Some questions about NTSC

[Lars Bunch]

1. Drop Frame TC keeps accurate time, but is inaccurate in absolute count of frames. Non-Drop Frame TC does not keep accurate time (it drifts a little over time) but it’s absolute frame count is accurate. So if you are cutting for broadcast, where knowing exactly how long your project really is is critical, then you use DF. But if you are cutting for eventual conform to film or exactly accurate time isn’t critical, then NDF is better. I prefer NDF because I’ve never been too keen on being on time.

2. The frame rate should be 29.976. This is the most accurate representation, but all these fractional numbers are referring to the same frame rate so you shouldn’t run into any problems no matter which you set it to. But as I’ve said, time isn’t a big deal for me so take that with a grain of salt.

3. If I remember right, the PD-150 doesn’t shoot 24fps so there will be no pull down to remove. The only time you would remove a pull down is on material that has been shot at 24fps but then interpolated up to 30fps.

4. I wouldn’t change any frame rates. If your final project is to be finished in PAL, but you are shooting in NTSC, you would be better off doing all your work in NTSC and only when you have a finished version, convert to PAL. That said, PAL is a better format than NTSC and if you could work in PAL, you’d probably be happier with the final results. Unfortunately NTSC means more frames per second to work with, but if you change frame rates back and forth, you’ll end up with weird, stuttery motion artifacts.

5. I can’t advise on ProRes422, but that won’t stop me anyway. If you are shooting with a PD-150, anything you do with the image afterwards can’t be much worse. Obviously working uncompressed preserves as much of your image quality as possible so you should use that codec for intermediate work. But once you have a final color corrected project that doesn’t need further adjustment, you’d probably be fine with a compressed codec. For example, shooting in DVCam isn’t a problem so long as you nail your exposures and don’t need to do much color or levels adjustments. But if you do need to modify the image, DVCam falls apart very quickly. I hear ProRess422 is pretty good so my guess is you’d be fine with it as a final delivery format, but it could be harder to work with in intermediate stages.

Hope this helps,

Lars

[Steve Roberts]

1. As Lars wrote, DF for broadcast, NDF for other situations.

2. If you’re just capturing in FCP, just capture. The material will be 29.97 on the tape. When you import into AE, let AE interpret the footage as 29.97 lower field first if you shot it 29.97 interlaced (60i). However, if you shot 24P, you need to remove pulldown in AE, since the camera took 24fps and added pulldown to make it 29.97. This would result in 23.976fps footage. FCP calls it 23.98, but you should call it 23.976 in AE.

Once you import and interpret, drag the footage onto the “new comp” button at the bottom of the project window to make sure the comp settings match the footage.

3. No no no. Just interpret the footage properly: if it was shot 24P, interpret lower field first (LFF) and remove pulldown, then drag onto the button as I wrote — you’ll have a 23.976 comp with no fields, if it was shot 24p.

There’s no need to use fieldskit for this. I’ve used Fieldskit before in odd situations, but I can’t remember when. Just interpret.

If however, the footage is 29.97 interlaced (60i), consider making a 59.94 fps comp, interpret the footage as LFF, and drag into this comp. This is what you do when you’re rotoscoping.

4. The PD150 shoots to the DV codec, no? Just capture in FCP, cut out the waste in a DV sequence if you want, then export the sequence with current settings. The footage will not have changed. Import into AE, do your magic, then render to the DV codec again if you expect to go back into FCP and export a complete timeline from there with no FCP effects.

However, if you’re not going back to FCP, then you should use AE to render to the Animation codec or 8-bit (or 10-bit) uncompressed, or Photo-JPEG before compressing to MPEG-2 in your compression app. ProRES and AE may still be dodgy together, but you may want to do some tests.

[Keith Hamilton]

One thing I’d like to add is that if there is going to be some heavy compositing, chromo keying, color correcting, etc., then I suggest you convert your footage from DV to ProRes 422 before doing your work in After Effects. Your footage won’t look any better than DV, but ProRes 422’s larger color space should produce better results for the large amount of post work you’re doing. I’ve done a lot of heavy post work (keying, color correcting, VFX) with ProRes 422 in After Effects CS3 and haven’t noticed any issues.

Also, if you are not going back to FCP, export a digital master out of After Effects by rendering an image sequence (TIF format using LZW compression). Then, create your MPEG-2 encoded video for DVD from that. Or any other output format required.

If you are returning to FCP, then render a video with the Animation codec. That’s what I usually do.

