Recently I've been wondering what steps people take when they create an asset (static mesh) for UE3. Anyone want to share from beginning to end the basic steps they take when creating a static mesh of UE3? Here are a couple steps to give you an idea of what I'm looking for. I'm putting these in no specific order.
- Gather Reference
- Create High Poly
- Create Low poly
- Gather textures
- Create Normal map
etc.
I know that you may think that this is different for everyone but I'm just curious on how different people work.
Thanks.
Replies
http://boards.polycount.net/showthread.php?p=678052#post678052
The workflow and process of others has always really interested me. A lot can be learned by simply watching someone else and how they do things. It took a while for me to really focus on a workflow for all assets and try to stick to it and not jump around (within reason, as no two assets / enviros, etc are the same). But when I did, it was much easier to hone in on areas where vast improvements in speed and quality could be made and are still being improved.
Not sure how in depth you wanted to get, but here is an overview for a basic asset that would end up in a game engine (UE3 for example):
1 - Gather reference photos, make quick sketches, acquire concept, etc.
2 - Break down how asset will be used, what makes it interesting (where detail can be placed), and identify any technical / artistic hurdles.
3 - If modular (not a unique set piece), block out quick cubes / volumes that are to scale to ensure asset will snap on grid and give you an idea of the space you have to work with, and plan how mesh will be broken up to increase reusability and reduce the number of materials to mesh ratio to as close to 1:1 as possible. Seriously, don't overlook a little planning. It will really go a long way and save your ass!
4 - Model high poly asset / pieces. Keeping in mind how asset pieces will be exploded for baking / paired up w/ low res counterparts. Mudbox elements where needed.
5 - Use high poly assets as base for the low poly. Sometimes removing constraining edges / misc geo to make low poly, majority of the time creating a new low poly inside the volume of the high.
6 - UV map unique low poly meshes. Parts that will be instanced / duplicated / used elsewhere on the model are added last, almost right before export. With elements of the model that need unique UV space, put all UV chunks / islands in the 0,1 space for baking later. Scale all UV chunks relative to each other to ensure even texel density.
7 - Setup projections for baking if used. Bake Normal maps, Ambient Occlusion maps, and masks (flat colors for different materials for easy selection in Photoshop) for all exploded elements. Bake AO map w/ all low poly elements combined as well for additional detail.
8 - Combine baked texture maps into one texture map per type (i.e. take all normal maps baked out from different element and combine them into one normal map for the model)
9 - Add any additional normal detail by modeling detail bits on a plane over the assets UV layout or cleanup any normal map errors / smoothing issues.
10 - Use AO map and masks as a basis for color maps.
11 - assemble complete low res asset (attach all elements, instance any duplicated elements etc). Export to engine and test smoothing, lighting, etc w/ normal map and base AO map on asset.
12 - Introduce photosource for color maps. Add well placed detail before adding grunge and othe misc overlays. At the same time begin thinking about and working on the spec map.
13 - Get texture maps in engine and begin tweaking a shader. Add any additional maps if needed (masks, reflection, emissive, height, etc). Iterate on textures and materials till its fabulous!
14 - create collision geometry and lightmap UV's if not vertex lit.
15 - profit.
And hopefully you're done. Man, writing that out is tedious. We do a lot shit to make some art players are just going to run on by!
Cheers,
Tyler
What i find easier (for myself) is to make a medium polygon model wich has a bunch of edge loops and extra geometry that a low-poly doesent need but helps me build a good high-poly model around i use it to help define the shapes that will be on the high-poly and then be baked down into the lowpolygon model.
i find this easier than making a highpoly model and then making a low-polygon model.
but it depends on what you are more comfortable with.
I really think making a highpoly first is a bad idea, except for in some situations or specific parts of the model. Really depends on what you are making!
Oh boy, that got a bit longer than i expected.