I’m really glad to see this kind of functionality going into WM! (see the developement blog) As a game developer, I’ve been thinking a lot about how to handle texture distribution on big portions of terrian. I’m wondering what the the resolution will be like at the texture transitions- is it dependent on the coverage map’s resolution? I guess it would have to be. But it would be super cool if it wasnt, because i can see all those coverage maps being quite expensive.
This topic also brings to mind another thing I’ve been thinking about- object placement maps. I was always excited about the probability device, but it would be sweet if it had more rules built into it like: clumps, density, noise, slope, altitude, etc. Or even presets for rocks, trees, bushes.
Regardless, i can’t wait for WM Pro! great job Stephen!
I think the resolution of the coverage maps (a normal heightfield used as mask) needs to be the same at the transitions and on the rest, because it is an image. If you need more resolution at a certain part of an image, you really need to increase the overall resolution. (no such thing as content-based mipmaping (Jpeg compression) is possible, I think) (at least there is no way to define it without specifying the full detail version first)
This topic also brings to mind another thing I’ve been thinking about- object placement maps. I was always excited about the probability device, but it would be sweet if it had more rules built into it like: clumps, density, noise, slope, altitude, etc. Or even presets for rocks, trees, bushes.
You mean, a device, that would chew up a "probability map" and produce coordinates for placing objects? (Note that WM doesn't have "objects", so any object placing should be carried out using a different program..) But I agree the probability device could have more juice in there.
I think Stephen didn't implement it like that because of some statistical issues (and additional noise distribuition functions need to be implemented to control it that way).. I believe the original intention of that device was to behave sort of like that, for placing the "spikes" on more desired places. But you can control it somewhat using the bias slider, and carefully operating on the input heightfield.. For a "cellular" type of behaviour, you may need to use it in combination with a Voronoi device..
This would meen that a preset for trees, for instance, would need to be a macro, and not a preset, for the time being..
Hmm…If there was only some way you could get around having huge mask maps. Like maybe you could “unwrap” a mountain for example somehow so that the like colors of the mask were packed together and only the areas that need more res ( the transitions) were represented (mostly) in the image. I think you could do this manually with your 3d app and photoshop but it would be cool if could be automated.
I’m really glad to see this kind of functionality going into WM! (see the developement blog) As a game developer, I’ve been thinking a lot about how to handle texture distribution on big portions of terrian. I’m wondering what the the resolution will be like at the texture transitions- is it dependent on the coverage map’s resolution? I guess it would have to be. But it would be super cool if it wasnt, because i can see all those coverage maps being quite expensive.
This topic also brings to mind another thing I’ve been thinking about- object placement maps. I was always excited about the probability device, but it would be sweet if it had more rules built into it like: clumps, density, noise, slope, altitude, etc. Or even presets for rocks, trees, bushes.
Regardless, i can’t wait for WM Pro! great job Stephen!
By the transitions, you mean the transition from a given type of coverage to another? Yes, the resolution of those would be bound to the map’s resolution. And yeah, uncompressed they end up pretty large. One handy thing, though is that by their very nature coverage maps and other types of metadata compress losslessly REALLY well (since they contain a limited number of colors in large single-color areas).
Regarding object placement: yes! Actually, one of the cool things about the Coverage Maker device is that you can use it to create coverage maps for textures definitely, but by piping the output from it through the probability device you can create object placement maps that obey any number of constraints (slope, height, erosional areas, etc).