DV in AE

[Bret Williams]

Always interpret lower field first for best results, unless you know for a fact that it’s progressive footage. It won’t make a difference if it is, just takes longer to render.

Render out as lower field first as well to duplicate the look of the original. If you want that strobey look of fake film look progressive, then render with fields off. I’d make sure you turn on motion blur for all other layers in your comp, which of course takes longer to render.

You mentioned rendering uncompressed, then putting that in a FCP DV timeline. That would of course be actually worse quality than rendering to DV. You’re simply running your footage through an extra compression step, then rendering the outcome of that back to DV. But probably visually identical. Just longer rendering and bigger files and who needs that.

In any case… always interpret your footage correctly and then let AE deal with it intelligently. Output to fields if you want smooth motion and frames if you want progressive output.

I never understood the confusion. If you’re getting soft footage in FCP, are you repositioning the clip vertically in FCP? If so, you have to keep the position to an even value for Y. 0,2,4,etc. FCP doesn’t do subpixel rendering. It claims it does. It doesn’t.

[Steve Roberts]

If your interlaced footage is not going to get rotated, blurred, scaled, keyed, moved or distorted, you should interpret it field separation OFF. Yes, off. In situations where the footage is going to be minimally altered, if at all (maybe color correction only), I’ve found that separating fields introduces some softness into the footage. It shouldn’t, but that’s what I’ve found. You should do your own tests with unaltered footage.

AE always renders, recompressing the footage. Always. If you choose a lossless codec such as Animation or None, the recompression is negligible … lossless. Most people here prefer Animation as the codec when they want to preserve quality.

AE doesn’t work like an NLE. In an NLE, if you have a DV timeline, and spit out a DV movie, the footage is unchanged if you have cuts-only. But it renders transitions and effects, recompressing the footage. AE is like that, since it renders everything. if you take DV footage and render to the DV codec, AE will recompress to the DV codec.

But Bret is right. You can render to Animation in AE, then import into your NLE, but your NLE will have to recompress the Animation movie to DV to match the timeline. You’re better off rendering to DV in AE so the footage matches the timeline of your NLE when you import it.

[Brendan Coots]

You cannot influence or create the cool, strobing look of 24p film by rendering to progressive vs. interlaced, just doesn’t work that way. It also doesn’t matter one bit whether you render to frames or interlaced, for any reason. It is merely a preference. No matter what you do AE/the NLE will do what it needs to do. The only difference between the two is that some slow moving vertical animations with sharp edges can suffer strobing (not the cool, 24p kind) if they are rendered interlaced. That said, if your show is interlaced you might as well render interlaced, but it’s not a crucial issue.

As for the compression issue, you can’t really call uncompressed an “extra compression step” since there is NO compression, hence the name. Yes older NLEs might push it back to DV, but it is the exact same visual result either way. Personally, I always render out of AE to the Animation Codec and let the NLE do what it may. Also, Final Cut 6 (almost a year old now) doesn’t transcode footage anymore, it allows full format mixing and matching so the Animation/Uncompressed method WILL result in higher quality results.

I would suggest you either render to the animation codec (no loss similar to uncompressed, but smaller files), use the ProRes422 codec throughout your pipeline or use DV/DVCPRO 50, but only as a last resort.

Brendan Coots
Splitvision Digital
http://www.splitvisiondigital.com

[Justin Vaillancourt]

Thank you for all your responses. After reading and doing some more tests, if you just take your DV footage from FCP into AE and render Seperate Fields OFF to DV compression, it produces a file that looks visually identical.

The best way to test this was going into Quicktime and toggling the ‘Use High Quality Setting When Available’ in the preferences. I believe you have to quit the program to get it to take a effect. By toggling this, the fields off footage was matching the original footage exactly, but by toggling the lower field first footage, you never saw the interlace lines from the original footage, it just always looked soft.

As far as compression, my 19 second clip was 66mb in DV and 536mb uncompressed. Seeing that I have to render to DV in FCP anyway (while creating an extra render file) and there was visually no difference in the files, it seems like a no brainer. I haven’t used FCP6 yet, so I haven’t been able to test the open source timeline.

Thanks for the responses!

Justin

[Steve Roberts]

If you ever want to compare two nearly-identical renderings, drop one over the other in an AE timeline, and set the transfer mode of the upper layer to Difference. If you see something, there’s a noticeable difference. (duh) Otherwise, if you see black, the differe3nce is not really noticeable.

To see the amount of difference, with the Info palette visible, roll your mouse over the blackness. If all you see in the Info palette is RGB 0,0,0, then the two layers are identical. The deviation from zero tells you how much the two layers are different. If you want to exaggerate the difference to see it better, try adding an adjustment layer above everything, with saturation boosted through the Hue/sat effect.

This can be used to compare relative lossiness in compression.