting rid of the audio, because there isn't that much audio to get rid of.

If you know you're not going to keep the audio, you can avoid this whole discussion and just uncheck the "Record sound" setting in Fraps.  Then everyone here will agree that your entire Fraps file consists of nothing but video, and nothing is being wasted on audio.


Edit:

The best practice for the highest quality end-product is to keep everything in the highest quality format until the end, when you render a video file.  If you transcode your Fraps files and then use the transcoded files as your source files for your video project, it will look worse than if you kept the raw Fraps files to work from.  

Furthermore, and you perform this experiment easily yourself if you wish, its been my experience that options that reduce Fraps capture quality, such as "Half-size" recording, are not worth the reduction in quality.  You'd think the files would be 1/4th the size, but they are much bigger than that.  Same goes for "Force lossless RGB capture".  I always enable lossless capture, because it looks far better and is only marginally larger on my hard drive.

Going back to my previous statement, "If you transcode your Fraps files and then use the transcoded files as your source files for your video project, it will look worse than if you kept the raw Fraps files to work from."  If you must save space and cannot keep the raw files, you're better off capturing full-size lossless, then transcoding down, than you are capturing half-size and lossy.  You definitely will lose a lot of quality if you both capture half-size and/or lossy, and additionally transcode down.  You probably want to avoid that if you care what your video looks like.

One place you can safely save some size is by reducing the FPS.  I see no need at all to capture at over 30 FPS.  Furthermore, if you're in AH Film Viewer and capturing at half speed, you can halve your capture rate (to 15 FPS, for example) because when you double the speed again in your video editor, it will go back to being normal FPS.  This technique will not have an affect on your video quality, unlike the other things mentioned.
« Last Edit: December 10, 2011, 09:35:16 PM by BoilerDown »
Boildown

This is the Captain.  We have a lil' problem with our entry sequence so we may experience some slight turbulence and then... explode.

Boildown is Twitching: http://www.twitch.tv/boildown

Offline MrRiplEy[H]

  • Persona Non Grata
  • Plutonium Member
  • *******
  • Posts: 11633
Re: Fraps performance hit?
« Reply #33 on: December 11, 2011, 03:55:09 AM »
Then 99% of those size savings came from compressing the video.  If you look at the properties of any video file in Media Player Classic, click the MediaInfo tab, then you can see how much of the file is taken up by video, and how much is audio.  I can't find any video files on my hard drive where the audio is more than 6% of the total size.  For Fraps files they are all in the 1% range, often rounded down to 0%.  The audio component is literally insignificant compared to video, and this is an indisputable fact to anyone looking at the actual numbers and thinking mathematically.

I'm sorry but if the audio is recorded in wav format it will take typically 172 kb a second. You call that insignificant? :D In 5.8 seconds you'll be already at the megabyte mark. Coded in mp3 it's going to be around 16kb/s. Those fraps recordings you've seen must have either had no audio or you already had a codec in use. Wav audio is huge, believe it or not.
Definiteness of purpose is the starting point of all achievement. –W. Clement Stone

Offline BoilerDown

  • Silver Member
  • ****
  • Posts: 1926
Re: Fraps performance hit?
« Reply #34 on: December 11, 2011, 05:04:54 AM »
I'm sorry but if the audio is recorded in wav format it will take typically 172 kb a second. You call that insignificant? :D In 5.8 seconds you'll be already at the megabyte mark. Coded in mp3 it's going to be around 16kb/s. Those fraps recordings you've seen must have either had no audio or you already had a codec in use. Wav audio is huge, believe it or not.

Just look at the Fraps file statistics I posted.  The Audio is 23.1 MB.  Out of a 3.9 GB file.  That's far less than 1% of the total file size, even with your 172 kb per second figure.  Lets pretend that you really gimp your video quality and take a Fraps file that's 6x the length that I take.  So you get 6 minutes per 4 GB instead of 1 minute per 4 GB that I get.  This is an extreme case by the way which won't actually happen.  You're still only at around 5% of the file being used to store audio.  Conclusion: Still insignificant.

Come on people, take some math classes.  And stop trying to substantiate claims that are just wrong.
Boildown

This is the Captain.  We have a lil' problem with our entry sequence so we may experience some slight turbulence and then... explode.

Boildown is Twitching: http://www.twitch.tv/boildown

Offline MrRiplEy[H]

  • Persona Non Grata
  • Plutonium Member
  • *******
  • Posts: 11633
Re: Fraps performance hit?
« Reply #35 on: December 11, 2011, 08:15:38 AM »
Just look at the Fraps file statistics I posted.  The Audio is 23.1 MB.  Out of a 3.9 GB file.  That's far less than 1% of the total file size, even with your 172 kb per second figure.  Lets pretend that you really gimp your video quality and take a Fraps file that's 6x the length that I take.  So you get 6 minutes per 4 GB instead of 1 minute per 4 GB that I get.  This is an extreme case by the way which won't actually happen.  You're still only at around 5% of the file being used to store audio.  Conclusion: Still insignificant.

Come on people, take some math classes.  And stop trying to substantiate claims that are just wrong.


On regular cases audio compression makes a big difference. Compressing CGI is extreme because it requires either a special codec or very high bitrate. I'd suggest trying out radgametools, their video codec is designed for CGI streams and produces crystal clear images with high compression.

I've used the free radgametools (Bink) compressor for making software tutorial lessons where it reduces the file sizes exceptionally without loss in quality. Enough so that choosing the audio compressor makes a large difference in file sizes. Moving game image might be harder case but OTOH it is named radGAMEtools. http://www.radgametools.com/bnkmain.htm

Quote
Some Reasons Why Folks Love Bink:


Bink videos look amazing! Bink can scale its data rate from 1200 kps for 1280x720p videos down to 75 kps for Nintendo DS videos. Bink will always make the best possible video for your data rate.

Bink is completely self-contained - it needs no special system software, it needs no other audio codec, it needs no other surrounding architecture. Just one small library and you are good to go - there are no external installation or dependencies.

Bink is super, super fast. In some cases, up to 10 times faster than other modern codecs. It's fast enough to use for in-game videos, as well as cut-scene videos.
« Last Edit: December 11, 2011, 08:26:13 AM by MrRiplEy[H] »
Definiteness of purpose is the starting point of all achievement. –W. Clement Stone