Metaphase Variation 1
The first variation on the "Metaphase" image originally discovered by Paul Derbyshire.
There is a lot to say about the background of this video and some of the technology that went into producing it. Please see the Comments below.
MP4 Files: | 320x240 6.3 MB fast download 320x240 24 MB preview quality 640x480 118 MB DVD quality 640x480 281 MB master quality | 192-384 Kbps 786 Kbps-2 Mbps 4-10 Mbps 10-20 Mbps |
WMV Files: | 320x420 6.4 MB fast download | 192-384 Kbps |
All these files were encoded with variable bit rate encoding to produce the smallest possible file size for a specific video quality level.
Date Generated: | 12-27 Aug 08 |
Final Image Size: | 1e-99 |
Resolution: | 640x480 |
Video Length: | 3:20 of fractal, 3:47 total |
Frames: | 6000 |
Rendering Time: | 360 hours |
Method: | Escape counts Adaptive color optimization |
Audio: | Acid Pro 6 custom composition |
The original Metaphase image led to an earlier deep zoom. The same overall zooming strategy, applied to different areas around the set, can lead to many other very beautiful zooms (I've so far found about five that I think are worth making into videos ... stay tuned!). This video shows the first such variation I've made on the original idea, followed by a pullback and very deep zoom (the final frame size is 1e-99) into a mini-brot in one of the "chromosomes".
The actual "Metaphase" image is seen at time index 2:15 in the video. In the sample frames above, it's the last image in the first row.
By zooming into an area at the extreme tip of a structure in the seahorse valley, this animation incorporates a theme of very long twisted fingers touching the central structures at each progressive level of zooming.
I kind of randomly stumbled upon this great c olor palette with these gorgeous deep earth tones, something I've not used much before, and I loved it. There's a delicious copper-brown that blends into a luscious dark forest green, followed by burnt orange, and a contrasting, cool, slightly bluish-green, all separated of course by blends into white alternating with black.
This is the first deep zoom to use my newest and most powerful adaptive colorizing method.
The problem that this colorizing method addresses has to do with finding the best way to convert the raw fractal data to colors in a way that shows maximum detail in the images while maintaining smooth transitions of color over time.
This animation turned out to pose new challenges for colorizing that required more sophisticated techniques than I have previously used. There are two major problems. First, this animation has a huge difference of dynamic range of the raw data. That is, in some frames, the range of raw data is a factor of many millions, while in other frames, the range of the data is only a factor of 2 or less. That span of raw data has to be mapped onto the color palette in a smooth, visually appealing way.
The other problem is that the raw fractal data is very noisy from one frame to the next. Any naive approach to fitting the color palette to the fractal data will result in color mappings that show unacceptable high-frequency jitter or low-frequency warble. These effect can be seen in a short test animation I created, RankOrderTest.
The new method used for this animation uses a global smoothing technique to generate a time-dependent color palette mapping function that smoothly adapts to the raw fractal data at each frame.
The bottom line is (does anyone recognize who I'm paraphrasing?):
This produces a taste, a smoothness, and a drinkability you will find in no other fractal animation at any price.
This time all the compressed videos use variable bit rate encoding. I think this gives a slightly better quality for the file size.
This is also the first time I've been able to get the MP4 and WMV files to be essentially the same size and the same quality. That said, I think both technologies fail badly at encoding the broad, smooth gradients forming the background of this and most of the other videos. Both MP4 and WMV are designed to track blocks of motion on a stationary background, and these changing, smooth color gradients really stress their algorithms to the extreme. You can see severe blocking artifact in all the videos. The codecs I am using are high-quality commercial products, so I am fairly sure these artifacts are intrinsic to the technology.
The short pull-back that starts at around time index 2:14 is the first utilization of the improvements to my animation scripting that I've been working on since May. I'm now able to smoothly transition from one key frame to another using a text file script, I can save, load, and edit the animation sequence from within the rendering program. This doesn't sound like much, but it has finally made complex story-boarded animations possible for me, and there will be some really cool videos coming with sequences of zoom-ins, pullbacks, and panning movements. I've previously made a couple, HiPrecPanTest and PanTest2, but they were very difficult due to the clumsiness of the software.
The music for this video was composed using loops and samples from Sony's Acid Pro libraries. This is the most elaborate such composition I've made, with 22 separate tracks mixed together to make the final music file.