That could distort an image based on a height map?

I'll explain with a few images what I mean.

Say that I have this image:



And I would apply this height map to it:



Then it would distort it to something like this:



So basically, the script would take the height map, and distort the image according to the information in the height map... the higher the value of a pixel in the height map, the more grow/bulge would be applied to the corresponding pixel in the image.

This would be really useful to me... Anyone know if something like this exists? If not... I might have a few ideas how to make it work, although I'm not sure how feasible they are...

The fish-eye filter in gmic > deformations does a similar thing but it make a gloppy center.

I like the one you did much better.

The standard displacement filter gives this:



using these displacement maps:



You will need to play with the displacement maps to see if you can get the shape you are looking for. All of the details for the maps are in the corner documentation block. (It's a file from when I was testing it last summer.)

There was a thread about this last summer. I think it was on this forum.

You can generate displacement maps from your heightfield using the emboss filter:


Just duplicate the hf and emboss it once as shown then again at and Azimuth of 90 degrees. These then are the X and Y inputs for your heightfield.

Note that you need to heavily blur the results of the emboss filter to get rid of all the 8 bit artifacting from the heightfiled gradient you provided before displacing.

here is my result:


Having the gridline centered on the dispalce was less than optimal

-Rob A>

Thanks guys. It works like a charm.

dd, just curious, which method did you use out of the 3?

The third one, RobA:s. It was the most optimal for my purposes since I needed to get the displacement to follow a very specific shape.

Well I tried to do it and I can't figure it out at all. It is as clear as mud. I kind of liked the effect.

nifty!

Ok, getting back to this... the method described here works pretty well for layers that have lots of transparency and only sparce objects here and there... but for flat, solid surface layers, it doesn't perform that well, since the pixels are moved away from one spot and that spot gets blurred by the mapping.

So I wonder if it would be possible somehow to render a voxel landscape based on the height map, colour that landscape with the corresponding pixel values of another layer, then shine a light on it and view it from top-down... does anyone know how to go about doing this?

Failing that... I've been wondering about another method. It would involve taking the height map, and dividing it into 255 layer masks, each based on a height value - for example, layer mask 0 would contain all the pixels with value 0 from the height map, and so on. Then take 255 copies of the layer, and mask each of them with the corresponding layer mask. Then scale them up incrementally, so that the first one wouldn't be scaled at all, and the last would be enlargened the most.

To make this work a bit smoother, the height map could be posterized to have less values, so you would need less layer masks and layer copies. I'm not sure if this would work in practice, though...

I wonder if this could be done with a text layer?
So you could get that effect of part of a man coming out of the text?
Like say his arm to hug something or someone.

Didn't work too well for me with lens distortion.

A bit better using the displace filter and then placing the displaced text in from of the original text.
Getting there.

I'd use a raytracer - POVray can do this really easily.

Otherwise you could try lighting effects. Specify a bump map and a light source...

Can you provide an example of what the result should be like? I have a hard time with your description.

-Rob A>

In my first post...

I'm confused as that does not match you description... you wanted "a voxel landscape based on the height map, colour that landscape with the corresponding pixel values of another layer, then shine a light on it and view it from top-down..."

Which in my mind is more like this:


-Rob A>

Well if you could align the landscape so that each voxel would be one pixel in size, so that effectively, with a flat height map, the resulting image would be the same as the original... Then with a non-flat height map, there would be bulges in it. Sorry, I can't explain it any better.

Maybe you want to check out Displace or Normalmap plugins. Both can be mis-used very coolly. I just mangled basic grid with tech texture using Displace.

If you want real-looking (but fake) voxels, Blender can help. Either get Freestyle-compilation of it or then use more modest edge rendering in official version.

No, I don't have very easy way to suggest, these are... cumbersome methods.

Five minute example with official Blender 2.59...

But still cumbersome, if you need to learn it.

I have blender installed, I just don't really have time to learn to use it...

As for the displace filter, like said earlier in the thread, it works pretty well for images that have lots of transparency, like grids and such... but not so well for solid surfaces.