Hi, I am looking to create a forest environment for my portfolio and i need your help. Since reference photography is so complex, i would like to see how game artists have tackled forests previously. I'm going to go for something with a bit of a magical/fantasy twist, and have already started on some patches of grass and bluebells. If anyone can suggest some forests in games for me to check out, that would be ace. Extra 'thank you!' points for screenies/links. Thanks, Dan.

I suspect you want to look at it from a graphical/artists point of view...

http://www.speedtree.com/
Speed Tree might be interesting for you.

If you want to know about forest in videogames you must find out speedTre (http://www.speedtree.com/), that's a middleware that lets you render large forest.

It's used in a lot of videogames as you can see in the gallery. (http://www.speedtree.com/gallery/)

Enjoy it.

Sorry guys.

I didn't make myself clear. I'm not so interested in technology (although speed tree is lovely), but more in the creation of the environment overall.

For example, interesting use of colour, ideas for deatials or perhaps how the artist spends polygons.

That's what SpeedTree and similar programs1 are about. It procedurally creates trees with 1000-3000 polygons from some tunable parameters and also bakes textures for texture-mapped billboards for far away trees.
Billboards are crucial for real time rendering due to the polygon count (insignificant if you're doing a still image only).


1Saying "similar" is a bit unfair, because there's really nothing even close to SpeedTree. But of course, there's not much that's as expensive, too.

Quote: Original post by samoth
That's what SpeedTree and similar programs1 are about. It procedurally creates trees with 1000-3000 polygons from some tunable parameters and also bakes textures for texture-mapped billboards for far away trees.
Billboards are crucial for real time rendering due to the polygon count (insignificant if you're doing a still image only).

While poly counts do have something to do with it, its more so because the over draw would be too much. Calculating the transparency for each leaf overlapping another causes a large hit in fillrate. Billboarding causes this to not happen nearly as much.

Quote: Original post by Jarrod1937
While poly counts do have something to do with it, its more so because the over draw would be too much. Calculating the transparency for each leaf overlapping another causes a large hit in fillrate. Billboarding causes this to not happen nearly as much.

Hmm... but since you need to alpha-blend the billboards so they look somewhat decent, I'd say that they're even more expensive in terms of overdraw, no?

Quote: Original post by samoth

Quote: Original post by Jarrod1937
While poly counts do have something to do with it, its more so because the over draw would be too much. Calculating the transparency for each leaf overlapping another causes a large hit in fillrate. Billboarding causes this to not happen nearly as much.

Hmm... but since you need to alpha-blend the billboards so they look somewhat decent, I'd say that they're even more expensive in terms of overdraw, no?

Well, most artist i've seen would be alpha blending eitherway. With or without the use of billboards. Billboarding just allows less individual leaf groups to be behind one another, resulting in less overdraw, and less poly usage.
Although i've been an advocate that simple alpha testing can be enough to get by without the more expensive alpha blending. But thats a matter of debate.