I took some time over the holiday break to organize some old files and while doing so found some stuff I wanted to post. These are two pre-rendered environments I did several months back.
This was a scene made for the box cover of the Nintendo DS release and PC rerelease of the game Build-A-Lot:
[ame=
http://www.amazon.com/Build-Lot-Nintendo-DS/dp/B001AHMM8A][/ame]
The wide version was meant for a fold-out PC box, but I don't think that ever got released.
And the below background was created for LUXOR: Quest for the Afterlife.
Replies
I like the foot prints in the snow, nice touch man
The cliff looks great and i like the color nuance that flowers bring to it.
Just the flower lower down the cliff (the one in the shadow of a bridge) seems to be not in place though.
I agree. I added it to match the concept the client gave me.
Forgot to post this as well. It's a loading screen for a canceled casual game I worked on awhile back.
Love the style of these mate!
They were rendered out separately and composited in Photoshop. I did that with some of the smaller items so that they could be moved around in the PSD by the client.
Sure, below is a brief overview of the project. A lot of the process is probably common knowledge to most artists here, but guys new to environment art may find it helpful. I usually go about creating prerendered environments by breaking it down into several phases:
Resource Gathering: Look at the finished concept and start gathering real life reference, possible texture images, and any old assets I can reuse (textures, models, Max shaders, lighting set-ups, scene effects, etc).
Rough Out Scene: For Luxor the 3d image had to fit the concept close enough so that the ball paths matched up exactly. I set the concept as my viewport background and started to block in geometry and find the right camera angle.
Texture Creation: I created all of the textures in the same method as current game tech (Diffuse, Normal, Spec).
Prop Creation: This scene didn't have many unique props, so all I needed to model was the bridge, some icicles, and then modify a flower I created for a previous game. I created these in a separate scene. The bridge I modeled out straight and used soft selection and FFD modifiers to get the right curves in it later on.
Final Surface Geometry: This was the hardest part and one in which I could have gone a different direction. I started adding more detail to the roughed out rock geometry using sub-d modeling. I wanted to get the basic shapes down and then bump out additional detail using a combination of Normal and Displacement maps. This worked out okay, but the Displacement map caused weird artifacts if pushed to far, so some of the rock faces turned out a bit flat.
Next time I'm going to jump into ZBrush or Mudbox for the detailing phase. I was worried I wouldn't be able to match the camera angle in those modeling packages and the paths would get distorted from the concept. I think that would still happen slightly, but if I projected it to a lower poly mesh I could easily fit the optimized mesh back into the right positions.
Mapping: I unwrapped the rock pieces, threw the tiling textures onto the geo, and checked to make sure all the colors/values worked well together. After mapping all the tileable textures to the right size for the camera, I then mapped the rock and bridges to a second UV channel to use for Blend masks (using a B&W TGA to blend between to materials). I painted the blend masks in Photoshop, creating dirt paths and defining where the snow would fall.
Examples of masks, textures, and masks applied to separate UV channels.
Prop Placement: I merged my bridges and flowers into the scene and moved them over any existing placeholders I have roughed out.
Flush Out Full Scene: I took about 5 minutes to block in geometry around the areas not visible to the camera, but that would have rocks there in real life. The Tessellate modifier works okay for making blockey shaped look more "rocky." This helps during the lighting phase to get more accurate shadows.
Lighting: I move really fast in this phase, just trying out different approaches to see what will get me a good result while using the least amount of lights possible. I keep the renders small, the sampling/light rays low, turn subdivisions off, hide most small props, and do anything else I can to keep renders fast until I have something that looks good.
In this phase if it seems like the set-up is going to need a lot of hackery to look correct, I save it off into another file and try something new. With practice you can usually hone in on the best approach pretty quickly. For this scene I used a Skylight for global illumination, a Direct light for the sun, and a few Omnis to light up key areas.
Rendering: This is when I'll turn up sampling/subdivisions/ray/etc, increase my render size, add post effects and start working towards a finished render. For this project one of the requests was that majors pieces be but on different PSD layers, so I had to render out the bridge, flowers, and rock outcroppings separately so they could be adjusted if needed after final delivery.
Post: This is just putting all the different renders together in one PSD, color balancing if needed, or add in more post effects like bloom/fog/DOF/sunbeams/etc. The fun thing about prerendered scenes is that what would take 45 minutes to do in the scene can be done in Photoshop in like 5 minutes (the footsteps in snow where done this way).
I haven't do much of hi-res enviro renders so that's cool that u posted some details about that
What are you working on now?
Cheers
Earlier in 2008 I got a job at Nerve Software; we just finished up some multiplayer work on Quantum of Solace. Before that I worked at MumboJumbo, which is the studio I did these for.