Hey guys, I need to make a list of tools for procedural content generation, including models, textures, animation; all of the ones i've heard about are already on Wikipedia:
http://en.wikipedia.org/wiki/Procedural ... generation
So what I'm wondering about is which ones are better than the others; anyone have experience with some of these more obscure ones? A lot of people have used Terragen, World Builder, etc.
I suppose the natural follow-up question here is - what, as modders/content developers, would you want to see in a procedual modeler for games? Lets say (f.ex) that you had a tank, and you could push a button and instantly generate a hundred different mutations o nthat basic design, with different treads, turrets, hatches, detail bits, etc. or something similar for airplanes.
would that be useful? or desirable?
Thoughts gentlemen/women?
[help] procedural content generation tools
Moderator: Moderators
Re: [help] procedural content generation tools
I'd suggest looking at the various tree-generation systems, if they weren't listed in Wikipedia. It's one of the more important ways that this idea has actually reached production stages. Trees are easy targets for automation, because, like terrain, they generally obey certain algorithmic rules.
Something that tries to do the same thing with stuff where artists are paid to make it as cool as possible, like tanks and planes... no. That's what artists are for- to make character and mechanical designs that are interesting and make for a memorable experience.
However... make something that makes model buildings procedurally at different polycount targets, using pre-mapped parts, and you'll probably have a serious resume item, Dragon. I know that Hollywood and companies like EA have these things.
2. When talking about true procedural stuff, i.e., put these bitz together, voila, a thingie... it's uvmapping, skinning, and animation where you'll find the problems really are. Not modeling.
Uvmapping is a non-trivial task that can't be easily automated (and is already automated a lot more efficiently than it used to be). Skinning can't be automated except in isolated scenarios (and even then, usually demands touch-up). Animation is engine-specific.
1. The times that you really want 500 variants of a thing... it's always the "boring" stuff. For example... a procedural-content generator that can build low-poly building exteriors that are pre-uvmapped and use prebuilt skins... that's really sexy, if you're a producer. If you can go to your art lead and say "hey, make 10 wall greebles, 10 roof sections, 10 windows", and then use some rules to suddenly create buildings ("this is a roof corner, this is roof top spine N/S", etc.) then you're providing a substantial tool for all of the areas where doing it by hand is slooooow and you'd like to save on costs (both money and time).I suppose the natural follow-up question here is - what, as modders/content developers, would you want to see in a procedual modeler for games?
Something that tries to do the same thing with stuff where artists are paid to make it as cool as possible, like tanks and planes... no. That's what artists are for- to make character and mechanical designs that are interesting and make for a memorable experience.
However... make something that makes model buildings procedurally at different polycount targets, using pre-mapped parts, and you'll probably have a serious resume item, Dragon. I know that Hollywood and companies like EA have these things.
2. When talking about true procedural stuff, i.e., put these bitz together, voila, a thingie... it's uvmapping, skinning, and animation where you'll find the problems really are. Not modeling.
Uvmapping is a non-trivial task that can't be easily automated (and is already automated a lot more efficiently than it used to be). Skinning can't be automated except in isolated scenarios (and even then, usually demands touch-up). Animation is engine-specific.