——————————
H. Keith Hamilton
Fellowship Bible Church (Day Job)
SkyGate Studios (Freelance)
kham4002@mac.com

http://www.skygatestudios.com

[Bohdan Stehlik]

Thank you all for your time and this detailed answers! Things seem much clearer to me now.

I worked on some NTSC edits in the past, but it was just capturing into FC, cutting and back to tape. This time while capturing, I got error messages about DF / NDF; strangely only on some clips from the same tape. This made me start worrying about those fraction frames, and how to handle them in AE. After your comments, I’ll stick to NDF, since time isn’t relevant in that project.

I know that converting to a different codec won’t improve the footage, but as Keith wrote, the idea was to have a larger color space for color correcting. So thank you for confirming my thoughts.

De-interlacing the footage with FieldsKit is because most of the final images will be composed out of many different shots, to give a “tiled” look, like in David Hockneys works. I supposed nudging, moving and slightly rotating around progressive footage is easier, because there won’t be any problems with the fields. Not sure if that assumption is correct.

Again, many thanks for your answers! Unfortunately because of the budget, we have to shoot in DV-Cam, the other option would have been a small HDV camera, but I know that keying HDV is a real pain!

Bohdan

[Keith Hamilton]

Shooting in HDV? Capture the HDV footage and then convert that footage to ProRes 422 using Media Manager or Batch Export in FCP.

Here’s a great article on when to convert to another codec from the native codec.
https://www.kenstone.net/fcp_homepage/when_to_stay_native.html

——————————
H. Keith Hamilton
Fellowship Bible Church (Day Job)
SkyGate Studios (Freelance)
kham4002@mac.com

http://www.skygatestudios.com

[Steve Roberts]

[Bohdan Stehlik] “I supposed nudging, moving and slightly rotating around progressive footage is easier, because there won’t be any problems with the fields. Not sure if that assumption is correct.
“

You are correct. If the footage were interlaced, its fields must be separated (interpret footage) if you are going to move/rotate/scale/blur/distort the footage. Fieldskit will give a more accurate result if the footage was 60i, because, technically, the full frames don’t exist in that kind of footage, and they have to be created by AE or Fieldskit.

However, if your footage was shot 24P, then the original full frames do exist, and Fieldskit will give you the same thing as separating fields.

[Bohdan Stehlik]

Thank you Steve for confirming that. I’ll use FiledsKit, because in the past I got better results with it than using AE for the job of creating full frames.
The PD150 can shoot in progressive mode, but only at 15i/s, which is not a option.

Keith, it was exactly that article which made me think of using ProRes422 instead of the Animation codec. I have to pre-render many parts of my comps, otherwise the final output would take to long, so I try to save some disk space. (Just hoping the ProRes won’t be to dodgy with me!)
As for the HDV, I made some chromakey test with footage in that format, but due to the heavy compression of it, the SD images seemed to give better results. Even with a good lighting, the HDV results were far to blocky.

Thanks to all of your suggestions and precisions, I think I’m ready to go!

Bohdan

[Darby Edelen]

[Bohdan Stehlik] “This made me start worrying about those fraction frames, and how to handle them in AE. After your comments, I’ll stick to NDF, since time isn’t relevant in that project. “

Sorry, I’m a stickler for details so here’s a quick note =)

There’s no such thing as a fraction of a frame (unless you consider a field a half a frame). In NTSC the frame rate is fractional. The reason this becomes confusing is because tape decks count timecode in intervals of whole frames, which is the whole reason that drop frame timecode was developed.

In NTSC-land, when your deck reads that 30 frames have elapsed (one second) only 29.97 frames have actually elapsed. As non-drop timecode progresses an error builds up between the reported timecode and the actual duration, this amounts to 18 frames every 10 minutes (30frames/sec * 60sec/min * 10min = 18000 frames… 29.97frames/sec * 60sec/min * 10 min = 17982 frames, a difference of 18 frames).

In order to account for this difference between the timecode and the duration of the footage, drop frame timecode skips 2 frames every minute except for every 10th minute, which accounts for 18 frames every 10 minutes.

Darby Edelen
Designer
Left Coast Digital
Santa Cruz, CA

[Bohdan Stehlik]

You’re absolutely right, of course. I never thought about it that way, since I live and always have worked with video in PAL-land, but your explanation makes it even more obvious.
Every time I got stuck thinking about the .976 of a frame, but it’s like explaining the image size / resolution-relation in digital photography; you can’t change the length of an Inch, but you can change the size of a pixel when modifying the dpi without interpolating.
Well, I think you understand the comparison; in every case one more light went up in my head, thanks to you.

Bohdan