My first (modern) attempt at a fractal movie — albeit, only a short study.

The purpose was to “prove out” the various technologies involved, and expose technical challenges. I’ve definitely succeeded, and even came up with something that I think is kind of interesting. It’s on the road to achieving my four quality points that make for a “good” fractal movie. I think it’s at least the equal of some of the fractal videos on YouTube. I’m pleased with the product as well as the process.

Quicktime Movies: 320x180640x3601280x720

Technology

The technology stack that I proved out is:

• UltraFractal 4, running in Parallels on MacPro
• UltraFractal 4 Server, running on some old Windows computers and in Parallels on various other Macs
• Shake
• FinalCut
• Logic Pro
• QuickTime Pro

The newest Parallels is sweet, with its Coherence mode. I now really don’t care that UF is Windows only. It feels almost like a regular Mac program, assuming you ignore the ugly Windows UI stuff. Getting the UF server up was trivial, although I’ve gotten spoiled by the Mac’s “Bonjour” auto-discovery and had to spend a couple of minutes typing in IPs. I’ve shared a Mac directory into the Parallels machine, so all my UF output can be seen from the Mac.

I used Shake to add just a bit of Hollywood gloss to the middle section. Shake has an interface only a unix geek could love — luckily, I am a unix geek, and am starting to learn my way around. I’ve ordered a couple of Shake books which should be here this weekend. That will help me go beyond my really simple start — currently, just a FileIn, a Blur, a level thingy and a FileOut.

FinalCut is still a bit of a headscratcher for me — not that it’s that tough of an interface, just it’s a deep program and I’ve only been driving it for a few hours. For this first attempt, all I did was add some titles, some intro and outro freezes/dissolves, and layer in the audio. Worked great.

Logic was terrific. The work I did back in December really paid off — I fired it up, and started composing in less than 60 seconds. I paid almost no attention to technical stuff, instead just concentrating on music. Lovely! The orchestral library setup I have worked great. It was quite liberating to think “hmm, an english horn doubling the lead violin would sure sound good” and be able to do that almost before I’d completed the thought. The sounds weren’t bad either, even though it’s the $99 Apple Orchestra JamPack instruments. Good value there for sure! As soon as EastWest brings out their orchestral libraries in Play format, though, I’m jumping all over them…

Quicktime itself worked fine for creating a variety of outputs — different sizes, compressions, letterboxing, etc. My biggest problem now is I’m still stumbling in the dark about the right combination of codecs and other settings to make the files playable while minimizing artifacts. I’ve got a lot of learning curve to climb here…

Process

My UF still images still have a ways to go before they’re compelling, but the class with Janet should help. Still, it’s so easy to come up with interesting stuff. I probably put in less than an hour coming up with the key images and roughing out the various layers. The hard and time consuming part was getting the animation right between them. The central piece is the “explosion” and this took about 3 hours of dinking around to get the gradients just right. Add in another 3 or 4 hours to get the various other texturing layers and orbit traps and so forth right… so maybe 8 hours to get the right look and feel and gross “camera” movements.

The worst part of this experience was the timeline editing. The visual discontinuities in camera movement were especially difficult to resolve. I should post more about this in a separate post… but it’s really irritating and makes for an unacceptable quality level in the final product. I never did fix it entirely, although I got close. (In fact, as I’m writing this, the new-and-hopefully-improved version is rendering right now, with smoother but still imperfect camera moves).

I also found UF’s timeline editor to be a pain to do complicated stuff in. I’m not yet sure if it’s my lack of experience or limitations in the program itself, but I imagine a number of things in my mind’s eye that would be *really* tedious to do in UF directly. I think I may well write a behavioral based animation pre-processor that feeds UF; this would keep UF in its comfort zone of extraordinarily powerful and intuitive exploration and “roughing in” while leaving the anal-retentive work to something I directly control. More on this later.

Rendering from UF is both simple and kind of frustrating. Simple, in that it’s all built-in, including network rendering. Frustrating, in that I can’t see exactly what frame is being rendered, I can’t see the results until the render is 100% complete, I have to use the UI instead of being able to script it. I wrote to UF’s author and was excited and impressed that he’d like feedback on this to potentially include solutions in future UF versions. I think I’m going to have *a lot* of suggestions around animation…

Hmm, rereading this makes me concerned that someone would think UF is bad or something. Quite the contrary — UF is the single most powerful and intuitive fractal program I’ve come across. It’s amazing and inspiring. I think I’m just trying to take it beyond its comfort zone, and adding to that is that I’m a newbie with it and it’s probably got a lot of smarter ways of doing things than I’m using.

Anyway, the complete fractal is 1200 frames, with something like 8 layers. It takes about 48 hours to render 720p HD (1280×720) on a MacPro, an iMac, a MacBook Pro, and a Sempron-based Windows PC. Tedious. I render to photoshop files to send into Shake.

I really learned the value of doing “quick” renders of a couple hundred frames at small resolutions with all fancy things like anti-aliasing and motion blur and so forth turned off. This was especially important in trying to figure out the jerky camera movement. I like being able to render directly to an avi so I can see the results without having to dink around with Shake, although I also routinely rendered 1/4 size low quality psd’s so that Shake could use them as proxies.

As far as the music, this was pretty trivial stuff. I dinked around with the harp sound for five minutes to come up with the theme, added too much garbage around it, and then removed the excess. It still is too busy, but it’s just a study, so what the heck. Predictably, it’s in an odd time signature (7/4) — I can’t seem to write anything normal without being really conscious of it. I record this stuff, and then have to count out what I did after the fact.

I *really, really* like having the movie in Logic’s Globals at the top of the arrange page. The “explosion” is the centerpiece of the movie, and I had to make sure the music built up to and declined from this point effectively. You’ll notice the trumpets hit their high notes right on the cue. I initially loaded the full movie (Animation codec, which is uncompressed, at 1280×720) and although Logic could handle it, it caused some glitchiness in streaming the samples of the disk even on my fire-breathing Raid-0 MacPro. It was much happier after I’d used Quicktime Pro to export a 1/2 sized version, compressed with the Apple Pixlet Video setting.

Results

So… overall the movie is interesting. My biggest complaint is the camera movement still isn’t what I want, and I can’t figure out how to do it in UF. Thus, I may well investigate coming up with my own spline-based camera preprocessor.

Here’s the original version, with really jerky camera movement. The versions above decrease the problem, but still aren’t what I really want. I also hadn’t decided on a domain name for this venture when I rendered this first movie. It showed “Fractaliae” but I was considering either “Fractanima” or “Fractanimus”.