Question about HD to SD and interlace>progressive>interlace

[ayim]

Hey, I'm wondering how I would go about resizing my 1920x1080 60i footage to broadcast SD size, I'm guessing 720x480 and I'd have to burn in to a DVD then. But the footage is interlaced, so I'm going to need to deinterlace before resizing right? I have After Effects and I'm wondering about this quote here:

"If you want to use interlaced or field-rendered footage (such as NTSC video) in an After Effects project, you get the best results if you separate the video fields when you import the footage. After Effects separates video fields by creating a full frame from each field, preserving all of the image data from the original footage.

Separating fields is critical if you plan to make significant changes to the image. When you scale, rotate, or apply effects to interlaced video, unwanted artifacts, such as crossed fields, are often introduced. By separating fields, After Effects accurately converts the two interlaced frames in the video to noninterlaced frames, while preserving the maximum amount of image quality. Using noninterlaced frames allows After Effects to apply edits and effects consistently and at the highest quality.


When you render a composition containing field-separated footage, set the Field Rendering option to the same field order as your video equipment. If you don’t field-render the composition, or if you field-render with the incorrect settings, the final movie may appear too soft, jerky, or distorted.''

Is seperating fields in AE different from deinterlacing? Because wouldn't deinterlacing 60i to 30p, then interlacing 30p to 60i lose smoothness from original source?

But anyway to get it broadcast standard, would it be simple as seperating fields in after effects, resizing to 720x480, and then rendering out mpeg2 with proper field so its 60i again? The footage is 16:9 aspect ratio though, and 720x480 isn't so wouldn't it distort the footage? kinda confused on how to do this

[poisondeathray]

Is seperating fields in AE different from deinterlacing? Because wouldn't deinterlacing 60i to 30p, then interlacing 30p to 60i lose smoothness from original source?

It's basically deinterlacing for your display. If you put it on a 29.97 fps comp, it will single rate deinterlace. If you put it on a 59.94 comp, it will double rate deinterlace (often called bob deinterlacing) and you keep the temporal resolution

But anyway to get it broadcast standard, would it be simple as seperating fields in after effects, resizing to 720x480, and then rendering out mpeg2 with proper field so its 60i again? The footage is 16:9 aspect ratio though, and 720x480 isn't so wouldn't it distort the footage? kinda confused on how to do this

That would be too easy. Adobe majorly screwed up here for several reasons:

1) It's deinterlacer produces low quality ; it leaves many artifacts and "jaggies". AE's deinterlacer with "preserve edge option" on is better than the regular AME or Premiere deinterlacer, but still pales in comparison to avisynth methods

2) It's resizing algorithm isn't very sharp. I think it's either a bilinear or bicubic resize

3) It doesn't re-interlace either using AE's render queue, or if you export a 60p or 30p lossless/uncompressed footage into AME. AE's render engine is more limited and doesn't even re-interlace at all. AME dupes the frames instead of properly re-constructing the fields (should be 60 moments in time, instead of 30 with studdering repeats). You end up with a temporal resolution of 29.97 either way = bad.

If you want more information and comparisons look through these posts:

https://forum.videohelp.com/topic373212.html
https://forum.videohelp.com/topic371951.html
http://www.precomposed.com/blog/2009/07/hd-to-sd-dvd-best-methods/

If you want to do it the "easy" way, just resize in premiere pro, but you will get a very soft image and artifacts.

[ayim]

Hey, thanks for all the information! I just got back from a trip and didn't have time to check here or anything.

So if I'm understanding correctly, if I want to add effects, titles etc to 108060i file in After Effects I should deinterlace it to 108060P so I wouldn't lose any of the frames when re-interlacing it?

So I would go Raw 60i.mts > Avisynth+vdub > Bob Deinterlace to 60P lossless > After effects > lossless > Reinterlace to 60i
And if I wanted I could add the resizing option durring deinterlace right? Oh, and could I utilize the 60P and make some slow motion before reinterlacing, or would that do some weird stuff?


