I'd like to start a series of posts called TestRun, where I test out either hardware or software applications, benchmarking where I can with utilities (such as FurMark, CrystalMark, 3dMark) and just playing and testing when there are none (such as with Battlefield3, where at this time there is no repeatable, actual Benchmarking tool, as opposed to say, other stand-alone benchmarking game utilities such as Alien Vs Predator, Crysis, or even games with demo recording playback for repeatability, at this time BF3 does not have).
This is also a 'Quick' Edition, a version of the testing and results that are text-only for now, where I may do a full version as time permits in the future, with screen recordings, graphs and other materials that I would like to share; but for now, hopefully the testing and results alone will help people out.
[Links to updates and Video Editions of these tests can be found at the bottom of this article]
The Experiment
For this TestRun then, I am pitting FRAPS against other screen recording tools, Dxtory and Bandicam.
I wanted to include MSI's Afterburner with its capture utility, but at this time I could not get it working with my Gigabyte Mainboard and ASUS videocard. Other utilites such as ZDsoft's Game Recorder only allows 60-second recordings in it's trial version and I wanted to test longer recordings, since one of the problems many people are having (including myself) is with the audio and video going out of synchronization over longer game captures *coughfrapscough*. Dxtory and Bandicam allowed unlimited trial version recording (with large watermarks embedded) so they were perfect adversaries for my purchased version of FRAPS.
Dxtory and Bandicam are both newer competitors in the game recording market, showing up in 2009 and 2010 and as such, seemed to take more advantage of modern hardware processing capabilities. For instance, both Dxtory and Bandicam offer recording compression on-the-fly. You can even choose various compression CODECs (compressor/decompressors) such as MPEG, XviD, and more. What the compression allows, is much smaller file sizes for your captures.
Fraps does a very light compression through their codec, and results in large file sizes.
Dxtory and Bandicam have configurable settings for the codec and the compression amount (how much to process the recording as you go, how small to make the files that you create) - but remember that the more compression you attempt on the data, the more you are straining your hardware to process it as you go - meaning that the smaller the output of the files you are recording will be, but the slower your system will respond while you are playing the game and recording.
Bandicam allows you to record the operating system screen itself, not just within an accelerated game, but the Windows GUI for instance, so you can create Screencasts or tutorials, recording what you are doing on the screen.
Dxtory does not do this, it will only capture the DirectX/OpenGL acceleration within a game.
FRAPS will record the gpu-accelerated interface (such as AERO in Windows), so you can record the screen like Bandicam, if you wish.
The Test
For this TestRun, I used the Crysis Stand-Alone Benchmarking Utility, the Unigine Tropics Demo/Benchmark, played Diablo2 [warming up for D3!] and some multiplayer Battlefield 3 (returning to the same server in a short period of time), since that is a nice modern game that exercises your hardware well. I also tried recording the standard Windows GUI, reading some articles and watching some YouTube videos and recording what I was doing. I am currently running an AMD 6-Core CPU and an AMD/ATi Radeon HD 6870 on a GIGABYTE 990FXA chipset mainboard. I assume this is an average-to-above-average system at the time of this writing and it could help out anyone with a similar rig in the future as well, looking for ideas on what may work for better for them.
Keep in mind, that when considering the apparent effect of lag on the system, it can vary, not only between systems but also between games.
For systems, one person may get only a loss of a few frames, while another person can experience a much larger framerate loss, the difference usually more dependent on the system hardware.
For games, sometimes a company programs/codes the actual game on one type of hardware or the other, then 'ports' it over/translates it [the portions of it that were not compatible] for the other family of systems (such as Intel-->AMD or AMD/ATi-->NVIDIA), the result being games then 'prefer' one or the other and will natively perform more efficiently on the former (this explains why some games will perform better in Benchmarks and Testing on one brand of architecture than the other).
[I chose these three game recording applications for now, as opposed to including PlayClaw, Xfire and others, as perusing game forii, these three seemed to be "The Big 3" programs that people were always talking about, suggesting, and using in their game recordings. They also all allowed Unlimited Recording Times, since I already owned Fraps. In the future, I hope to do a larger TestRun of all of the currently available screen recorders I can find and create a nice Review of them all!]
The Data
FRAPS:
Game recordings from this Australian Heavyweight Champion, as everyone already knows, are huge. One minute of in-game recording at 1680x1050 [the current limit of my monitor at the time of this writing] was over 4.5 GB. That's a total filesize which will take up more than a DVD. For 1 minute. The data stream average for the video overall was about 550Mbps, meaning that it was writing data at about 70MB/s to the harddrive [on a drive that can handle up to 150MB/s]. That means, that recording an hour of gameplay (beginning a long session, or an average Conquest match in BF3) would be well over 250GB of data.
