Interlaced Material... Driving Me Crazy

6 testclips later:

No matter what I set there, all clips are progressive.

Okay, so...
Let's asssume for a moment that Compressor cannot create Interlaced MPEG-4 files.
I will now have to get back to trying to de-interlace my material.

De-Interlacing is imo the last resort.

Can you set the "Field Dominance Override" in FCPX to "Un-modified" and export a short video directly from FCPX with the "Video Codec" set to "Current Settings" and see what you get?

For future reference I can confirm that the "De-interlace:" option in Compressor only (should only) happen when the "Output Fields:" are set to "Progressive".

I just taken some time to look into this. Going through Compressor I can't get your file to export as Mpeg4 interlaced either, no matter how I set it. And this issue doesn't relate to the MBAFF interlacing as I got the same results with other interlaced source footage and, Compressor still goes progressive no matter what.

Testing with these files was done sending to Compressor from FCPX and opening directly within Compressor to the same results, so it's not an FCPX issue. But, exporting these files directly from FCPX as H.264 went progressive too.

I've no need for interlaced video with those codecs, so to me it's not an issue, but something has me curious, these files all originated on a PC, and when they got changed to progressive they changed very cleanly. But none use MBAFF interlacing, when you next import new footage from your camera into FCPX can you let me know if the same issues arise?

I've learned a few important things for my AVCHD workflow today.
In accordance to what my English teacher one said: "never write 'in my opinion' - everything you write is your opinion", I will skip that phrase. The following WORKS and makes very much sense to me.

a) The AVCHD directory structure...
...is technically NOT needed for video editing. The additional files do not contain additional information that is needed to deinterlace or to decode a video. MTS file stream files can be played, used, edited independently (as I said, Windows 7 can play them natively, and a lot of software can edit them natively).

So this leaves us with the problem on an Apple computer, I want to open MTS streams without recoding them first.
(If anybody can make a valid proven point that recoding is better for the quality, I'm listening. I have until not seen a good explanation of why it would enhance quality.)

b) What ClipWrap does!
...is to remux the stream from the MTS container into a MOV container, WITHOUT changing anything in the original video file.
i) I have checked the files on a binary level and it is true.
ii) I contacted the support, they confirmed.

Then why does a "rewrapped" MOV file display itself as 1888x1062 instead of 1920x1080?
The reason for this is a metadata entry. The video image itself is NOT changed. NOT cropped at all. ClipWrap support referred me to this Apple specification . I have to say, I think thats a totally waste of resources, this functionality.

But, for me:
My rewrapped AVCHD clips are now fully usable on an Apple computer. No problems, no odd stuff happening. Very happy with this step.


What's left...
...is a very old problem for editors. To De-Interlace or not to De-Interlace.

Background:
The camera I had (Canon HF 100) had a shitty quality. It's that easy. Lot's of artifacts, very blurry. Not good. (Yes, I set it to high quality FPX, 17 Mbps, as high as it can.)
But I left it to shoot interlaced, I should have set it to progressive. With a flat screen TV, no one needs interlaced.

But that's what I'm dealing with now.
The best quality can be achieved if the material stays interlaced. So that's what I wanted to do. Two options:

Find a de-interlace function within FCP X or compressor that leaves a satisfying image quality, end export directly to MP4.

Or, export interlaced to Apple Pro Res 422 HQ, take that huge file to my Windows PC, and recode from Pro Res to MP4 there. Because THAT WORKS.
I will remove the links to my sample downloads (and the files from S3) now.

I thank you very much, NocoDave for your input in this thread. I will browse through the pages later today to apply the bona fide "Thanks".

Thanks BenB, too. You've kept me on my toes and pushed me to find a solution.

Dear Apple, I'm still disappointed. Very low-level consumer oriented software. I have now 20 days left in my FCP X trial and hope that the other functionality (found some cool stuff for surround effects) will make me a happy user.

"Format H.264, 1920x1080 (1888x1062)"

1888x1062 is a display size and not the way the media is processed.

You can deinterlace in FCP either on a clip by clip basis, on a project basis, or you can deinterlace in Compressor. Do not use the filter. Use frame controls.

The best quality can be achieved if the material stays interlaced. So that's what I wanted to do. Two options:

Find a de-interlace function within FCP X or compressor that leaves a satisfying image quality, end export directly to MP4.

I'm finding de-interlacing with compressors frame controls works very well. Something else I'm looking into is this Macroblock-Level Adaptive Frame/Field interlacing. I can't find much about it, but picked up (as yet unconfirmed) that Freeview HD here in the UK use it for transmission as it can carry both interlaced and progressive content in the same encoding.

I'm wondering how this relates to being captured in camera, and if it's possible that material shot interlaced could have the benefits of de-interlacing with zero loss in quality.

If anyone knows, or can point me to somewhere with more information it would be greatly appreciated.

So this leaves us with the problem on an Apple computer, I want to open MTS streams without recoding them first.
(If anybody can make a valid proven point that recoding is better for the quality, I'm listening. I have until not seen a good explanation of why it would enhance quality.)

I think I'm with you on that now, but as before, for me it's not an issue. Recoding can't really improve the quality of your footage, but it can keep it the same and recoding from Long GOP to intra-frame will make things a whole lot easier on your system, that's the main benefit.

Something else I should have added to that last post. While re-encoding won't make your clips any better quality it will stop them becoming worse if your doing anything other than straight editing, particularly with colour work.

That's interesting. But why?
Will 'optimizing' during import suffice?

Optimising during import is what your looking for. Long GOP video doesn't hold all the relevant information on a per frame basis and refers to other frames for information that doesn't exist within certain frames that you may be wanting to make changes too.

The biggest issue is that AVCHD works with a 420 colour space, optimising to ProRes422 not only builds all the missing information back into each frame but changes it to a 422 colour space. It's not the same as capturing 422 but is more solid to work with. If your doing some pretty extreme compositing then ProRes4444 offers even more control over adjustment, but to what extent 420 upraised to 4444 over 422 has I'm not sure. I know 422 to 4444 is worthwhile.

Tom Wolsky wrote: You can deinterlace in FCP either on a clip by clip basis, on a project basis, or you can deinterlace in Compressor. Do not use the filter. Use frame controls.

Hi Tom,
thanks for that, but I cannot find deinterlacing anywhere. All I see are the frame controls where I can change the field dominance - is that an implicit deinterlace function?

Here's my workflow:

a) Rewrap with ClipWrap
b) Edit in FCP X (maybe optimize during import, haven't decided yet)
c) Export as ProRes 422 HQ
d) Convert that on my Windows machine to a pure, but interlaced, MPEG-4 movie.

So far all de-interlacing tests have been disappointing, so I'll stick with the interlaced format when I have the sources. Newer material which I filmed progressive is not a problem at all, of course.

So: 4 days before my trial period ends, I'll buy the license.