-
Rendering long credits – best practices?
I’m doing a rather long credit crawl sequence, aimed ultimately for DVD production. As a first stab, I made the mistake of putting the text of the entire crawl as one long block of text into a Sony “Text” generated media event, continuous for over 2 mins in a 2min 20sec event. The long text section is keyframed over most of the middle 2 mins, mostly panning slowly from top to bottom.
The result was a 2.5 min patch of video that took an hour and a half to render on a Quad Core machine (at 640×360 resolution in preview!! quality to an mp4 for upload to vimeo for proofreading). I’m guessing the slowness is because i’m doing the equivalent of pan&scanning small segments of what probably would amount to a 300MB still image?
Here’s the result, if you care to look, a few huge glitches in the transition from a smaller one-line text bit to the big roll of names and such. I know the badness is there, but I needed to get my collaborators something they could actually proof for typos in the names, titles and general layout.
This is a machine that usually renders most video at close to actual running time or faster for some streaming resolutions, and takes maybe 10 mins to render a fairly complicated piece of generated media (such as the ProType Titler sequence from the opening credits of the same project). That project (maybe a 10 min render?) is here:
It still needs some tweaking too, but the 10 or even 20 min render I can live with there, if it’s paid off in apparent production value. An hour and a half tacked on to what will ultimately be a fairly long render without the end credits, not so much… 😉
I’m thinking this may speed up if I break the text into smaller chunks?
But I’m also wondering in general what people think are best practices when creating a very text heavy credit sequence like this one?
I chose to avoid the “Credit Roll” generator, by the way, mainly because it tends to be so inflexible when tweaking individual lines and sections of text, and because (I’m lazy) it would take a long, long time to cut and paste all the necessary bits of text from the document I’m using as source reference, into individual “cells” in the credit roll generator.
If I can’t shorten the render time significantly by breaking the text into sections, I suppose the next step will be to pre-render the finished text sequence alone to a clip that I can lay in on top of (or underneath) other elements?
Again, I’m looking mainly for war stories and novel ideas that others have found for making the render time more reasonable without sacrificing too much resolution.