[Peter Whittaker]

Hi Dan,

Thanks. I’m not a scripter at all, so I might have to finally learn!

[Peter Whittaker]

Sorry for not getting back sooner, was isolated from the interwebs!

Wow Filip, impressive. To clarify, what happens is that the names themselves don’t move, but just change opacity. The opacity change moves up the screen, and each name fades back to, say, 10% opacity.

The speed of the “scrolling” is about 15 names per second, so based on your calc that’s 70 mins / 15 = ~5 mins total comp time.

Some backstory: this is a list of names and dates of death. We have it running in Quartz Composer in OS X. We have to do something similar with a layer of pre-rendered text for the background munge, but its the actual names in the right positions, just copied and pasted from the Quartz composition into Photoshop in smallish chunks.

Re the assumption that the X-pos isn’t important, you’re mostly correct. The interesting thing we’ve found from having an accurate X-pos per name is you see changes in density across the screen, showing as lighter or darker bands. This adds some natural variance to the texture of the screen, and also shows that, for instance, more people die in winter than summer.

[Peter Whittaker]

wow, incompatible with PS3 beta. GG adobe. I’ll wait until they sort that one out before going near it….

[Peter Whittaker]

well what was the solution?? You know, for others who might find your question via search…

[Peter Whittaker]

well whaddya know, AE is stupider than I thought. Turning off mblur for the static layer saves me another 8-10 secs on a 60-sec short render.

So maybe it is worth splitting everything. That said, I could stop benchmarking and I’d have it all rendered by now!

[Peter Whittaker]

Well the effects start on each layer, fade out, then come back in at the end. So I’d need to do two splits for each layer. It’s my own damn fault for being too tricksy!

>Motion blur’s another big-time render hog.

Hmm, I’d assumed if the layer wasn’t moving it wouldn’t process it.

*runs off to find out*

[Peter Whittaker]

Good idea!

..but I’d need to set up 24×4=96 different render jobs.

The alternative is to just hit Render once, and go home for the night.

Which would you pick? 😉

[Peter Whittaker]

Dave, I’m using image sequences for a lot of the precomps, but I’ve been sticking with QT Animation files for final output. The final QT is 73 gigs: I haven’t tried a PNG sequence but IIRC it’ll be a lot larger. Second catch: I have a lot of my stuff prerendered, and in that case the render itself is a lot faster if you output to a single file format rather than a sequence (i.e. its mostly I/O rather than effects processing).

3rd catch: using QT Pro to stick image sequences seems to take FOREVER! Now I use AE to stitch em, at least it gives me a proper progress bar.

In short, I’m using a mix of the two approaches. I definitely see the benefit of rerendering only a small sequence of images rather than the whole lot, and its saved me a few times already on this project. Maybe I’ll just go the whole hog and see how big the final PNG sequence would be. Need to get a bigger HD though! 🙂

[Peter Whittaker]

hmm, nice tip…trying it now. I’ve been also trying Episode Pro, I’ll report back with results!

Thanks,
Peter

[Peter Whittaker]

I don’t feel comfortable unless I know the machine is slaving away while I’m at home 🙂

In this case though cutting the effects down means I can get two 20-minute sequences rendered in one night rather than just one. Given how close I’m working to a deadline, that might be a lifesaver! Now if only AE’s H264 encoder was as good as plain old Quicktime Player…