It's a combination of the two really.
[Zbrush] sculpting
First, I model the folds in Zbrush and make sure you can tell what the big ones are on the silhouette. Put a flat black shader on it and rotate the model to make sure the major forms are recognizable.
Next I make the secondary folds, and I check their readability by changing my shaders in zbrush. Just try a bunch of different shaders, rotate the model and see if any detail becomes hard to see.
[Topogun / 3D coat]
For retopology, I cut the biggest folds in to the mesh, so they really stand out. Try to keep it In areas that don't bend much. Leave the rest of the folds to the normal map.
[Xnormal / Photoshop] texture baking, texture painting.
I add some shading to them on the texture map, using Ambient Occlusion and maybe a cavity map, paint by hand if you need to.
On the color map, I make the edges of different parts a little lighter or darker. Just so you can see a separation between the forms.
[3Danimation software] 3ds.max, maya, etc.
In the material shader, I'm mixing the Mental Ray Ambient Occlusion shader as a multiply layer, or what ever looks good in the diffuse color channel. Using a composite map.
Then I have a normal map on it, strength set to around 200 in 3ds.max. If you set it higher than 100, the details show up more, but don't push it too far because it'll look too obvious.
The light rig helps also