I remember a discussion about this sometime in the past but I think it may have been on the old boards since I just searched through all 4 pages here and could not find it.

The discussion was about creating high poly meshes for normal maps and whether people created thier high poly mesh first or thier low poly mesh first. I seem to remember that most people felt it was best to create the high poly mesh first and that there were some good reasons listed for this. Can some of you who have dealt with this please post which way you work and what your reason is?

Thanks.

I think for best results build the low poly last. This lets you tweak the topology perfectly, and dosent give you any limitations while building the high res.
That being said, in production I dont have time to do that very often. So usually its build and unwrap the low, then take it to ZBrush to build the high res and generate the normal map. This works fine, but you cant make extreme changes mesh you are working on. If you do your mapping will get stretched. You cat't just cut loose in zbrush when using a pre-existing low res with mapping.
If you are very happy with you lowres, and you are just adding stuff like surface details, wrinkles, pockets ect. then the Low ---> High method works great.

Here

It depends on the individual model, in my opinion. Organic surfaces lend themselves to being turned from low-poly into high poly, but you lose some flexibility when going from low to high.

Hybrid approaches can work real well also...for example roughing out a figure in Zrush, then building a clean low-poly model from that, then re-importing the model into Zbrush to do the detail work, which ends up being turned into a normal/displacement map for the low-res model.

If the high-res and low-res models are being done without much communication (e.g. the high-res models are outsourced, while game-res models are done in house), then doing high-res first is probably much better.

How'd you do that?!?!

Thanks!