View Full Version : weird idea for forest texture

02-09-2010, 02:58 PM
weird idea for a forest texture. I was scanning a convex old photo for a customer, and I removed the scanner lid and threw my overcoat on top to cut out the light without bending the photo and noticed something. Though the fake lamb's wool liner was cotton colored, it looked a lot like a forest texture. So after I finished scanning the photo, I scanned just the liner, applied a greenish tint and exported as a JPG. I also added a smaple of it in use. What do you think? It is useful?

Edit: I see I need to tweak the main texture itself, some white is showing through, I will fix and upload later.


02-09-2010, 04:09 PM
That's a pretty funky idea! I know i tend to use moss for my forests, but that's a cool direction to take it.

Now you got me wanting to look through my closet for interesting textures.. whee!

02-09-2010, 04:18 PM
Nice thought on a start, but it looks like you may need to get a cleaning scan. There are what appear to be hairs in the image. Here is my own real quick play with your original image.

02-09-2010, 04:32 PM
Yeah, that's my dog's hair! I need to wash the coat and do it again. It was a spur of the moment idea.


02-09-2010, 04:41 PM
Could someone potentially create a script that creates random "blobs", like bubbles? Think of a thin sheet of soap bubbles... I think that could work. Hard to photograph, I can do the general math for it on paper using set theory but the specifics of converting it into code is far beyond me.

02-09-2010, 05:39 PM
yeah its a pretty cool idea and does look pretty good! Indeed some big furry animal hanging around in those woods losing his hair ;) My 1st association was cracked mud so maybe if you give it a sandcolor and scale it down it could work for that too. Jugg, the math behind that would be very interesting, been a long time though i coded graphics so cant promise i'd step in ... cheers

02-09-2010, 06:23 PM
I would think that a distance function applied to random points with a Poisson distribution for your desired radius followed by a few tweaks like thresholding, exponent, uniform noise, and erosion would generate a result fairly close to this. The Poisson distributed point set (a jittered grid is a simple to compute approximation) gives you the centers of cells that are roughly the same size, the distance function gives you a smooth distance between points with nice cellular boundaries between blobs, the threshold ensures the cellular boundaries become clear, the exponent to make the blobs a little less linear, noise and erosion to add the fine-scale stochastic element. I'll see what I can do this evening about synthesizing this sort of thing.

Any spectral sythesis fractal generator that has a cellular basis function should be able to give similar results, especially if the cellular result can be applied only to the first octave or so (think crocodile skin or cracked mud shaders).

02-10-2010, 03:41 PM
I think it looks terrific, GP. Get out your lint roller and do it again!

And I'd also love to see the results of an algorithm if someone has the inclination to write the code.

02-15-2010, 08:10 PM
I tend to think that this may also make a nice texture for mountain peaks, used selectively. I think I would combine this with a second, smaller texture to create an even more realistic resource, but you're right, this would work great!