So far, so good. Shake composites the individual layers very well for straightforward merge modes (add, multiply, etc). Some of the modes in UF seem to be non-reproducible in Shake. For example, “Difference” would appear to be equivalent to ISubA (Subtract, but take the absolute value) yet the visual discrepancy is obvious.
[Update: see my entry on FinalCut compositing that solves this problem.]
I’m excited by the results. The ability to change parameters in merging layers in Shake is huge — it’s way faster to to these things (both in terms of composition as well as rendering) in Shake than in UltraFractal. It also sets me up to do Study #4 (changing the Shake parameters in time with the music) in a very powerful way that would be difficult and time consuming to do directly in UF.
This makes me more committed to leveraging UF as a composition tool, but leaving all the details of final production outside of it. Define your keyframes in UF, but then let the details roll with other tools, and be willing to accept differences in final output as serendipitous inputs to the creative process.
For example, I took an image that ships with UF (Tendrillon, by the incredibly talented Janet Parke) and rendered it flat, and each of its layers individually. I brought them into Shake, and made the composite as illustrated here:
The first set of nodes, with the IMulti, reproduced absolutely perfectly. There was zero difference between the results produced within UF and with Shake. The second set of nodes, layering in a Difference, produced something different — but not necessarily worse or bad. I like both versions, and would be very open to simply rolling with the Shake version in a production. Shake has a cool way of comparing two different nodes at the same time, so you can see the Shake version on the top, and the original UF version on the bottom:
The Shake version makes the dark spots turn white, and the purple area turn blue. Ok, I could live with that. It may not be what Janet had in mind, but it’s still very nice.
Here’s the final study render. It has zero artistic value, but does illustrate the possibility of animating layers outside of UF. Note that this animation was done with a single frame from UF, just rendering separate files for separate layers, then animating them with Shake:
Finally, here’s the Shake recipe to do it, and a picture of the animation curves:
This study is good enough to proceed forward with other things. I still have a challenge resolving some of the UF blending modes to Shake’s, but basic things are good enough and fancy things will simply require dinking around with single-frame tests.
