The point is to capture as much surface detail as possible in the normal maps so that you can use less geometry. This applies to games as well as movies since increased scene complexity increases render times and can also leave a computer unusable so that nothing can be rendered at all.
In movies as well as games, LOD(level of detail) is very important to keep the computer running smoothly. When a model is further away, there is no need to use the most complex version of the model since none of that extra detail can be seen by the eye.
Normal and AO maps are ideal for these lower level of detail models since they are efficient but still look good.
Normal and ambient occlusion maps can display a lot of surface detail with very little performance cost. To display the same detail with actual geometry adds a substantial burden to a machine even when there is still only one model displayed.
The following page does a really great job of showing what normal maps can do. This page is specifically for Lightwave3D but the premise applies to all DCC software and games. http://amber.rc.arizona.edu/lw/normalmaps.html
The reason for this work around is to avoid the use of baking a high poly to a low poly using Selected To Active. It’s difficult to set up and is prone to artifacts. This method avoids all that.
Lets say that you sculpt a very detailed model using the multi-res modifier. If you bake out proper maps then you can apply that modifier at level 1 or remove it entirely and instead use the maps that you generated.
Now you should have no viewport lag, and also very fast render times while still having a very detailed model.
whoah thnx man, hey with the rebake, possible not .exr anymore?
(also another trick, 2d space normal map (scratches, little details) developed with plugins from PS and Gimp, and added to normal map, plug the normal again and rebake to have the 2d space normal map develop its own RGB (xyz) information, cheers!)
of course there’s a little problem with this, in dyntopo you have to retopo in order for us to even talk about games, in multires in most case you also have to retopo, the ghosts of selected to active haunts again!
@iceking: What’s your thoughts on having to retopo a multi-res model? I’ve never encountered this problem.
So far as the .exr thing goes… It is overkill to use this but just to be on the safe side. Really, floating point textures for normal maps are only necessary if the object is super shiny, otherwise the low-precision round-off errors are imperceptible. Diffuse texture maps also do a great job of hiding these errors.
On a similar note: displacement maps should also use the non-color data setting, same as normal maps.
The Decimate modifier is supposed to respect the UV coordinates since several versions of Blender.
By example to the left a manual retopo (with an applied Subsurf) from a Dyntopo sculpt that i then unwrapped and used a texture on it (4912 quads, so 9824 triangles)
To the right the same thing but on which i applied a Decimate modifer , resulting in 875 triangles
The UV is not too badly respected despite the heavy decimation :
Hey sanctuary.
You better post the good doctor, he looks great.
It is also a great example of what likeness means. Practically speaking.
About normal maps, try the new cycles implementation first.
It is experimental, it doesn’t support cage baking yet, among other things.
Still, it sounds much better than BI. In any case, better object space normal maps.
About this method (on topic)
I find it pointless.
You can do better by using displ exr maps and convert them to any kind of normal map you like.
I tried to explain it, many times.
It is all about the margin around the UV islands.
So far in blender,
As we still don’t have a real cage map baking support.
@marc dion: retopo for multires when resulting sculpt have totally stressed the base level resolution. for example: humie monster becomes horned, tailed, lots of changes overall monster, then retopo as needed. but yeah in some case there really is no need to retopo, btw, i’ve been using this tech for a year now, just thought everyone knew lol.
and also if your presenting stills, object normal map could also be okay even for characters. only in fully dynamic animating object is tangent absolute.
Play nice and stay on-topic, folks. We don’t need personal attacks or rebuttals to personal attacks… or even replies to this very post of mine.
Any further off-topic posts to this thread will be deleted. If you have a problem with this, you’re welcome to send me a PM or post in the Blender Artists Website Support forum.