Fields – do I understand correctly?

[Derek Shin]

1) I can’t think of a real reason to separate fields when working in AE, maybe someone else could chime in. Do you mean reading “in” fields? Just make sure when you import your footage, you interpolate it correctly. And changing field order does not affect frame rate (i.e., PAL 25i vs Pal25p is the same frame rate).

2) Again, you should not change frame rate unless deliberately changing speed (i.e., slow motion). If this is for standard broadcast, you can simply import/work/render in standard Pal 25 interlaced. AE is good about dealing with all the video format issues for you behind the scenes.

3) Yes, you would want to export with progressive frames for web video. YouTube specifically uses Flash. You can use third-party compression programs such as Squeeze or Compressor that automates the compression for you. I usually use either Flash or H.264 for web.

[Chris Gleinser]

Thanks for your answer!

For 1), well the way that I understood is that although I’ve got interlaced 25 fps material, AE does me the favor to display it as 50 fps by doubling the lines (or interpolating) so that I can edit the video in a better manner than if it would show me the real 25 interlaced frames. But the video itself IS still interlaced, of course, and when I render my final video, those “doubled lines”, and all the effects that I made onto those single frames, will be put back together into 25 fps interlaced.

So that way, there should be no information loss to my original material and my final 50 half-frames should still be like before (if I didn’t use effects of course 😉

I hope that’s correct, because I produced some material with a PAL camera and I’d like to have the same interlaced quality when I watch the final thing on a TV screen.

So, again in a nutshell, video resolution and fields etc remain untouched, but for my convenience, AE displays me 50 full frames with doubled lines or interpolation, so that I can better deal with effects and so on.

I hope now it was a bit clearer 🙂

[Sam Moulton]

If you move or scale interlaced video in the comp you must separate fields. Moving interlaced footage up or down an odd number of pixels will reverse the field order and really screw things up if you don’t separate fields.

Any time you use a camera and place interlaced video in a 3D layer that’s not perfectly square with the camera and at the default distance you must separate fields.

About the only time you can not separate fields and be guaranteed that you’re not loosing quality and risking all kinds of output problems is if you just drop the video into the new comp icon or into your empty comp window or timeline to create a new comp with exactly the same properties, and then you don’t touch any of the transform options other than opacity. OTT, you should always separate fields.

If you want to see every field, you double the frame rate of the comp.

When you render you can choose to introduce fields or you can choose not to introduce fields. This is an artistic decision and has no other effect except for how the rendered product will look. Interlaced video for the web generally looks terrible. Interlaced video for TV generally has a smoother feel to motion. Non interlaced video for TV means that the set will still display fields but both will be the same moment in time so the results will have similar motion characteristics as film. They won’t be exactly the same, but they will be similar because of the different way the shutter works in film cameras vs video camera.

There are even differences in the way motion looks between film cameras. Most motion picture cameras have rotary shutters that have the shaft to the right side of the frame and they rotate counter clockwise exposing the film from the bottom of the frame to the top, but others have the shaft below the frame exposing the film from left to right. These two types of shutters produce different motion artifacts just as a video camera, which scans the frame from top to bottom every other line, then goes back and picks up the missing lines twice per frame and 60 times per second (50 pal) in the standard way produces different motion artifacts than a video camera that’s truly progressive and records the entire image for each frame. When shooting, especially panning, you have to take great care not to introduce a stroboscopic effect with both film cameras and progressive video cameras. When creating moving elements for non interlaced or progressive output you have to be very careful that you don’t end up with the same problem. Whether or not to separate fields is a much easier question to answer than whether or not to render interlaced for TV. Move, scale or rotate, separate. Render with fields, depends the motion in the frame and on how you want it to look on TV.

[Chris Gleinser]

Well, for my project I don’t think that I’d have to scale and rotate very much (but who really knows that in advance :), so the main stuff that I’d have to do would be color correction, cutting, blurring, well stuff like that. Of course adding some extra layers for text and doing some masking stuff, but the footage itself is shot in a way it should also look in the final product (regarding size, zoom, and so on).

I think I’ll have to experiment a bit…

[Steve Roberts]

Regarding separating fields, I’d add that you need to separate fields in **all** situations except:

– when you do nothing to the footage at all
– when you change colors, color-correct or change levels/curves

Basically, any time the original pixels or groups of pixels change shape or position in the frame, you need to separate fields. This includes the times when you are blurring or distorting the footage.

Of course, if you want to render progressive, you need to separate fields even if you’re doing nothing to the footage or just color-correcting, etc.

[Bret Williams]

Don’t know what you guys mean by separate. But you must always have AE interpret your footage correctly as either interlaced or field rendered. Trying to second guess AE’s render image will leave you with crap.

It doesn’t matter if you’re transforming, moving, scaling whatever! Moving your interlaced video up one pixel won’t hurt anything. If you tell AE how to interpret your footage, it will render it correctly. This is AEs job and it’s single job. To create great looking images.

Interpret your footage correctly. Create what you want. Then use the render cue to render to the size and format you wish. Whether it be interlaced or non interlaced.

[Steve Roberts]

[Bret Williams] “Don’t know what you guys mean by separate. But you must always have AE interpret your footage correctly as either interlaced or field rendered.”

“Correctly” is a matter of, er, interpretation. Sometimes, if you are doing nothing to background footage, and AE has separated fields automatically (by interpreting the footage as interlaced), the footage is softened slightly in the render. In that case it would be “correct” to override AE’s separation and set separate fields to “off” for the footage.

“Interlaced” is the same as “field rendered”, by the way.

Regarding “separating fields”, that is often done automatically when AE interprets on import. In general, we just need to make sure that AE has done its job. However, depending on the format and the interpretationsrules.txt file, AE might not separate fields automatically. In that case, we would need to select file>interpret footage>main and separate fields manually.

“Moving your interlaced video up one pixel won’t hurt anything.”

Moving interlaced video up or down one pixel would switch field order, which would not be a good thing. You might want to do some more research on the subject.

[Chris Gleinser]

Well with AE automatically separating fields I indeed meant the interpretation of the footage.

When I interpret footage as interlaced, AE won’t show me the typical lines in fast movements, so I guess it somehow interpolates, and my first thought was that AE shows 50 frames instead of 25, but those 50 with interpolated lines.

Now I saw that it’s still 25 frames, so I don’t know exactly what AE is doing. But when you set the framerate manually to 50, you can scroll through the whole animation and it seems incredibly smooth, so I guess in that case it’s done like I usually thought.

For me, it would be important that my interlaced footage will be still interlaced in the final product and having no information loss (let’s say if I wouldn’t do anything to the footage, just put it into a comp and export it back).

I also saw some tutorials about removing fields (or de-interlacing) manually, but I think that’s only important if you want to mix interlaced and progressive footage or if you want to produce videos for being watched on a computer.

[Steve Roberts]

When AE separates fields in 25 fps footage, it displays 25 frames per second. To make up those frames, in high quality mode, the fields are interpolated, and in draft mode, the fields are doubled.

When you choose to render interlaced, AE accesses the frames that it did not display (they were *not* discarded) and uses them to render the footage as interlaced again.

Here’s a doc:
https://livedocs.adobe.com/en_US/AfterEffects/8.0/help.html?content=WS3878526689cb91655866c1103906c6dea-7f37.html

Why don’t you just do a short test? Take some interlaced footage, do nothing, separate, don’t separate, move it around, blur it … do a number of combinations .. and render as interlaced. If you want to check the field lines on rendered test footage, re-interpret its field sep as “off”. Ideally, you should also view the result on an interlaced monitor.