Itoo Software Forum

Forest Pack => Forest Pro (*) => Topic started by: simmsimaging on February 07, 2012, 07:33:12 AM

Title: not so random randomness?
Post by: simmsimaging on February 07, 2012, 07:33:12 AM
I'm wondering about another possible function. I get that randomness is clumpy and that is how it is, but when dealing with something like a crowd of people then showing repetitions of the same object too close together is really obvious.  The current project (aside from grass) needs a crowd of fans in a stadium and I am seeing a lot of areas where there are sometimes 2-3, sometime 4-5 copies of the same person right beside or very near together.

With trees etc it's not so bad, but in this case it is really apparent.

So: is it possible to add a variable that limits the "repeat distance" for objects in a FPP set? It would be I guess more "stochastic" than truly random, but it would avoid really apparent repetition if I could, for example, say "no repeats within x units"  - that kind of thing?

Possible at all?

b
Title: Re: not so random randomness?
Post by: iToo on February 07, 2012, 10:04:23 AM

We have received this suggestion for other users as well, but unfortunately there is not an easy way to do it. The algorithm would be too complex, too slow or both. Anyways, we will think on it.

If you are using 3D people, one alternative to add more variation, would be to change their cloths randomly using Forest Color. You would tint it from a discrete range of colors, or use several textures. Something as this:

(http://www.itoosoft.com/img/forest/ref/tintmap2.jpg)

Title: Re: not so random randomness?
Post by: simmsimaging on February 07, 2012, 06:25:43 PM
That method won't work for me unless i want to custom make 70 texture map variations for each colour change - the tex maps includes skin, shoes, hair etc so randomizing the colour would affect the whole person :)

If the maps were setup differently your method would improve things though.

Regarding the algorithm: hopefully you guys can come up with something. As you well know, I am not a programmer, but I was wondering if it could be an adaptation of the collision system? Like an added layer of collision detection that would be 'self collision' for individual geo: then you could raise the detection radius to a large percentage (like 1000%) to make sure no repetitions can occur within roughly 10 body widths etc. 

I understand your point that it would be slower, but so is collision detection and there are cases where it would definitely be worth it!

Is that a possible way to go?

/b
Title: Re: not so random randomness?
Post by: iToo on February 08, 2012, 09:29:03 AM

May be it would be possible, i'm not sure. The collision detection algorithms are well studied and documented, but this is something completely new. We will think on it.

Title: Re: not so random randomness?
Post by: simmsimaging on February 08, 2012, 10:03:12 AM
Great - thanks

/b
Title: Re: not so random randomness?
Post by: GF1991 on February 10, 2012, 03:53:16 PM
Hi Brett,
how about using this?

http://www.vizpark.com/shop/color-tools/ (http://www.vizpark.com/shop/color-tools/)
Title: Re: not so random randomness?
Post by: simmsimaging on February 10, 2012, 04:13:40 PM
Thanks for the link. It's interesting, but doesn't really solve the core problem :)
Title: Re: not so random randomness?
Post by: Atomica on February 20, 2012, 08:29:10 AM
We agree with you on this and really need the ability to have say 100 people planar objects in a forest scene and it only places 100 objects (no repeats)

Any updates on this ability guys ?
Title: Re: not so random randomness?
Post by: iToo on February 20, 2012, 09:26:12 AM

No, sorry. It is in the wishlist, but the list of requested features for Forest is fairly long. It may take some time.