Intermediate CODEC Workflow
by Dale Paquette
on
Jul 9, 2009 at 4:21:33 pm
I have read a bazillion posts on this general topic, I apologize for rehashing a bit. The consensus seems to be that doing effects work on mpeg2 is undesirable at best due to its inter frame compression. However, that's what I have and that's what I would like to do. Is this a good procedure? Open an mpeg2 file (the HV30 actually outputs m2t and I assume its HDV mpeg2 inside). Anyway, it looks good on the PC, even plays back fairly close to real time in AE. I understand the difficulties of inter-frame compression. So, I output the footage using the animation codec (or some other) and end up with a very voluminous mov clip. I then delete the mpeg2 version and work with the mov version (all set for best quality btw) to do effects based work. In other words, I assume that the mov version has decompressed the inter frame compression. When I'm done, I can render out to mpeg2 for my NLE (which is already mostly populated with mpeg2). Guess my question is does the file>export route decompress mpeg2 inter frame video so that AE can simply work with the decompressed footage one frame at a time? Thanks.
Re: Intermediate CODEC Workflow by Dave LaRonde on Jul 9, 2009 at 4:52:15 pm
[Dale Paquette]"...The consensus seems to be that doing effects work on mpeg2 is undesirable at best due to its inter frame compression. However, that's what I have and that's what I would like to do. Is this a good procedure?"
Despite whatever you would like to do, the answer is no. It's not.
[Dale Paquette]"...I output the footage using the animation codec (or some other) and end up with a very voluminous mov clip..."
What are you using to do this outputting? It should NOT be AE, but a different application. Otherwise, you won't be accomplishing anything, because AE doesn't work well with this kind of footage.
And yes, the file sizes will be enormous compared to what you're used to. But then, HDV codec compress the living daylights out of video, making them EXTREMELY lossy. The file sizes may be tiny, but they won't look any dood if you re-encode them, as you propose to do.
"High Quality" and "Small File Size" are mutually exclusive terms.
Dave LaRonde
Sr. Promotion Producer
KCRG-TV (ABC) Cedar Rapids, IA
Re: Intermediate CODEC Workflow by Dave Johnson on Jul 9, 2009 at 6:08:11 pm
In addition to the things Dave pointed out, it'll also be important to understand that it isn't possible to "decompress" compressed video since that would mean somehow regaining all of the data that was discarded during compression. In fact, as far as I know, there's no such thing as "decompress" in the context of digital media files ... there's "decode", but that's different in that it just refers to the computer using the proper codec to read/play or convert an encoded file, which doesn't involve regaining discarded data.
Re: Intermediate CODEC Workflow by Dale Paquette on Jul 9, 2009 at 7:28:27 pm
Thanks. Perhaps I didn't ask the question properly. I understand that "regenerating" data in between the keyframes of mpeg2 does nothing about improving quality. If I don't have the data to begin with, I'll never have it. Rather, the idea is to enable AE to do effects work on a particular (non key) frame without having to reference to a previous and/or future keyframe. In this way, I would think the render process would be much "faster" despite the bigger files. I agree, no increase in quality.
I also misspoke about rendering out to mpeg2 for use in the NLE. I don't do that. I use the somewhat lesser capable Elements 4 but it seems to work well with whatever format the footage is in. So, "non-affected" footage stays with its native mpeg2 compression and I edit it together with what ever I render out of AE. I go back to mpeg2 only from PE4 when I render from there so that I can burn to BD. Other than titles, I don't do any effects with PE4.
So, would the workflow I described do anything wrt speeding up the internals of AE? Thanks again.
Re: Intermediate CODEC Workflow by Kevin Camp on Jul 10, 2009 at 5:34:07 pm
[Dale Paquette]"ould the workflow I described do anything wrt speeding up the internals of AE?"
yes. converting any inter-frame compressed footage to a codec that uses no compression or only intra-frame compression will speed things up (noticeably) in ae. it will also be a bit more stable and you'll have much less frequent issues in your final renders.
note, you don't have to go lossless if you can't afford the drive space, you just want to avoid codecs that use inter-frame compression. but, by going lossless, you'll avoid further loss in image quality.
Re: Intermediate CODEC Workflow by Jesus Ali on Jul 10, 2009 at 11:29:14 pm
If you were working on the Mac and Final Cut Pro, you could Transcode the HDV footage into Apple Intermediate Codec or (preferably) into Apple ProRes 422, or even ProRes 422 HQ (10-bit).
Yes, you don't re-introduce quality that was lost in the HDV compression, but you have a better color space (4:2:2 in PR vs. 4:1:0 in HDV) in which to process special effects, transitions and color correction.
Working in a compressed color space is like multiplying a grid of long decimal numbers but only being able to use Whole numbers as their result. Fidelity and Granularity is lost.
I don't know anything about the Elements NLE, but I heard that Sony Vegas has a format that's similar to ProRes and can be transcoded to on Ingest (capturing to computer from tape).
Maybe it is called, "Cineform" ?
If your NLE can transcode to DVC PRO HD on ingest that would be a better format than HDV, but still more compressed than Animation (Lossless).
Also, are you working in the HV30's 24P or 30P mode? I've read on the Cow that After Effects may be able to remove the Pull Down from this footage after it is captured as 29.97. Working in Progressive is really great. You can set your AE Comp to 23.976 fps and save rendering 6 frames every second and the results and effects look spectacular.
Re: Intermediate CODEC Workflow by Dale Paquette on Jul 11, 2009 at 1:30:30 pm
Jesus, thanks, I'll look into your thoughts and ideas. I do shoot in 30p as that solves, I believe, some problems with quality loss in the de-interlacing process (as I understand it!). As for 24p from the HV30, it does indeed shoot 24p but apparently uses a "proprietary" algorithm to handle 3:2 pull-up thus exporting it as 30i. I have experimented with all of the pull-down options in AE to no avail. The HV30 forum does have a pull-down hack but their version requires a bunch of open software and some java software writing. Since I am not striving, at least now, for the ultimate cinematic look, 30p is fine. I do do my work on windows PC so I'll have to check into what is available besides animation for intermediate work. I also need to look into PE4's options, but I believe they are limited. That said, I have found plug-ins for other specialized functions such as 30p capture that are not publicized by Adobe. Indeed, the ones I found and have used successfully are for Premiere Pro. I do have CS2 for that and, who knows, there may be some there I can use. I'll keep all posted. Thanks again.
