OK, back to the thread topic, but with a twist. This time I've just one Burnout Paradise capture of the fun Togue Sport Burning Route, though comparing Xvid and x264 compressions. As mentioned, I'm new to x264 as far as hands on compressing, so bear with me. Like I said elsewhere, consider this a preliminary test, as I've yet to master the use of it.
I kept the resizing of the 512x384 clips the same as before at 640x480 via Lanczos. As for why 4:3 that was mentioned earlier, it's not me, it's the game. You see I generally capture at 16:10 even though I have a 4:3 CRT, because every game I've captured up until now plays fine at that aspect ratio on my monitor. This one for some reason has lots of proportional distortion I could not adjust out via screen resizing. I've never seen the Takedown 4x4 look so silly, tall and skinny.
CRF was left at the default 20 and no change to qp_min, as my minimal reading of x264 indicated CRF is better and the default setting about right. I found the 2 Pass !Fast compression profile to be roughly identical to Xvid's twopass as far as time to compress, so that's the one I used. Target file size was set to 41 MB, which resulted in a pretty accurate 40.8MB. Conversely the calculator in VDub came out at 39MB, so I did some manual calculations, readjusted the bitrate, and made another compression.
For audio compression I used the LAME codec at 192kbps stereo, merely for the purpose of using the exact same codec for each video codec, so as to not skew the results. All other settings were left at their default values. I decided to make a 3rd compression just before posting this, to see if indeed x264 could have better image quality in a smaller file size. Though this time I tried the AAC codec at 160 vs 192kbps. I know, that's going against pitting video codec against video codec a bit and slightly skewing the results, but I was also wanting to compare VDub to AutoMKV, and that includes making use of some of it's features.
I tried a half size compression, 20MB, which was noticeably worse in quality, so I scrapped it. Something I should note though. I didn't actually complete that compression. When I started the encode process, I got an error saying a BAT file could not be found. In clicking OK it continued encoding anyway. The only difference was I used the default output file location, vs one I had designated before. I checked the partial compression to look at it's quality out of curiosity, and it was not as good.
I settled on a 30MB file size, which did not result in an error after starting the encode process. I had the feeling it was something about that 55MB or so Temp file contents the app produces upon each encoding. Everything else was left the same on the 30MB clip, only the target file size and audio was changed.
To my eye, the two 41MB clips are close, but the x264 one is noticeably better when zoomed. You can see more detail in the foreground, and no banding in the sky, as was evident in the Xvid clip. The nearby textures that quickly draw into view (trees, walls, etc) are also sharper in the x264 clip.
The 30MB clip however appeared to me to be a mixed bag. Whilst the sky is still sharper on the x264 clip without the banding the Xvid clip shows, the nearby textures drawing quickly into view are less sharp. You can see it if you look closely at the tree foliage and the brick walls of the tunnel. However amazingly the foreground including the pavement and the car's silhouette seem to be sharper in the x264 clip, despite being less than 75% the file size. Again, ALL these differences I'm noting are while the videos are zoomed full screen on a 20" viewable 4:3 CRT.
I will leave it for those of you whom download them to judge though. Some people see things I don't, and perhaps vise versa. Overall though I have to say x264 is the clear winner, even though this is just a prelim test done by a relative noob at using it. Though it doesn't seem as huge a difference to me as it sounds dopefish claims to be possible, I still can't get over how much I was mislead by the talk and sight of large file sizes with this codec. Perhaps some don't even know how to use it as well as this first timer.
Truth be told though, I really think it's more that the sites that offer such downloads cater two type types, those that want small files due to slower ISPs or time constraints, and those whom prioritize HD video quality. From what I've heard there's no comparison between the x264 videos and the Xvid videos on such sites, there are simply none of the flaws readily visible in the Xvid ones. I saw an x264 compression of the Fantastic Four: Rise of the Silver Surfer Blu-ray movie once on a 720p TV and I swear, you'd have thought it was actual Blu-ray.
I get the feeling that at 720p and up resolutions, which obviously I'm not showcasing here, the differences might be even more dramatic, perhaps in both image quality AND file size.
Togue Sport Burn Rt Xvid
Togue Sport Burn Rt x264
Togue Sport Burn Rt x264 30MB