Workflow for Deinterlacing for Vimeo

[Michael Gissing]

As you have interlaced material and a digi beta broadcast deliverable, I would stay with an interlaced sequence.

To solve the title issue, you could export just the textless from a copy of the final, use Compressor to create a deinterlace QT and then drop that back into a progressive sequence and add back the titles. Export that progressive sequence for the Vimeo compression.

A short test should indicate if that helps with both image and title quality

[Jon Smitherton]

Thanks Michael.

I suppose I could put the Text on V4 say (would I have to do this for graphics as well?), then nest video in a interlaced video sequence, de-interlace and change the original sequence to a de-interlaced one.

Only thing is that I’ve got different field orders – upper,lower and none…if I remember properly – de-interlacing effects the progressive footage and gives it jaggies.

Unless anyone has got a failsafe compressor setting, I might have to get fieldskit.

Thanks,
Jon.

[Michael Gissing]

Don’t nest and deinterlace in FCP. You get a better result doing the deinterlace in Compressor to a self contained QT file and bringing it back into a progressive sequence. Then add your text. This should also minimise rendering.

[Julian Engels]

Hey Jon,

we shoot a lot on DigiBeta interlaced and output to web formats. So we always have to deinterlace. We do this with compressor. To reduce render time while maintaining quality, we set the sequence to progressive and send the final edit to compressor for deinterlacing. We found that render time is half when field dominance set to “none” compared to render times of an interlaced sequence I can’t see any difference in quality, when using this method. If you do it this way, you have to make sure to set all the frame modes of each clip to what it was actually shot (DigiBeta clips to interlaced odd, HDV/DV to interlaced even, progressive material obviously to progressive etc.) Text overlays are automatically set to progressive (since sequence is progressive) and therefore will not be deinterlaced and should look nice in final output. On the other hand, if you set the field dominance of a progressive clip to interlaced it will look awful in final output, same with field dominance to a incorrect field (odd/even). With “Edit” -> “Object Setting” (from memory) you can check and correct if you have to.

Hope that helps,

Julian

[Russ Carlin]

Hi all,

When you say send it to compressor to deinterlace, are you just meaning that you’re compressing it to whatever output you need and checking a deinterlace box at the same time, or do you sent it their just to be deinterlaced?

Sorry if that doesn’t make sense, or if it’s a stupid question!

Carrrrrrr Carrrrrrrrrrrrr

[Jon Smitherton]

Hi Guys,

De-interlacing reduces resolution especially with the de-interlace filter in FCP – it halves the vertical resolution. Watch on a broadcast monitor when you apply the filter. It is also a stylistic decision as the motion is not full – this may or may not be desired.

To get full resolution with de- interlacing duplicate the clip on top – so same clip on V1 and V2 – then add shift fields +1 to one and shift fields -1 to the other with a 50% opacity on the top clip.

For this reason – too much work for this technique and wanting to deliver a Digibeta of the film for broadcast – we decided not to de-interlace. Which leaves us with my original post – how to de-interlace with text looking good.

Michael, thanks have been doing some experiments with RE:Visions field kit and it looks great. Thanks for the tip though!

Cheers,
Jon

[Michael Gissing]

Jon. the Nattress deinterlace plugins do a better job than the manual clip stacking that you describe. They blend fields with control and do not just throw resolution away. I prefer that method to Compressor but that is because I can tweak individual shots with the plugins, but a lot more time consumming and render making.

I agree about the need to balance the look of deinterlace with motion smoothness.