Welcome to Fractal Forums

I finally got this working at decent framerates last night. Rendered on an nvidia gts250 gpu, compressed with x264 simultaneously in realtime.

All movement control is done interactively on the keyboard, no splines or keyframes! (think: Descent (http://en.wikipedia.org/wiki/Descent_%28video_game%29))

http://www.youtube.com/watch?v=hN8vOCK6QPw

 - ld

Here is another recording showing the limit of 32-bit floating point precision. At the beginning of the video the camera position runs out of precision, causing jitter, and at the end you can see the surface turning into voxel cubes. This is as zoomed-in you can get without doubles.

Almost as if somebody planted all those trees there, in an obvious formation...

http://www.youtube.com/watch?v=6W30MbpEBU0

 - ld

So this is moving around in realtime? What kind of spec machine are you running this on?
Seems like it would make an ace plugin for Mandelbulb3D for navigating around before rendering a high-res version (as well as being great fun by itself obviously!)

Intel i5-750 (mostly spent just encoding the video) with an nvidia geforce gts 250. I'm also running it on another machine with a core2 cpu and a geforce gt 220, where it's somewhat less smooth but still nice.

But it's all about the controls.

Edit: one more thing; the videos are somewhat choppy because the rendering framerate is not exactly 30fps, but the rendered frames are forced to that frame rate when encoding the video.

Nice, i wish i know how to implement that!

to increase resolution, perhaps it would be possible to simulate multicore on the graphics card. if the framerate can go to 100 for example, you could multiply the screen into 4 by using graphics to cache the frames and combine them.

hope that you checked out kali s channel on shadertoy, it's full of amazing ways to optimise and refine gpu rendering with help of shadertoy community.