I am starting a project to be displayed on a web browser ( a vitual tour of a place ) and it will be my first attempt using Unity.
Therefore, I want to avoid as much problems as possible during my production pipeline and the first question I am wondering is if Unity handles multi-sub objects / multi materials of 3dsmax and if not is there an alternate way of having several textures assigned to a same object.
Also what would be the best/suggested max size of textures for a web application, 512x512 or with the powerful compressor of Unity there are no problem with 1024x1024?
I don’t know about the Max side of things and wouldn’t want to say either way. But as for texture size for web use, your best bet is to keep as low as possible as there are far more people online with dialup than it seems (I’m not, but I’ve never forgotten what it’s like lol).
You could always share textures between objects, go through your textures and see where there’s lots of wasted UV space and double up UV maps to use the same textures/different parts.
Yeah I’ll try to keep it as low as possible indeed.
Also another question came back to my mind, what happens if there are normal maps used in the scene but the graphic card of someone viewing it doesn’t support it? Would the scene appear to him the same minus no normal map displayed or would it create some weird glitch like an invisible object (the one with normal map on) or something?
the shader will fallback to the specified fallback which commonly means a lower one or vertex lit. But you will have some hard time finding anyone able to play the game at all but impossible to use NMs as even intel GMA 900 / 950 can do that
All(most?) of the built in shaders take older setups into account, those will fall back to something that does work, may not look very nice, but it’ll work.
Custom ones however may or may not have fallbacks, depending on if the person who made them thought to add one or not. Best to go through all the shaders you use, check them for fallbacks, adding some if needed, and also run the graphics emulation settings too, and test on as many variations as possible, and in as many browsers as possible.
Aim for a baseline usergroup and stick with that. For example, if you’re aiming for gamers, you should be fine focusing on the latest effects. If you’re aiming for anyone, then assume people with very old hardware are going to be viewing it and work accordingly.
If your using multisubobject materials and exporting them, internally your basically creating two seperate materials. Whatever the engine does, when it comes to render time the object will be:
split to seperate the two materials
vertices along the seam will be doubled.
two seperate materials will be drawn.
So your video card will not be running as efficiently as it could. Each material means an extra mesh or draw call. Any smoothing groups or materials on the same mesh will be split to create a seam with double the verts on those vertices both materials share.
If your using per pixel lighting and shadows you each material/mesh combination will get their drawcalls multiplied by the number of lights in your scene, and again if they cast shadows.
So your best bet is to try and combine all your materials into a single material UV set, if possible multiple objects in a single material UV set like a texture atlas. Depending on which 3D app you use this could be simpler than it sounds.
For example 3ds max 2010 or later can UV map and pack multiple objects to a single UV Channel/texture page texture without having to work with each individual mesh separately.
Unity’s more advanced shaders like bump. specular, alpha etc do have fallbacks for older hardware so ancient video cards will fallback to diffuse + alpha and ignore Bump/specular etc.
The only problem with older cards like some intel 9xx is that they support PS2.0 but do vertex shaders on the CPU which slows things down a lot and normally hardware profilers only check the higher level, see PS2 support and do they best they can with the most advanced rendering despite not being suitable.
I’m not sure how well Unity’s GPU profiler works and whether it is more sophisticated than something like Ogre3D’s.
Lets say i have a scene that contains many many little detail objects. Trashcans, tables, chairs, roadsigns and etc…
The smart way would be unwrap all these objects on one texture atlas but this is big job. I texture my little objects like this. I add UVW Mapping modifier and then choose box, cylinder and etc. Every element gets their own UVW map and their own texture. Then i collapse the stack and import them into Unity. In Unity everything is OK. The textures are on object as it should. Of course i have no idea what goes on behind the scenes. Do you have any suggestion or ideas?
Is this the smart way of working or should i really start to manually unwrap every single object?
Each prop should be unwrapped. Some things (like simple crates or barrels) can be left as-is since simple uv mapping is enough for them (if you’re not using external sculpting tools).
Of course, unwrapping a crate should take no more than 2 minutes so you’re not saving a lot of time.
Think of your scene as a reusable lego city instead of a unique piece of mesh and things should “lock” into place.
I’ve seen Kevin’s stuff before, that guy has stupid amounts of talents. But had never seen the work from the second link, omg that stuff looks so good! I love the style!