Low to High / High to Low ???

Incomitatum

The current project I am doing will require some SubD, hard edge, modeling. I was going to just -go- for it, model the crap out of it, add some details in Mudbox, and then use PolyCrunch to distill a LowPoly out of it.

But, some say this is wrong (though they have fundamentally said why).

So then were do -you- all get your Low Polys from?

Do you build them first, preserve them, and then add details to them to get your High?

Or do you (and please say "no" as this would drive me nuts) Go from your High and build a Low -over- it, hoping the topography matches up well; if so, how do you keep from punching yourself in the balls while doing it this way?

Do these different methodologies matter at all? So long as it -looks- good and has an acceptable count, in the end.

sampson

For me highpoly to lowpoly is the best method. Just gives you alot more control and you arent always being restricted by your lowpoly..

Incomitatum

Thank you Sampson; but how do -you- chose to get there then? Do you rebuild the low, on top (if so what method and tools do you use to keep from hanging yourself (it just sound like -so- much more work)? Do you use something like PolyCrunch and derive -from- the High -to- the low?

pior

Retopology workflow is where its at. the task can sometimes be reduced down to less than an hour depending on the asset.

Topogun!!!

ImSlightlyBored

Either way works man, I wouldn't worry about things and worry more about the end result is all... if it looks good, it looks good.

I do both, normally high to low with low being a deconstructed version of the sub-d cage (ie remove loops, optimised by hand etc) but I've done low to high before now, and even UV to modelling so it is what it is

Incomitatum

I am the first to admit how daft I am. By Retopology do you mean manually deleting loops and verticies and generally dumbing down the High poly by hand?

CompanionCube

for some bits i reduce the polycount by hand, but mostly i use the freeform tools in max 2010, or similar tools in which ever program you use, theres also programs like topogun

sampson

yeah if im doing just like a hard surface asset, ill just get rid of my turbosmooth modifier, remove the supporting edgeloops, tweak the mesh and your done no need to rebuild an entirely new lowpoly (but it depends on what your making)

Tom Ellis

My workflow for hard surface stuff is usually:

1.Model High Poly

2. Model Low Poly - Clone high poly mesh and remove excess loops, add bevels/adjust geometry in preparation for baking.

3. Unwrap low poly

4. Bake Maps


For Characters it usually goes like this:

1. Base mesh in Max

2. Sculpt Base Mesh in ZBrush

3. Export Sculpt and Retopologize in Topogun (create a new mesh by creating faces over the high poly in Topogun / Retopo Tools Max plugin)

4. Import retopologized mesh into Max.

5. Unwrap / Bake.


I notice your website's Twitter feed says you are watching the Old Damaged Pillar DVD... while it's a great DVD, I'd ignore the bit about Polycrunching the mesh to create the low poly. Generally that's not a recommended workflow and seen as a bit lazy. It's acceptable in that case because it's a pillar and doesn't require precise texture placement when working with the UVs, which would otherwise be pretty tricky with a Polycrunched mesh.

Restrict your use of Polycrunch to smaller assets where it seems excessive to build a complex low poly... rocks, sheets of cloth etc etc.

[HP]

sampson wrote: »

yeah if im doing just like a hard surface asset, ill just get rid of my turbosmooth modifier, remove the supporting edgeloops, tweak the mesh and your done no need to rebuild an entirely new lowpoly (but it depends on what your making)

Exactly my workflow. (Keep in mind though, I use this for environment pieces, not sure about chars)

Yes, the pillar DVD is great, if anything, you learn great little tricks with it.
I agre that Polycrunch shouldnt be used on most assets, but on more organic assets like that destroyed pillar, is ok I'd say.

Psyk0

Have a look at this tutorial, it's essentially what creationtwentytwo described for character workflow:
http://area.autodesk.com/tutorials/street_cop_workflow_by_mashru_mishu

You'll probably need to register to get access tho.

stimpack

I just do env. but i assume same principles apply..... I do a med rez. From that i extract off geo to make my High then use med again to make my low from. If you dont get too crazy it should all line up.

Clockwork

Incomitatum wrote: »

I am the first to admit how daft I am. By Retopology do you mean manually deleting loops and verticies and generally dumbing down the High poly by hand?

You put your hi poly in the background, then lay out new topology over it. Poly by poly. With good retopology tools it's a breeze.

I generally only do this for character modeling.

mLichy

I do HP model first when it comes to hard surface stuff, then I just use that mesh most of the time for the Low, unless it seems like it's going to take too much time. I try and setup as many loops though in my HP as possible to make the HP to LP model cutdown alot faster.

Richard Kain

I'll usually use basic extrusion modeling to flesh out my volumes and "rough" the model into the general structure I'm looking for. Then I'll do a pass to add more geometry to the areas that I know will need it. (head, hands, etc...) Then I start throwing on subdivision levels and sculpting as I go. Once I've got the sculpture looking the way I want, I will create a new object, and start using retopo to create the low-poly mesh based off of the high-poly sculpture. Then I go over the retopo-ed low-poly model and check its joints to insure that everything is good for rigging. Then I'll UV map the low-poly. After that I bake the normals from the high-poly onto the low-poly model's UV map. Then I just work on painting the diffuse map for the low poly and call it a day.

Ghostscape

I'll go high to low for subd, hard-surface modeling.

For complex meshes I'll usually rough out a very basic (boxes/spheres/cylinders) mesh that aproximates the mass so that I can get an idea for what the proportions are, then I'll start making all the high poly pieces and mash that shit together.

Once its done, I build a low poly cage around it, with a combination of edge-extrusion modeling, snap to vertex, and a home-grown tool we've got here.