MPEG-1 vs MPEG-2

[Todd1]

I've read that, "VideoCDs are MPEG-1. They are naturally frame-based, as they are at 240 vertical resolution. Super VideoCDs are MPEG-2. MPEG-2 is always field-based, as it has a vertical resolution of 480."

I do not create CDs. I use MPEG out of Power Point for presentations. But the above bothers me because I create MPEG-1 videos at 704x480. They look good but I wonder if I'm kidding myself - is the MPEG-1 coding tossing out half my resolution? Or is it using the full 480 lines?

If I encode the same 704x480 YUY2 source file in each MPEG-1 and MPEG-2 with the same data rate will the quality be the same? I use MPEG-1 for ease of use on many computers.

[Eric VanHeest]

If you need to convince yourself that you're getting the full resolution, just make a half-res MPEG-1 file, and you'll see the difference, I guarantee it.

As far as I know, MPEG-1 doesn't really have anything to do with frame vs. field; it just takes video streams at any resolution (subject to some minor restrictions) and encodes them as best it can. VCDs are 240 vertical pixels, and happen to use MPEG-1 for the encoder; there's no other correlation between MPEG-1 and "240 lines."

MPEG-2 is definitely considered "better quality" than MPEG-1 (specifically, I think, at higher bitrates, such as the 3000+ range). MPEG-2 "understands" interlaced video and doesn't produce the same artifacts that a "blind" compressor would with the interlacing (meaning that you don't need to de-interlace a video before compressing to MPEG-2 if you are going to be displaying it on something that also "understands" interlaced video, such as a television).

[Todd1]

Thanks for the frame vs. field clarification.

When you write 3000+ bitrates I assume you mean 3 megabytes/second for a final data rate. My personal rule of thumb is to keep the video playback data rate under 1 megabyte/second because I sometimes get stuck with real clunker computers. I'm very happy with my results at 250 kilobytes/sec to 350 kilobytes/sec.

I also assume that when you wrote that MPEG-2 "understands" interlaced video and doesn't produce the same artifacts that a "blind" compressor would with the interlacing (meaning that you don't need to de-interlace a video before compressing to MPEG-2 if you are going to be displaying it on something that also "understands" interlaced video, such as a television) - that this does not apply to giving presentations with a PC and a projector.

Based on what I'm using video for (and your input) it would appear that I'm not losing anything using MPEG-1 over MPEG-2.

[Eric VanHeest]

-----------------------------------
When you write 3000+ bitrates I assume you mean 3 megabytes/second for a final data rate.
-----------------------------------

Unfortunately I neglected to place a unit after my number, but I was referring to 3000 kilobits per second (not kilobytes), resulting in about 375 kilobytes per second for playback. I don't know what MPEG-2's "sweet spot" is as far as bitrate vs. MPEG-1, but I do know that MPEG-2 requires a lot more processing power to decode.

I'm not very familiar with what technology PC projectors use, but since I can (usually) read them at resolutions of 1024x768 and higher, I assume they aren't using anything related to broadcast specifications (which are impossible to read at anything other than 640x480, e.g. the G400's TV-out), and would therefore be susceptible to any interlacing effects visible on a standard PC monitor. I assume that you either de-interlace your input video somehow, or have generally low-motion videos.

[Walrus]

Todd1: MPEG-1 can support vertical resolutions above 1000 pixels (I *THINK* the max is 4096), but you are correct it is designed exclusivley for frame-based encoding. MPEG-2 can do either field or frame encoding. As for MPEG-1 at full resolution... it's encoding everything, but if you have interlacing issues it will blur a lot around fast moving issues since the interlacing artifacts do not encode well. MPEG-2 *can* get around this by encoding every other line into a seperate field and thus avoid having to encode the interlaced image as a single frame. (note: not all MPEG-2 encoders support this) As for the quality... the standards are essentially the same. (MPEG-2 has a bit more data blocks in the basic format, but essentially it uses the same compression mechanism). If the settings used are the same, the MPEG-1 and MPEG-2 file at the same datarate should look the same. (though the MPEG-2 file should be a few bytes bigger do the slightly higher data format overhead).