After Effects CS3 pulldown removal automation

[homomech]

Hi, the script below will automatically guess pulldown removal for all clips selected in the Project window. It is as if you right-clicked each clip individually, selected "Interpret Footage/Main", and pressed the "Guess 3:2 Pulldown" button. Except it all happens in one click.

function ReverseTelecine() {
    var sel = app.project.selection;
    if (sel.length == 0) {
        alert("Please select clips in the Project window.");
        return;
    }

    for (i = 0; i < sel.length; i++) {
        if (sel[i] instanceof FootageItem) {
            sel[i].mainSource.guessPulldown(PulldownMethod.PULLDOWN_3_2);
        }
    }
}

ReverseTelecine();

To use the script, just copy-and-paste it into TextEdit, then save it as .jsx file (eg, ReverseTelecine.jsx). You should save it to your scripts directory. (On OS X, that's /Applications/Adobe After Effects CS3/Scripts). Then you can run the script from the pull-down menus in AE by selecting File/Scripts/ReverseTelecine.jsx.

I haven't tested this script on Windows, but it should work the same.

Workflow:

1. Capture your clips to .mpeg files using Premiere. (This is fine for me, as I almost never want to capture all the footage on a tape. I usually have a log sheet telling me which clips I'll actually need, so I have to set up the capture manually anyway. YMMV.)

2. In AE, bulk import all those .mpeg files to the Project window. They should all be selected after they import.

3. Run the above script.

4. Drag all the clips to the "Create a New Composition" icon in the Project window.

5. In the "New Composition from Selection" dialog that appears, make sure "Multiple Compositions" and "Add to Render Queue" are checked, then hit OK.

6. Open the Render Queue tab and hit the Render button.

7. Have a beer.

8. Edit the rendered clips in Premiere.

For this to be effective, you have to set up After Effects so that its default render settings are what you want (eg, Prores 422). So you might want to render one clip manually before doing the above, so the render defaults and target directory are correct.

-Paul

[lordtangent]

Awesome. Thanks for sharing!

[Eugenia Loli-Queru]

Now, that's good stuff.

[hazemabdulrab]

THANK YOU.....U ARE THE MAN......NOW QUIICK question this is required only for 24p right?..........and i think depending on audience..web,tv,dvd?

[appy]

You 're AWESOME!!!

[Max Goldberg]

Does this work well? I heard After Effects just guesses the pulldown?

[DaFireMedic]

This is FANTASTIC!

I just tried this in AE 6.5 with Vista and it works!!!!!

To create the script in Windows, open Notepad and copy homomech's script into it. Now "Save As" and when the dialog box opens choose UTF-8 in the "encoding" field. Save the file as Reverse Telecine.jsx and follow the instructions just as Homomech wrote.

(oh, and in 6.5 I have to rename the m2t file to an mpg extension to get it to import the file)

I have 7.0 on my other computer so I will check it with that but I see no reason why it wouldn't work in that as well.

Thanks again!

[homomech]

THANK YOU.....U ARE THE MAN......NOW QUIICK question this is required only for 24p right?..........and i think depending on audience..web,tv,dvd?

You're welcome. Yes, this only required for 24P. And yes, the decision to use 24P depends on your ultimate audience.

[sfdex]

To answer Max's question, yes, it seems to work on all clips, so long as you don't have more than one shot within one captured clip. I mean, when you were shooting, you stopped the camera, but when you capture in Premiere, you capture across the cut. The HV cameras seem to re-start their cadence with each push of the record button.

Awesome work, Homomech.

[homomech]

Does this work well? I heard After Effects just guesses the pulldown?

In my experience, AE always guesses correctly. However, that's because of the way I capture the clips. I manually set up clips to capture in Premiere, because Premiere does not correctly determine clip boundaries for HV20 footage.

If you hook up your camera to Premiere and tell it to capture the whole tape, it will just give you one giant file with all of your clips in it. That won't work for AE pulldown removal, because AE only guesses once. The first clip in the file will work correctly, but the subsequent ones may not.

So, for the above script to work, you have to spend a lot of time in Step #1 setting it up so that Premiere captures 1 clip to 1 file. Again, for me this is not a big deal, since I have to do that anyway (about 80% of my tapes are useless takes I won't need). For you, though, Premiere/AE might not be the best solution.

The ReverseTelecine script should work for files captured through programs other than Premiere, though. Maybe one of the other posters can see if that's the case.

[Max Goldberg]

How about HDVSplit? That automatically makes clips, right?

[netrate]

Just read this news on HDsplit, maybe it has been seen before :
http://www.dvinfo.net/conf/showthread.php?t=125559

So, does one have to capture using HDsplit first and then run through AE? How about capturing through Vegas?
Also, this business of capturing clips only, do you mean scenes? Can you dump a whole tape or do you have to work in small clips or something?