The thing with Fraps recordings however, is that the Quality is superb. It almost is WYSIWYG (What You See If What You Get), but the file sizes show that as well. You could not record days and days of game time and edit it all up at the end; at least not at Full Resolution, playing at 1080p or higher, unless you had the hard drive space to deal with it all.
Another issue with Fraps is that is very demanding on the system*. Because it is powerdumping huge amounts of data onto the drive, it strains the colon of the system bus with its girth, filling all bandwidth with video and audio data, taxing the CPU and RAM as it traffics it all, even though it is only very lightly processed.
End Result: your entire rig (especially if older) is slower in response to pushing this huge amount of data around and hence the many complaints about 'Fraps lags the game' on forums everywhere*.
*[Note however that this does not happen to everyone and not on all games]
I found that playing Battlefield 3, on a 64-player map, while recording with Fraps, my framerate drops from 70fps down to 30fps, even with "Lock Framerate While Recording"
not toggled on. The same thing happens with Skyrim. The resulting videos are smooth and look great though, they are just...big.
There is not much that you can configure with Fraps. You can change the framerate (to whatever your system can handle) and record in Full or Half resolution (for instance, if you are playing at 1680x1050, your recordings would be at 840x524), if your system cannot record at your full game size. You can also simply lower the in-game display to something smaller, like 1280x720 for easy editing and upload without having to resize if you are sharing your videos at that resolution anyway. this also helps if the strain is too much for higher resolutions.
[These two 'tips' above will also help anyone experiencing lag or slowdown or choppiness while recording during gameplay]
You can record multiple sound inputs, if you are doing a game commentary for example, and there is a built-in "benchmarking" utility, where it will keep track of the in/out/average frames per second and output that to a file, if you want to do some testing.
I purchased Fraps long ago, when it was the Go-To app for game recording and benchmarking and in many ways it still is; however through recent system upgrades (from a single core to a dual core and now up to a six-core processor) Fraps seems to have another problem - audio and video desynchronization: the sound doesn't match what you see on the screen, getting worse and worse out of time with the video the longer the recording is. As my older system (for instance, the dual-core) does not experience this, I assume it has something to do with the hardware and calls that the application is/is not making. Not being a developer for Fraps, I do not know exactly; but without fixes in future versions, many people (not all, only the ones experiencing this problem) are going to be looking for alternatives to this classic and wonderful veteran program, origin of the phrase "FRAPS, Or It Didn't Happen!"™.
DXTORY:
Dxtory is a challenger from Japan, touting faster movie capture and distributed hard drive writing (taking turns over different drives, so that it lightens the strain on all of them by sharing it, if needed) as it's battle weapons. Highly configurable, you can actually use most Codecs that you have installed on your system, finding the best one that may work for your setup. It comes with its own Dxtory Codec as well, a high quality adversary for the Fraps codec.
One minute of recording at 1680x1050, with the High Quality setting of the Dxtory Codec, came out to just over 5GB, averaging about 600Mbps. With the 'Compress' checkbox ticked, I assume this is the Fraps equivalent, because it looked just as good with the slight compression; but the file size was just as large as Fraps as well.
Dxtory has a couple of tricks up its sleeve however. For one thing, recording at that high a bitrate in game in Fraps, my framerate took a big hit. With Dxtory, it was a lot smaller: from about 50fps in a busy 64-player BF3 multiplayer area, down to about 40fps. And then to see the quality of the video file is almost the same as Fraps? Fantastic. The other thing is, you can specify different levels of quality when using the Dxtory Video Codec to save disk space and lower the strain on your system. For instance, trying the Dxtory Low Quality setting, the file created from a minute of gameplay was about 2.5GB, with an average bitrate of 330Mbps - about half that of the High Quality recording. Although it wasn't bad, you could tell that it was lower quality however, with some Chromatic Aberrations around text at lower resolutions [I may provide screenshots from the video in a future post]. It was good enough for recordings at 1050p or higher though, especially if you are resizing down to 720p for upload to a video sharing site somewhere.
You can also use other Codecs that you have installed on your system. I tried XviD, x264 and some others, but putting too much compression/small size restrictions on the recordings created more lag in the game and resulted in 'choppier' video output/playback.
[Although the built-in codec is great, I settled on installing the UT Video Codec (YUV-4-2-2) for it's balance of speed and size versus its great quality - and that's the one I would use if I were to purchase Dxtory.]
