HDV to Uncompressed w/Letterbox

[John Pale]

This is normal. Unstretch it with the motion tab. You will notice much of your image is now cut off at the sides of the 4:3 frame. This is what they are asking for. Full frame, 4:3 non-letterboxed video. You can leave it “center cut” or use keyframes to pan and scan to keep important important areas in frame.

[Max Kaiser]

Thanks John,

The thing is, they’re fine with it being letterboxed, and that is what it really must be, not full-frame, pan and scan, 4:3. They just need it in 720×486 overall. Is the solution then to pull and push it into place using motion – that seems pretty hard on the video.
Thanks!
Max

[Billy Stuart]

Max,

Take your imported “.mov” and open it in the viewer. click the “motion tab” and open the “distort” tab. in the Aspect ratio enter “-33.33”.

This should give you your letterbox. then export you file again. check it with Quicktime and it should be correct.

good luck!

Billy Stuart
Post Supervisor
National Boston Video
617.734.4800

[Jim Martin]

can’t you just drop your HDV sequence in a new 8 bit SD 4×3 sequence and render?

Jim

[Max Kaiser]

Billy,

I tried you method and it worked amazingly. When I outputted that version to quicktime via export>quicktime>make movie self-contained, and compared it against the HDV to DV version I’d created by going HDV to Camera and back via camera downconvert it looked MUCH better on the computer screen. I was just amazed at how much clearer the graphics were, the colors, everything. I thought I’d done it. It was what I’d hoped the difference would be.

Then, I wrote both out to DVDSP via compressor using the BESTDVD 90minute 2pass setting and being careful to make sure my fields stayed lower. The resulting DVD didn’t look so hot on TV. The HDV-DV version looked pretty good as it usually does, but the Uncompressed 4:2:2 looked stuttery, but only very slightly.

Also, when I watch the HDV-DV in my firewire connected television monitor it looks pretty good, but the Uncompressed 4:2:2 version again looks like the image timing is just off. Is this because I really shouldn’t be looking at uncompressed video via a firewire television monitor connection? Still, how do I explain the stuttery DVD. I’ve checked my fields repeatedly and can’t seem to find the mismatch. I know HDV is upper dominant, and DV is lower, but I wasn’t sure which way to play Uncompressed, so I went with lower since this is what the FCP preset has.

Jim, when I tried your method, again, it worked fine on the computer monitor, but when I watch it on the firewire connected (as opposed to a kona card or such) monitor it looked very, very stuttery. Ditto when I wrote it out to compressor and DVD.

Basically, I can get the image to look great on the computer, but can’t seem to get the settings right to make it look good on the TV.

Any ideas? Sorry to belabor this, but I really need a great solution for working with HDV in a 4:3 letterbox timeline so I can add text, etc. to the bottom bars – preferably without going back to tape which so far is the only way that has yielded acceptable results.

Max

BTW, I’m using 1080i60 HDV captured via firewire from an FX1/

[Jim Martin]

don’t use 2-pass for anything in compressor. There are known issues

Jim

[Sean Oneil]

Use a de-interlace filter. Try the one that comes with FCP or try Stib’s free filters (Google that phrase). You may be able to add the filter to the clips on an NTSC timeline, or you may have to do it on an HDV timeline first, then nest that sequence into an NTSC timeline. I believe that will solve your problem. Just figure out which setting on the filter works best.

Normally these filters are not the proper way to do a downconversion. But in this case it should be fine.

There are two methods of deinterlacing. I forget what Final Cut’s filter calls them, but you should be able to figure it out. “Weave” method combines the two fields together and is perfect for a true 30fps video. It will make it a perfect 60p – suitable for re-sizing to whatever you want without any flickering artifacts.

But if the video was shot at 24fps (I think the Sony Z1U calls it CinemaMode or something), this adds pulldown so a weave deinterlace will not work. You’ll get nasty flickering. If this is the case, you’d need the “Bob” method of deinterlacing. This method throws away every other field instead of combining them. This essentially cuts the vertical resolution in half. So using Bob on NTSC turns 480i into 240p, and then blows it back up to 480p (but with the quality of only 240p). 1080i becomes 540p, blown back up to 1080p etc. Throwing away half the resolution is generally bad, which is why you want to avoid Bob with standard def footage. But since its 1080i and you’re downconverting to NTSC anyways, this method should be ok.

This is actually the method the FCP canvas and viewer both use, and it is why your video looks fine on the computer monitor – but it flickers on the external video monitor.

Again, I foget what terminology the FCP and Stibs filters use. I think they use their own terms rather than bob and weave. I’m not at work so I can’t check. But again, you should be able to figure it out w/ trial & error. I’ve downconverted HDV footage doing this and it works great.

Or you can hire a Teranex and not worry about any of this stuff :).

Sean

[Andrew Commiskey]

Which 4;2;2 uncompressed setting are you using? Make sure your frame is 720 x 480 and not 720 x 486.
DVD is 720 x 480. if you have 486 lines it will squeeze them together and the results are crappy. You can use compressor to crop 2 lines off the top and 4 off the bottom to solve the problem.
Hope this made sense.
Drew

Chaos is the beginning of everything.

[Max Kaiser]

First let me say, thank you very much for your post. It was very informative. I followed it to the letter but didn’t come out any the better. As for the stuff about DVDSP, I think that is further down the line so I’m not worried about that, first I gotta get it to look good in FCP.

Here’s what I did.

Took the 1080i60 HDV (upper field) footage dropped it into a 8 bit Uncompressed (lower field) timeline. Applied the the FC de-interlace filter. Tried choosing both Upper, and Lower and actually, all the other “minimalize flicker” settings as well.

I tried the same thing with stib’s filter which basically has only an upper or lower setting, and then derivations of that based on interpolation or duplication.

I also tried simply changing the field order in the UC timeline.

I also tried applying the de-interlacing filter in the HDV timeline, and then moving the clip.

In each case, sometimes things got better, but never was it as smooth looking as when I simply sent the HDV out to camera, did a camera downconvert, and pulled back in as DV. A very unwieldy solution when the client has a small change.

I notice that some people are using Natress filters, perhaps because FC and free de-interlacers don’t do a good job? I’d be happy to pay the $100 to solve this problem as it is important to our workflow. But I’m not sure that the filter could do the job. Graeme said he thought it was because my fields were out of order – I didn’t want to bore him.

Basically, I’m super confused about fields. Do I want to match them, or obliterate them? Is the de-interlacer changing an upper field to a lower field? Does my video played out through firewire have to be lower? Could there be another problem beneath the field issue? As I say, it does get better just not great.

Thanks again to everyone for their ideas!

Max

[Sean Oneil]

Sorry it’s not working out. I think there’s a “shift fields” filter. Instead of changing the field dominance, it just places the video one line higher or lower than center – which has the end result of changing the field dominance.

Sean