Hardware vs software captureing.

[snipper_cr]

Is it correct to say if i am "hardware" encoding, say, MPEG 2 it means that the capture card is doing all the work? While if i do software encoding the processor and system is doing the encoding?

I assume hardware is better than software?

I can capture two ways: With my Radeon X850XTPE via the Rage Theatre chip or via a breakout box of my HDTV Wonder card.

Are either of those two hardware encoding?

[Cornucopia]

Side note: Even in "hardware" encoding, there's "software" in the "firmware", so in a way it's ALL software.

The difference is quality of algorithms, and whether the algorithm has to work on a live/linear stream or not.

Hardware encoding (during capture) is stuck working with "live" streams and not files, so it obviously can't do 2-pass VBR or extensive motion search (without huge buffers). Exceptions to this are the VERY PRO cards which control tape machines, use timecode, and so can do 2 pass, etc.
Algorithms vary from sub-par to above average (excepting pros again).

The good thing is the off-loading of the task from the CPU (allowing the use of busy or less capable CPUs/systems) and the real-time nature of the capture.

Software encoding has the benefit of working on a random access file, making multipassVBR and extensive motion search very possible. Then, it's all in the quality of the algorithm, which vary from fair to as-good-as-PRO Hardware. The downside is, of course, at least ~2x realtime (for CBR/1passVBR) or ~3x realtime (for 2-passVBR)--although faster systems will decrease this. The other downside is high system utilization (that also is lessened with better systems).

Scott

[mikiem]

Welll... depends on which ATI docs you read.

Long story short, hardware compression at capture hasn't really been necessary since cpu's were less then 1 gig & then the format was usually mjpeg. I'd guess that on your PC you're probably using somewhere around 20 - 30 % CPU capturing mpg2, which is more than fine as you really shouldn't be doing a lot while your capturing, IMO anyway.

So, you've likely got 3 choices: Using your X850 you should be able to do hardware mpg2 compression, but the latest ati software is botched so your mileage will vary -- probably considerably. Or you could use software mpg2 compression. Or could use your other card. Test all 3 and see what *you* think looks best.

[snipper_cr]

Okay so for all intents, using the Rage Theatre inputs of my X850 it should use hardware MPEG2 compression. As far as my HDTV wonder i dont know if it will use hardware or software. Id like to think hardware but either way id just use Radeon for capturing since it has hardware... but who knows what the cat drivers will do... Thanks for input.

Now i just need to get around Macrovision.

[jman98]

What Cornucopia is not telling you is that he is referring to software re-encoding of files and not doing live VBR captures, which in no way at all are better than hardware encoding and may, in fact, be worse depending on your PC and what you are using. He's not really comparing apples to apples here. He is comparing a hardware encoded capture, which he assumes will not be re-encoded with something like CCE, to a capture that will be re-encoded with software. What the original poster is asking about is more like asking if Hauppauge (hardware encoding) is better than ATI (software encoding) for on the fly capturing.

I believe very strongly that hardware encoding IS superior and there is nothing to stop you from doing a hardware capture at, say, 9000 Kbps CBR and then re-encoding it in software down to 4500 VBR. You will get much better results from that than doing an on the fly capture at 4500 VBR with either hardware encoding or software encoding on the fly.

I think the Hauppauge PVR-350 is hands down the best capture card I have ever owned and it does hardware encoding. ATI will always have its fans who think it is better.

[mats.hogberg]

you really shouldn't be doing a lot while your capturing, IMO anyway.

...with software encoding. With a hardware encoder, you are free to continue to use your computer just as normal regardless of if you're capturing or not. It doesn't affect the capture process whatsoever.

/Mats

[edDV]

I can agree with both Cornucopia (post encoding) and jman98 (hardware stream encoding) here and both techniques have a place. Post encoding has the luxury of not having to work in realtime and the huge advantage of "knowing the future and the past" depending on how it detects motion. Most encoding software doesn't bother with the latter advantage because it slows encoding.

I'm a bit confused by mikiem

Welll... depends on which ATI docs you read.

Long story short, hardware compression at capture hasn't really been necessary since cpu's were less then 1 gig & then the format was usually mjpeg. I'd guess that on your PC you're probably using somewhere around 20 - 30 % CPU capturing mpg2, which is more than fine as you really shouldn't be doing a lot while your capturing, IMO anyway.

So, you've likely got 3 choices: Using your X850 you should be able to do hardware mpg2 compression, but the latest ati software is botched so your mileage will vary -- probably considerably.

...

Yes, MJpeg cards used hardware compression.

But what software encoders work with 20 - 30 % CPU (P4 2.8C) while MPeg2 capturing? The Mainconcept would be running closer to 50-70% within the buffer. I'm assuming fairly high 6500-8500 Kb/s CBR or single pass VBR bitrates. Higher compression would peg that CPU.

Also, is there a version of the ATI x850 that supports video capture? That card supports playback decoding but I can't find a model that supports hardware encoding input.

[Cornucopia]

What Cornucopia is not telling you is that he is referring to software re-encoding of files and not doing live VBR captures, which in no way at all are better than hardware encoding and may, in fact, be worse depending on your PC and what you are using. He's not really comparing apples to apples here. He is comparing a hardware encoded capture, which he assumes will not be re-encoded with something like CCE, to a capture that will be re-encoded with software. What the original poster is asking about is more like asking if Hauppauge (hardware encoding) is better than ATI (software encoding) for on the fly capturing.

I believe very strongly that hardware encoding IS superior and there is nothing to stop you from doing a hardware capture at, say, 9000 Kbps CBR and then re-encoding it in software down to 4500 VBR. You will get much better results from that than doing an on the fly capture at 4500 VBR with either hardware encoding or software encoding on the fly.

I think the Hauppauge PVR-350 is hands down the best capture card I have ever owned and it does hardware encoding. ATI will always have its fans who think it is better.

jman98, you're right. I was seeing the question through my experience and preferences and not the OP's. I agree very much with what you said, although my preference is to ALWAYS capture as high quality as possible. Which, to me, works out in this order (best to worst):

• Uncompressed YUV 4:2:2
  MJPEG compressed ~2:1 ... 3:1
  DV compressed
  Very High BR MPEG2 compressed (I-frames only, if poss.)
  ...all the rest

I much prefer going to an editable format, and then (re-)compressing to the final format via a file or frameserve. Yes, it uses up more space, and yes, it adds to the TOTAL amount of time spent on a program, but 99% of the time, I think it's worth it for the quality I get. I can do other things with other computers while that's chugging away anyway.

Scott

p.s. jman98, can you do ~15Mbps MPEG2 I-frameOnly with that card?