Bitrate 720x480 vs 352x480 concept question

[CaZeek]

Hello,

I am playing around with my capture settings to see what the different file size / quality combinations are. The source is a Disney XD cartoon from FIOS digital cable (not HD). I'm using the ATI Theater Pro 550 Hardware MPEG-2 encoding.

Trial 1: 352 x 480, 4MB/s avg (5MB peak) VBR capture --> Resulted in a ~ 4.2MB/s capture
Trial 2: 720 x 480, 4MB/s avg (5MB peak) VBR capture --> Resulted in a ~ 4.2MB/s capture

The quality difference is virtually indetectable across these two captures, yet one has over 2x the resolution. I expected the slower scenes to be slightly sharper in Trial 2, but fast motion to be blocky.

Assuming the source is approximately the resolution of Trial 1, does it not matter if I just throw extra pixels in the capture? I assumed that switching from frame to frame with twice the pixels (even if they are essentially just doubling each pixel from the lower res source) would require ~ twice the bitrate for similar quality.

[jagabo]

Cartoons compress quite well because of the low detail and slow animation rate. FIOS is probably transmitting at a digital resolution closer to 352x480 than 720x480. The smaller source captured at twice the frame size will not require twice the bitrate. Unless there's lots of analog noise.

[CaZeek]

Scene I'm checking is the intro to X-Men (used it to make sure I had some fast motion).

To get an idea: http://www.youtube.com/watch?v=VxOQTaLTFrU

[lordsmurf]

4.2MB/s is not enough for good Full D1 MPEG encode. There will be artifacts.

[jagabo]

There's lots of motion but many shots have little detail. I don't know if it's a bad capture or conversion, but although the FLV file from Youtube is 30 fps, the animation frame rate appears to be about 12 fps. All the duplicate frames make it much easier to compress.

[CaZeek]

4.2MB/s is not enough for good Full D1 MPEG encode. There will be artifacts.

Agreed- but capture is not producing artifacts that I can see. I'm side-by-side comparing to the half D1 video w/ VLC media player, and I really can't see a difference. I think I have a pretty good eye for this too, that's why I'm confused.

Just to be clear- the youtube video is not my video.. was just posting it so people can see what scenes are in the video I'm capturing.

[lordsmurf]

Put it on a 55" screen and check again.

[jagabo]

4.2MB/s is not enough for good Full D1 MPEG encode. There will be artifacts.

It's fine for low frame rate, low detail, cartoons over low resolution digital broadcast.

[Nelson37]

Cartoons are a somewhat special case which won't really give an accurate answer on bitrate and resolution.

Because they compress extremely well, the differences you are expecting just do not show up.

Try the same test on a live-action source.

[CaZeek]

Thanks for the replies- so bottom line on the theory is then that encoding to 720x480 "should" require about 2x the bitrate for same quality of 352x480, even if the source is (for argument's sake) exactly 352x480 already?

[lordsmurf]

Put it another way:
You're not gaining anything by making this 720x480.

[jagabo]

Thanks for the replies- so bottom line on the theory is then that encoding to 720x480 "should" require about 2x the bitrate for same quality of 352x480, even if the source is (for argument's sake) exactly 352x480 already?

No. A 720x480 cap of a 720x480 source should require about twice the bitrate of a 352x480 cap of th 352x480 source. A 720x480 cap of a 352x480 source should be somewhere in between those two. It has twice as many pixels but not twice as much information.

[jman98]

CaZeek - If you're happy with the 720x480 capture then keep doing that. If I was me, I'd capture at that resolution too. 4.2 Mbps can be fine if you have a clean source. I've done TV captures and encoded them at roughly that bit rate and I and others who saw them were pretty happy with the results. If you're pleased with what you are doing then keep at it. The only person you need to worry about pleasing is YOU.

[CaZeek]

Thanks guys- got it. Basically wanted to know whether bitrate was pixel-dependent or information dependent. Answer is somewhere in the middle (aren't answers always like that?). Been capping in 352x480- so will stick w/ that for now, but was more just curious on the science behind it.

Actually one twist which adds a little more confusion. I capped a full episode in 352x480 @ 3.0Mbps avg, 5.0 Mbps peak, the result was a 3.3Mbps MPEG-2. Comparing that to a 4.0Mbps MPEG-2 version, I actually did notice a very slight difference in artifacting during fast motion, as originally expected. That leads me to believe the best bitrate for 352x480 is somewhere closer to 3.5Mbps-4.0Mbps, which confuses me again, b/c if the 4.2Mbps was working for 720x480 (b/c of whatever reasons that allowed the cartoon to be compressed so well), then shouldn't the 352x480's bitrate for the same quality be significantly lower than that?

[El Heggunte]

...

then shouldn't the 352x480's bitrate for the same quality be significantly lower than that?

From my personal experience, it seems there is some relationship between the pixel aspect ratios and the achieved compression efficiency. If you use TMPGEnc Plus for example, it's easy to verify that a 480x352 encode gives you a smaller filesize than a 352x480 one.

[CaZeek]

...

then shouldn't the 352x480's bitrate for the same quality be significantly lower than that?

From my personal experience, it seems there is some relationship between the pixel aspect ratios and the achieved compression efficiency. If you use TMPGEnc Plus for example, it's easy to verify that a 480x352 encode gives you a smaller filesize than a 352x480 one.

That's interesting.. maybe the card is better suited for 720x480 caps also then.. I believe that is the resolution standard to the card, as it was the only one available before installing new drivers.