High resolution and low resolution models
polycounter lvl 5
Noob question: I am wanting to create a high resolution model of an object, and applying it's normal map to a low resolution object but i am confused on how to go about creating the high res and low res objects. I understand how to transfer the normal map but how do i go about making the models. Do i create amlow poly and then add more resolution. Do i make a high poly then take away resolution? How similar does the geometry have to be for a normal map to be applied.
Replies
Typically it's best to block out the model and roughtly capture it's shape and porportions with basic primitives and focus on large details, look at this article, http://www.nextgenhardsurface.com/showthread.php?401-NGHS-Article-The-Importance-of-a-good-blockout
After the blockout you then model the highpoly model. Then typically you model the low poly from scratch to match the high poly as close as possible. Some parts of the model might be easier to create by reducing the polycount of the high poly object, but that's not ideal for most parts of a model.
I make a base object in 3Ds Max, for example. Then, depending on what kind of a model it is, I may want to work on it in Zbrush, where I sculpt upon it. Then I return the model to Max for baking. Now, if I have changed my model's shape by a very small amount, I may reuse the base Max model as the low poly. If I've changed it a bit more then maybe some Max vertex moving will adapt the lowpoly to the highpoly one. But I achieve the most approximate result if I use some Zbrush methods to make a low poly mesh instead. Any one of these will do with differing results.
1. Go down a subdivision level
2. Use Dynamesh with a resolution of choice.
3. Use Decimation Master to remove redundant tris
4. Use Zremesher if you have it, it's something between both 2. and 3. but with more quads.
This is assuming the Max and Zbrush workflow, cant tell you more til you specify your tools; but what I meant to say is that there are different ways to go about this. But for complex models, automated retopology will probably be the most painless sort.