For more details/information on various codec settings (using as an example, H.264/AVC), I have written a more in-depth article, talking about the different settings (for MPEG-4 Part-10) and why you may or may not want to enable them, here (text only, "Long Version"):
http://gametipsandmore.blogspot.com/2013/05/game-recording-with-mpeg-4-using.html
Another weapon in the arsenal of Dxtory is its configurability. You are not restricted to just Full Resolution or Half, you can record in 75%, 50%, 25% and even other increments [I found 80% to be a good trade off between quality and file size for the default codec] or set your own resolution to record at - even while you play in another totally different rez, to help with reducing recording 'lag'. Nice. It also has multiple audio inputs (compressing the audio as well, if you wish) and timed screenshot-taking in multiple formats, if that's something you need. It also has it's own little 'Disk Benchmarking' tool, so you can test each drive and find out which one is the fastest to record with.
BANDICAM:
Bandicam is a contender from Korea that can record not only DirectX/OpenGL acceleration in games, but also the GUI itself or portion of it (such as recording what you are doing in Windows for tutorials). Another highly configurable program, you can change the resolution and quality level you are recording at (playing at one resolution while recording in another, as with Dxtory). You can choose a compressible the sound format and even embed your own logo as a watermark, with various sizes and transparency settings on it.
One minute of recording at 1680x1050 with its default setting came out to 0.1GB - you read that right, 100MB - with an average of about 15Mbps of data. Ten straight minutes of recording BF3 multiplayer came out to about 1GB. That means to record an hour of gameplay (an average Conquest match in Battlefield 3 or a Full Run of an Episode in Left4Dead) would take up only about 6GB!
[That also means, for the amount of space 1 hour takes up of a full-size Fraps recording, I could record over 40 hours of gameplay!]
I still can't believe it.
I immediately thought that the resulting file must look like crap on a cracker, but the video is actually comparable to both Fraps and Dxtory. It's not 100% quality, but then, the Default setting in Bandicam [at the time of this writing] was set at 80% Quality of MPEG-1, Variable Bit Rate ( which uses more bitrate when needed, using less when less is going on and not needed) encoding. The resulting video file did indeed look about 80-85% quality of a Fraps/Dxtory(HighQuality) recording, but that means it looked just fine, with clear text and textures, and it could further be turned up to 100% if you so desired.
['Videophiles' will notice some typical mpeg artifacts such as Gibbs Effects at lower resolutions/quality settings and some Temporal Recursive Noise and other compression artifacts when using MPEG-1 at low quality settings, but these can be alleviated by increasing the VBR percentage (to 90% or higher) or switching to recording in MJPEG [which is PlayClaw's recording codec of choice, for example, at the time of this writing] using a high quality setting (again 90% or higher) or once could even use the built-in YV12 codec if space is not an issue]
You can also record in YV12 or RGB, which give "FrapsQuality" recordings, but are also huge in file size to match (for instance, recording in YV12 colorspace, 1 minute of recording was about 5GB and 1 minute of RGB was almost twice that size).
[For filesize savings at a decent quality price, I think most people will be happy using [what originally was the program default of] the built-in Bandicam-optimized MPEG-1 codec at a high variable bit rate (80-100%) setting - and that's what I would use if I were to purchase Bandicam.]
The effect of apparent lag on the system was somewhere between Dxtory's 'Barely There' and Fraps's 'Omg I'm Gunna Lag', taking a 60fps Battlefield3 match down to about 40fps while recording. It was surprising to not have it affect the framerate more actually, considering the quality of the resulting video file, its small footprint the output file was going to have, and the processing that must be going on, behind the scenes to have that all happen. I also found that when recording the Windows interface, the resultant video was less choppy than a Fraps recording of the same thing (and with Bandicam you can choose to record only a portion of the screen, if you wanted to, for creating tutorials and such).
Bandicam is also configurable like Dxtory, so if recording at 1080p resolution, of 100% quality, at 60fps is a bit much for your system, you can turn down the quality (80% is the default and looks ok), or record at a smaller resolution while playing the game at a larger one, or record 'full-size' and just go and reduce the in-game resolution to something smaller, or you could turn off sound compression and many other things... but for the most part it works great as is, especially with a more modern desktop system.
With no audio/video sync problems during long recordings (tested up to 48 minutes) and with the additional option to record general Windows screen output when needed, this is a serious contender, especially for those who must have 'disk space usage during recording' as a consideration, as I do.
The Conclusion
Fraps vs Dxtory vs Bandicam?