A few quick questions:

How much quality loss do you get from deinterlacing? And doesn't all movie productions have to deinterlace to add the effects?

What's the best script to reinterlace the footage? And how would I deinterlace it in the first place to 60P

I tried understanding what was doing what in your script and I'm guessing these are deinterlacing to 60P right?

"Directshowsource("20090912161022.m2ts", fps=29.97, audio=true)
AssumeTFF()
interp=NNEDI2(field=3) #double rate, starts with top
YadifMod(order=1, field=-1, mode=1, edeint=interp) #TFF,field set to order,double rate"

Also thanks for that UT video codec, I'm going to have to try that, anything to speed up editing is a gem.

Thanks to you I won't have to buy cineform lol

[poisondeathray]

So if I'm understanding correctly, if I want to add effects, titles etc to 108060i file in After Effects I should deinterlace it to 108060P so I wouldn't lose any of the frames when re-interlacing it?

Yes, you have to bob deinterlace to 60p; if you standard single rate deinterlaced to 30p, you would lose 1/2 the temporal information , and you wouldn't be able to re-interlace it later for your DVD

So I would go Raw 60i.mts > Avisynth+vdub > Bob Deinterlace to 60P lossless > After effects > lossless > Reinterlace to 60i
And if I wanted I could add the resizing option durring deinterlace right? Oh, and could I utilize the 60P and make some slow motion before reinterlacing, or would that do some weird stuff?

Yes , for "best practices", I would bob deinterlace to 60p, do your effects, overlays, slow motion etc..., export as 60p then resize and reinterlace in avisynth. You can take any number of "shortcuts" along the way, but each shortcut you take will impact the quality. If you go that that last page in that link above, you can see the difference between an "all Adobe" workflow, and a more complicated avisynth workflow. For some people, time maybe more important, but at least you have some choices here.

AE's scaling algorithm is decent, but you have many more options and control in avisynth. Avoid AME at all costs for scaling and cropping. You can see some comparisons here:
https://forum.videohelp.com/topic376711.html#2030745

How much quality loss do you get from deinterlacing? And doesn't all movie productions have to deinterlace to add the effects?

Deinterlacing ALWAYS causes quality loss. Higher quality deinterlacers will give you better results. Again avisynth has dozens to choose from, from very fast, lower quality, to glacially slow high quality ones. You can see some examples of deinterlacing differences below, but there is always a trade off between quality and speed:

https://forum.videohelp.com/images/guides/p1934885/stockholma_0-520_q3_yadif_mvbobmod_t...mca4_tdtmm.avi
some screenshots comparisons here:
https://forum.videohelp.com/topic376565.html
https://forum.videohelp.com/topic354397-30.html

What's the best script to reinterlace the footage? And how would I deinterlace it in the first place to 60P

There is no "best". Each source is different, and some deinterlacers work better on some sources than others.

I tried understanding what was doing what in your script and I'm guessing these are deinterlacing to 60P right?

"Directshowsource("20090912161022.m2ts", fps=29.97, audio=true)
AssumeTFF()
interp=NNEDI2(field=3) #double rate, starts with top
YadifMod(order=1, field=-1, mode=1, edeint=interp) #TFF,field set to order,double rate"

This would work for AVCHD, you might have to make some modifications depending on your source footage. Also , another deinterlacer might be "better" in your specific footage.

Note in that thread, some of the advice was very specific for foochuck's case. e.g. there was a field swap because he wanted to match BFF footage with DV. You wouldn't do that field swap if you didn't have BFF footage.

Also thanks for that UT video codec, I'm going to have to try that, anything to speed up editing is a gem.

Thanks to you I won't have to buy cineform lol

This is still significantly slower than cineform. The difference is that it's mathematically lossless. However, in AE with effects etc. you usually never render at full speed, regardless of format anyway

Cineform is still significantly faster for "normal" editing, eg. in vegas or premiere

[ayim]

Dang this has got to be the best site I ever visited, thanks again for the heaps of information!