this height field thing

  • 86 Replies
  • 2038 Views

0 Members and 1 Guest are viewing this topic.

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #75 on: June 17, 2020, 10:51:50 AM »
I was made aware of Yann Lby on YouTube. That channel has quite a few animated Mandelbrot height field zooms, among other fractal and non-fractal videos. Looks to be a custom height field renderer in isometric perspective. No orbit traps (yet?), but if you liked my few animations here, there's definitely something to see there.

https://www.youtube.com/channel/UC1abJGxA_o3V432zRSNchRA

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #76 on: July 21, 2020, 04:56:02 PM »
"It's like watching paint cry."

What was I thinking? Rendering an animation takes forever. But around my place, the first heat wave of summer has ebbed. And I replaced most of my computer's fans. So the silicon slave is good for another bit of toiling. Yay!

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 2226
« Reply #77 on: July 21, 2020, 06:37:15 PM »
I was made aware of Yann Lby on YouTube. That channel has quite a few animated Mandelbrot height field zooms, among other fractal and non-fractal videos. Looks to be a custom height field renderer in isometric perspective. No orbit traps (yet?), but if you liked my few animations here, there's definitely something to see there.

https://www.youtube.com/channel/UC1abJGxA_o3V432zRSNchRA
I tried using  MMY3D but it crashes so often I gave up on it.

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #78 on: September 08, 2020, 12:43:29 PM »
"LiquiDifference Engine"

One would think that I'd know better. "Watching Paint Cry" took a few months to render and wasn't really worth it. I remember I wrote here that I wasn't going to animate mirrored height fields.

Turns out that with a few compromises, enabled by reduction of noise, render times with specular reflections are bearable. It should only take one or two weeks to finish this.

The aesthetics are a bit controversial to myself. Is this the shining chrome of glorious retro futurism? Or is this oily and slimy yuckiness? Maybe that ambivalence causes my own disgusted fascination. :)

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #79 on: September 17, 2020, 02:30:10 AM »
And there it is. "Slimy Chrome".

Online quaz0r

  • *
  • Fractal Friar
  • *
  • Posts: 128
« Reply #80 on: September 18, 2020, 12:45:22 AM »
very interesting dr hobold

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #81 on: September 18, 2020, 02:47:06 AM »
I am not a doctor.

(Owning a Tardis would be kinda nice, though.)

Offline C0ryMcG

  • *
  • Uploader
  • *
  • Posts: 239
« Reply #82 on: September 18, 2020, 07:25:56 AM »
Delightful!
It's quite satisfying to see the bubbles move around, it shows a bit of what's actually going on that the still images hide a little bit.

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #83 on: September 18, 2020, 10:52:34 AM »
The obvious next step is to zoom while moving the orbit trap. Traps are unwieldy, though, so there will not be much "art direction" when I produce such an animation.

Oh, and I stumbled on something that is perhaps another degree of freedom for the camera (in this setting where all of the animation is happening exclusively inside the height field data). Analogous to https://en.wikipedia.org/wiki/Thales%27s_theorem it should be possible to move the camera around the circle, while maintaining FOV (field of view) and keeping the exact same top and bottom edges in view.

Not sure if this camera path is actually a circle in general (for a FOV other than 90 degrees), and probably there are more precautions necessary to avoid looking beyond the height field somewhere along the other axis. Still, this might be good enough to pretend that the camera is flying freely.

Offline C0ryMcG

  • *
  • Uploader
  • *
  • Posts: 239
« Reply #84 on: September 19, 2020, 05:53:39 AM »
I never knew that theorem's name.  I remember that fact being very useful for something for me in the past, although i can't remember what.
I take it you're thinking of moving the pov-ray camera object around in a controlled way that always keeps the heightfield in view?
With this technique, I'm sure the top and bottom center pixels will be where you want them, but I wonder about the corners of the image... It's hard for me to picture this in 3 dimensions...

Well, here's one way to think about it. The camera's view is a cone, and a cone can be described as having a 90 degree tip. So if the tip of the cone is on the surface of a sphere, and the plane being viewed is intended to take place in a circle bisecting the sphere, you could predictably always have an oval as the intersection of the view cone and the plane. It's large axis would always be equal to the sphere's diameter (based on this theorem).

I think this could work if your image plane was INSIDE This 90 degree cone, although a big portion of the heightfield will be unseen often, especially at shallow angles. But with forshortening and whatnot, I think this is unavoidable anyway.

Another option would be to fake this kind of movement in the heightmap generation step, which will be quite math-y. I think I have some reference materials somewhere on perspective-warping calculations. Then you have to make sure your pov-ray camera can imitate the necessary tilt shift to get convincing FOV effects (Should be possible? Maybe as post-processing), and ideally, the heightmap would tilt, but I doubt that's an option in Pov-ray.

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #85 on: Yesterday at 03:56:47 AM »
Inspired by a recent bug, I experimented with remapping of elevation values by some non-monotonic function. Normally I had only slightly tweaked the height profile from valleys to peaks. But this time I am using a wavy profile. No mathematical justification whatsoever, but looks kinda nice.

Offline hobold

  • *
  • Fractal Flamingo
  • ****
  • Posts: 330
« Reply #86 on: Yesterday at 01:26:41 PM »
"Now this might get a little weird."


xx
Mandelbrot height map

Started by birational on Fractal Image Gallery

1 Replies
266 Views
Last post February 21, 2020, 12:46:25 AM
by Duncan C
question
importing height maps

Started by Kammcorder on Noob's Corner

1 Replies
97 Views
Last post August 06, 2020, 11:06:00 AM
by Sabine62
clip
Height mapped Mandelbrot

Started by pinballpsycho on Fractal Image Gallery

1 Replies
215 Views
Last post February 05, 2020, 09:30:55 PM
by Dinkydau
xx
depth of field problem

Started by Tsarvek on Mandelbulb3d

1 Replies
433 Views
Last post August 17, 2018, 06:36:21 PM
by muinami
xx
Depth of field usage

Started by Micha1982 on Mandelbulber

8 Replies
673 Views
Last post April 07, 2018, 07:44:31 AM
by Micha1982