Thomas de Groot <tenDOTlnDOTretniATtoorgedDOTt> wrote:
> On 24-1-2012 2:12, Samuel Benge wrote:
> > I'll try to post some stuff generated by the utility soon. Hopefully I won't get
> > sidetracked like I usually do :)
>
> Very good! Nothing wrong with sidetracking though ;-)

OK, here's one.

For this image, I took the output from strands.exe and expanded the height data
outward from its own density several times using a function similar to the
vector analysis stuff in math.inc, lowering the resulting value and recombining
it with the original image at each step (I finally found out how to perform this
process during an animation without producing a folder full of numbered images:
..bat files are created at each frame and called after each, in order to delete
and rename previous images). This "filled out" the original lines and eliminated
nasty artifacts without removing all of the smaller details. The height data
used throughout had a resolution of only 512x512 pixels, except for the texture
baking step which was twice that. The texture employed a 2D proximity pattern to
deepen the areas between the "roots," and fake subsurface scattering
(diffuse-occluding absorbing media) to help give it that living look. The HF was
copied nine times, so what you see is actually tiled, and repeats.

The water's meniscus was created by taking the height data at the water line,
making it white, and blurring it. The blurred image was simply mapped to a
watery plane and used in a normal block with a few tweaks. The interior of the
water uses absorbing media.

For both the proximity pattern and blurred height data, I used points created
with Fermat's spiral, which I've discovered produces better results for 2D
pattern blurring than other arrangements, and is more flexible.

The result is... kinda gross. Possibly more animal than plant :/

Post a reply to this message

Attachments:
Download 'tubers.jpg' (217 KB)

Preview of image 'tubers.jpg'