[Max Goldberg]

Each time you press record has to be a separate file.

[netrate]

With Vegas, it has a scene detection feature when capturing, I think that might be the way to do then. Why does HDVsplit do that out of curiousity?

[sfdex]

Premiere Pro has scene detection, too, but for some reason it doesn't work on HDV footage from my HV30. I suspect that is a factor of the odd way Canon flags (or doesn't) the footage for cadence detection.

So, yes, to use this script successfully, Max has the right answer; every time you press record while shooting has to be captured as an individual file on your hard drive.

[CLJ891]

Thank you so much! This is exactly what I needed. Worked perfectly with Vista and AE CS3.

[DaFireMedic]

I have 7.0 on my other computer so I will check it with that but I see no reason why it wouldn't work in that as well.

Thanks again!

Update: I have since tried it in 7.0 and the script works fine.

This really helps me out cuz for some reason I have not been able to get HV20pulldown.exe (which is a great program) to work on Vista as other people have, so I have been using AE on my Vista desktop for pulldown removal. This script really speeds it up, making the process as fast as the TMPGEnc that I bought for my laptop.

[visugeek]

Great script, I'm using it with cs3 on xp64, it works great! Thanks for sharing!

[resx]

why dont my mpeg files captured from premiere open in after effects? Everytime I try it says file format is not supported.

[Max Goldberg]

I got my editing computer back today and tried this, but I can't get it to work right at all.

It looks good in After Effects, there is no interlacing, and the aspect ratio is right.

When I try to render it, I've tried it a few different ways, I get a variety of incorrect videos. There is interlacing, and the aspect ratio is totally wrong. Also, it won't play smooth and the file sizes have been ranging from 200mb to over 2gb. The original file was 91 mb... Can anyone help?

[DaFireMedic]

why dont my mpeg files captured from premiere open in after effects? Everytime I try it says file format is not supported.

What version of AE are you using?

Mine open fine in 7.0, but in 6.5 I have to change the file extension from .m2t to .mpg.

[DaFireMedic]

I got my editing computer back today and tried this, but I can't get it to work right at all.

It looks good in After Effects, there is no interlacing, and the aspect ratio is right.

When I try to render it, I've tried it a few different ways, I get a variety of incorrect videos. There is interlacing, and the aspect ratio is totally wrong. Also, it won't play smooth and the file sizes have been ranging from 200mb to over 2gb. The original file was 91 mb... Can anyone help?

Try it with one file to get it dialed in, and make sure that it was shot in 24p, if you haven't already. Before applying the script, the file will be 29.97 fps. After applying the script, the file should now show as 23.976 fps while sitting on the timeline, so you know that the script has worked. Now send it to your render queue and check your settings. I always export to .avi using Lagarith as the codec (big file output) or uncompressed (VERY big file output). It will always end up significantly bigger than your original file unless you export in a very "lossy" format, which you don't want to do. Also, these "lossless" files are not going to play smooth, that is just the way it is with lossless codecs. After you edit it in Premiere and export to your "delivery" format, it will play smoothly. As for the aspect ratio, make sure that you are exporting as square pixels (1.000).

[homomech]

It looks good in After Effects, there is no interlacing, and the aspect ratio is right.

When I try to render it, I've tried it a few different ways, I get a variety of incorrect videos. There is interlacing, and the aspect ratio is totally wrong.

How are you playing back the files? After Effects and QuickTime Player don't really get along. I often find that clips exported from AE have the wrong aspect ration in QuickTime Player. But, they play back fine in Premiere, so it's just an annoyance for me. I've never encountered an interlacing issue though.

Could you list the export settings you're using in AE? Might help narrow down the problem. Also AE version, operating system, capture program, etc.

[Max Goldberg]

Well, I've tried a few different settings, but I'm just curious as to why the new 24p footage is way bigger (in file size) than the original 60i footage. I would lossless would not lose any quality, why would it gain file size?

So I want to do square pixels at 1930 x 1080? Or 1.3333 aspect ratio at 1440 x 1080?

I still just don't understand how a video file with less frames has a larger file size...

[homomech]

I still just don't understand how a video file with less frames has a larger file size...

Because there's more information per frame when you save loseless.

It might help to consider the size of a fully uncompressed HD frame of video. For a 1440x1080 frame with three (RGB) 8-bit color channels, you need 1440 x 1080 x 3 bytes to store just a single frame, or about 4.5 megabytes. So, for a single minute of uncompressed 24P footage, you'd need 4.5 x 24 x 60 megabytes, or about 6.5 gigabytes per minute.

So if your one-minute clip is compressing to 2 gigabytes, that's actually pretty good.