I always like to say to everyone: try things out on your own, on your own system, and see which one is the best for you (they all have Demo versions available).
Perhaps you are more concerned with 'the best quality' and not concerned with 'hard drive space', then Fraps may be more for you - if your system can handle it. If you have an older system and are experiencing too much lag during recording, the configurability of Fraps to help, with options to turn down, are simply not there, so perhaps one of the other two, offering more configuration and recording options, is more your speed. Every system performs different and everyone has differing opinions on 'what looks good enough' for what they are doing; but, after testing all three choices of game capturing software personally, here are some overall considerations I can give with confidence:
Fraps
++
Excellent Quality (No other program records in as good quality by default)
--
Quality is not very configurable (aside from half-resolution recording)
--
Huge File Sizes
--
Large Performance Hit ([*not everyone] when recording, can lower to half-size to help perf.hit)
Dxtory
++
Great Quality (or Excellent or Low, depending on your settings and restrictions)
++
Quality is Configurable (compress with various codecs, compress audio, any resolution/size)
-+
Variable File Sizes (depending on settings, YV/RGB is possible which is similar to FrapsQuality/Size)
--
Small Performance Hit (when recording using UT Video YUV-4-2-2, otherwise Medium perf.hit)
Bandicam
++
Great Quality (or Excellent or Low, depending on your settings and restrictions)
++
Quality is Configurable (compress with multiple codecs, compress audio, any resolution/size)
-+
Small/Variable File Sizes (Smaller even at 100% Quality, YV/RGB is possible which is similar to FrapsQuality/Size)
--
Medium Performance Hit (when recording, can lower quality or size to help perf.hit)
Have fun testing and making your own choice!
Personal Short Version/Opinion:
- Fraps has great quality and always has, but the file sizes are just too big for me right now. I could record at half-resolution, but then the overall quality is too low in most games. And recently, there is a problem with the audio being out of sync with the video in the recorded files for me. Sorry Fraps, I have to look for something else until you get that fixed (or I do on my end).
- Dxtory has great configurability, I can have the quality of Fraps with the right settings, and without the lag that Fraps creates in some games! But then, recording with those settings, the file sizes would be just as big anyway. Since a concern of mine right now is disk space; sorry Dxtory, you're not my pick today.
- Bandicam, in it's default setting at the time of this post, does not have the exact quality of Fraps, but it can be made to have closer to the quality of Fraps by increasing the Variable Bitrate higher, or changing the compressor to MJPEG with a high (90-100%) quality setting. I could utilize the even better looking YV12, but then the file sizes would be just as big as Fraps... The MPEG-1 codec, as it's being used/optimized in Bandicam, especially with the VBR setting, only uses up disk space as it needs - and it compresses well even what it does use. The result is that it seems I can record everything, for hours and hours, and then I can pick and choose the parts later that I want to edit/keep. The quality of the Bandicam-optimized MPEG-1 (or Bandicam-optimized Xvid if you prefer to use that) at a high setting (80-100% VBR) looks fine for action games to me, with a few small nit-picky parts in areas where there is not much going on (flat color areas or slower games/movement), as is the nature of Variable Bit Rate recording. This can be fixed however, by filtering or resizing during editing - or it can not even be a problem at all by simply increasing the bitrate/VBR setting, or changing the codec to a high bitrate MJPEG setting [90-100% quality], which would improve things for even slow, non-action-oriented areas and games (like web browser games). Changing the codec to MJPEG also uses less overhead [performance demand/'lag'] and your recording will also be more compatible with editing programs and allow you to cut/start anywhere in the video and have it respond/seek much faster [I personally like the film-like look of MPEG-1 and how it compresses so well (small file size recordings); but the crispness, low overhead and compatibility with editing apps MJPEG offers is very useful, I am finding as I was doing this testing]. Being able to also record the desktop for tutorials, on top of all of that disk space I'll be saving, simply won me over. I am now a proud supporter (Registered User) of Bandicam.
See you in the games!
[ Update 1 : If you are looking for a TestRun with Video Examples of these programs and the quality that they produce when recording, the first Video TestRun comparison of them can be found at the blog here:
» Look for upcoming TestRuns, which will include PlayClaw's Newer Demo Version (that allows higher quality) and more Game Recording Apps - like Mirillis' Action, OBS, SmartPixel and more!
Update on this - a QualityTest/'Shootout'/Versus between ACTION, Playclaw 5 and Plays.TV Coming Soon™ in 2016!
[N.B.: I am not a developer for, nor affiliated in any way with any of these programs or companies]
