DV Progressive Capture via ATI Radeon

[Mark Fredrickson]

I have a progressive scan DV camera that I want to capture using the camera's analog outputs into my ATI Radeon video card encoding to MPEG2.

My question is what will the MPEG2 video file be - interlaced or progressive? If it gets changed to interlaced, when does that occur, when it leaves my DV camera via the analog outputs or when the Radeon processes it? I'm trying to figure out what settings to use in the ATI MPEG2 capture settings.

My reasons for this insane method of capture is speed. I can capture directly to MPEG2 via the Radeon in absolute real time. I'm then authoring to DVD. I want to transfer my home videos to DVD in the minimum time possible (with the least expense). I have tried software encoders (Mainconcept MPEG Encoder) via my firewire and they take too long (buffering and buggy). A year ago I was in the DivX camp and got burned out spending 10+ hours encoding 1 hour of video, and after encoding 40 hours of home video, I still have 200 more to go.

Thanks,

Mark.

[bb]

So you're not looking for quality ;)

If your DV is interlaced, it will stay interlaced even after transferring it over firewire. When encoding directly to MPEG-2 you have to make sure that progressive encoding settings are chosen. Most MPEG-2 encoders default to interlaced encoding, which is not optimal for progressive sources. You should be able to flag the frames as progressive, and to choose zigzag scan instead of alternate scan as encoding options. Unfortunately I don't have access to an ATI Radeon with MPEG-2 encoder, thus I don't know where to find the appropriate settings.

bb

[Mark Fredrickson]

Hi bb,

I remember conversing with you back in my DivX days. Yes, I have changed from quality to speed. The main reason is because I discovered that while I was engrossed in quality, my children were growing up and I still had not given them the ability to view their own home videos.

Oh, I could show them the best CD video on my computer, but all I had was only half of them 20 years ago (I have 4 girls). Now my oldest is off to college and I realized that maybe it would be best if I just gave up on quality and concentrated on creating something they could watch.

Back in my DivX CD days, I installed an X10 RF unit that allowed me to broadcast my PC monitor output to the TV via RF. I thought it was real cool, but did anybody else in the family use it? The answer is no. It was then I realized that I had to give them something very simple, and that was DVD.

So I made a conscience decision to start all over and encode all my 25 years of video to DVD as quick as possible but still allowing time to edit and authorize the DVD's with titles, chapters, etc. I've haven't quite given up with quality. I'm sill encoding at 8mb bit rates so the VHS/DVD looks reasonalbly good. DVD's are cheap now, so I'm only shooting for 1 - 1.5 hours on a DVD.

So back to my original post. I'm now assuming that when I record using Progressive Scan, that my DV camera will output Progressive Video either via the DV 1394 port or the RCA jack. It doesn't care, just that the DV 1394 port will send higher quality video because of the nature of the wires (4 wires for 1394, 1 wire for RCA).

With that in mind, I'm assuming that my camera is sending the Progressive Video through the RCA jack to my ATI Radeon video card (albeit in less quality than 1394). If this is correct ???, then I do have the ability to tell the Radeon that the source is Progressive.

So, more questions if you don't mind...

When I tell my Radeon that the source is progressive, how do you think it will encode the MPEG2 video? Will it stay progressive, or will it change it to interlace? In other words, will my DVD recording of my DV camera progressive video be progressive or interlaced? (is this confusing or what????).

The Radeon also has the capabilty to select a 3:2 pulldown option. I believe that this is only applicable to Film and that my DV camera is videoing at 29.97 so I should not use this option. Is this correct???

Can't think of anything else quite yet. It's late and I have a ton of Christmas shopping to do tomorrow. I can wait on the DV recording questions for a while since I have many, many, many hours of VHS encoding to do first, but I would appreciate figuring all this out now so I can rest my brain :)

Thanks,

Mark.

PS. I was born in Weisbaden but left when I was a wee baby. My dad flew in the Berlin airlift. I'm looking forward to coming over someday and seeing my birthplace.

[Sulik]

You can use "Progressive source" when capturing with ATI MMC, it will not deinterlace, and will mark all frames as progressive.
You need to be sure that your source is progressive, though -> since you're going through Analog, each field is sent separately.

If you see interlacing artifacts when playing back the capture, I'd suggest capturing interlaced instead.
It's much less destructive to capture progressive material as interlaced than capturing interlaced material as progressive.
Another way if you don't want to try twice is to deinterlace - since ATI MMC uses adaptive deinterlacing, it shouldn't damage the progressive frames (if anything, it might help compression).

Do NOT use inverse 3:2 pulldown unless your source is a movie (24fps converted to 29.97), which is very unlikely if your source is DV.

Also, if you're using the latest MMC 8.8 you can experiment with capturing directly to MPEG-4.

[Mark Fredrickson]

So, progressive video is transmitted differently depending on the connection (1394 versus RCA)? What would be the best way to tell if my capture is progressive or not? I could capture straight to AVI if that would help determine. I don't mind playing around with some short captures.

Also, how does progressive video relate to HDTV? I'm planning on getting an HDTV in a year or two. How would you recommend capturing progressive source video (via ATI MMC) to playback on an HDTV. My DVD player has a progressive scan button which will convert interlaced to progressive for HDTVs, but would it not be better to have my home video DVD already progressive?

[bb]

Mark,

capturing DV is merely a lossless data transfer from your camcorder to the PC's hdd. The DV data stream on the tape is the same bit by bit as in the "captured" DV AVI. It's more of a copy operation similar to restoring a backup from tape, so the word "capture" is a little misleading.

Analogue capturing is completely different. The analogue signal gets sampled scanline by scanline, and although this is done on a field basis, the captured video is progressive if your source is progressive. Due to the field capturing you may encounter a so-called "phase shift": if the wrong fields get recombined to frames, you'll get interlacing artefacts, similar to those you see in "real" interlaced sources. But unlike "real interlacing" the phase-shift can easily be corrected - unfortunately I have no idea how this could be done in the single-step process you desire. So let's hope you don't get a phase-shift. If you get a phase-shift, you'll see that in every scene with horizontal movement as comb artefacts (to try it you should capture a small sample uncompressed).

If your source is progressive, you should always encode progressive.

bb