Pagan Hammerite on 29/1/2011 at 13:16
I have just played T2 using my new USB sound card, a Cakewalk/Edirol UA-25 EX which gives you studio quality audio.
Intense, cinematic experience. If you are a serious taffer you should consider it, $300 or so.
I'm going back in.
Sulphur on 29/1/2011 at 13:32
Given that the source audio for Thief wasn't recorded at 96 kHz, and the game doesn't support direct output at that resolution anyway, I don't see what kind of difference it would make. If anything, the sound card having to resample everything to 96kHz would make it sound slightly worse.
Pagan Hammerite on 29/1/2011 at 13:43
I don't know how it works but when Garrett gets hit by an arrow you can feel it in your guts.
Made me jump out of my chair.
Sulphur on 29/1/2011 at 13:56
Well, it's probably because it's an all-round better sound card that whatever your previous audio solution was. Or simply different bass and treble settings. The 96kHz resampling certainly isn't helping the cause of clear, pristine audio if your sound card is forcing game audio output at that particular resolution.
lost_soul on 29/1/2011 at 16:24
It could make the sound better. Quake 1 shipped with 11 KHz sounds, and yet since the source code is available, people have added upsampling to the engine. For example, try Darkplaces and then normal quake and notice the sound quality difference.
Add this on to the list a mile long of things that could be improved if they would release the source code to Thief.
Sulphur on 29/1/2011 at 17:44
Resampling/'up'sampling any sound source to a higher resolution does nothing to the source audio. It doesn't make it sound better by itself, and in many cases, as in Creative's range of X-Fi's, if the hardware/software resampling algorithm sucks it actually makes your audio sound worse. This is exactly what happens with the X-fi, as the card internally resamples everything to 48 kHz then mixes and outputs it, unless you force it to 44.1 kHz to skip the resampling step. I've done that, and along with forcing bit-matched output, there's a world of difference to be heard on a good set of 'phones.
If the resampler's good enough, you should barely hear any difference whatever the resolution you choose.
lost_soul on 29/1/2011 at 20:30
Upsampling is a lot like (
http://en.wikipedia.org/wiki/Bilinear_filtering) Bilinear Filtering for textures.
It can improve things a bit, but you can't add detail that isn't there to begin with. Still, the techniques used to do these things are impressive.
The best part of upsampling audio is the reduction in "telephone-quality trebble". (phones use 8KHz)
The best part of texture filtering is an image that degrades gracefully, rather than becoming giant blocks.
Sulphur on 29/1/2011 at 20:41
I know what bilinear filtering is, chum, and upsampling isn't bilinear filtering. The analogy you're looking for in the image world is called 'rescaling'. You know, blowing up an image from a small resolution to a higher one, with an interpolation filter so that the final image doesn't look like a large mass of pixels. Bilinear filtering may share the interpolation step, but it and mip-maps belong in the world of 3D texture projection.
Like you said, upsampling can't add detail that isn't there to begin with. Upsampling doesn't clean up a dirty source either, but it can make it sound more muffled and less grating, which may have a perceived quality 'increase' when comparing the resampled audio to the original low quality source.
Child Of Karras on 29/1/2011 at 20:59
You are right, 96000Hz don't make the sound better. Not even after my sound update. BUT! What could be better is the dynamics and the signal-noise-ratio. Especially older onboard sound chips operate at 16 bit which give a total dynamic of 96 dB, while a human can hear up to 120 dB. I suppose your sound card operates at 24 bit, which is 144 dB dynamics. When the signal volume is close to the first bit (very quiet) it gets very noisy. And also the signal-noise-ratio is very bad at onboard chips. Computer internal noises cause buzzing and crackling on the line-out port, which at least can be reduced by deactivating Microphone and Line-In monitoring in Windows mixer, but it will never be as good as with a better PCI or external sound solution.
Sulphur on 29/1/2011 at 21:14
Yeah, like I mentioned it's probably because he's using a better sound card in comparison to his older one.
Crackling still happens even on PCI cards (X-Fi's, at least) sometimes though, but it's usually dealt with by muting the mic or, in my case, changing the flexijack to digital I/O.
