Low- & High-poly modeling?

[Flinet]

Hi!

Just wonder how you doing when you are going to model something in both high and lowpoly? I mean du you begin with the lowpoly mesh and finish it and the remake the lowpoly version to the highpoly one? Or otherwise?

I'm just curious wich workflow that is better

//Jimi

[chrizz1]

Well its reallly your own prefrence as to what you do, I prefer to build a high poly first then make the low poly, as I find building a high form a low tricky at times, experiment with both to decide every one will say something different.


chris

[Flinet]

Aha ok, but if you make the highpoly first, couldn't there be a problem to place some lowpoly objects into the same place as the highpoly version? If you want to take the normals i mean. Hard to explain but you don't want one part of the mesh to stick out more on the one side.. Sure that's what modelings about, to fix things, but sometimes it's harder to get the exact result And I belive it might be harder to get this if you model the highpol first.. Or what do you think?
Sorry if im blurry

[sickboy]

Hi,

Personally I'd build the high res version too and then reduce a copy of it. To be perfectly honest though for making a normal map I'd build your low/med res model then use mudbox or Z Brush to create your normal map.
Calculating it from two models which is what I gather you mean can cause a lot of faults, plus the fact it's so time consuming!

[Talon]

Usually, I build the high poly first. Then a mixture of reducing the high poly mesh to meet the low and remodeling parts to fit around the high poly.

I find that ensures that the low poly is a nice, close fit to the high poly. There's not as much guarantee that it will work as smoothly the opposite way.

[Flinet]

Aha nice! Well i have been doing the opposite, will try the other way on the next project!

But then if I would like to take the normalmap from the highpoly version, is it better to do that in Zbrush? Cause Im using Xsi's Ultimapper. Zbrush might be faster?

[retardedmonkey]

Yeah I tend to do the HP first as well now. I tend to block in the model, which is pretty much the LP...then go into finishing the HP. The nice thing about that is I can go back to any of those block-in saves and grab lower res parts to optimize. Just make alot of saves and you'll be fine :P

Altho if I need to make the model faster, I tend to make the LP first and then just make a new save so that I can convert the LP to a fast turbosmoothed model for baking. Tends to take a day or two which is faster for me than making a whole HP model as I get carried away with them when I do.

As far as Zbrush, from what I've seen the Zmapper does work quite well...and fairly fast. I haven't had a fair test with it tho as what I rendered out I had to project and it didn't like it that much. Just make sure you keep your lower subdivision levels and you'll be fine.

[Flinet]

I see. Yeah that's true, a long as there is alots of saves I could just go back that way

I have a plan of making a robotic character as the next project, would the Zmapper be a good choise to use then? If there is alot of "hard surfaces"? Or is it mostly organic stuff I should use the Zmapper for? I feel like it shouldn't matter but I haven't used Zbrush so much. But that will change today

[Ratteler]

I do Lowpoly first. It's the only way you can ensure you're not ditching details while staying in your polygon budget. You can always cut in more detail for the High res model, but it's a lot harder to take it out again later. This also makes it easier to manhandle the UV maps. I've never seen a well done UV that was automatically generated.

[thanoz]

Low poly first, then High poly. Let me explain why this is the only way (even though I know some do it otherwise).

First off UV MAPS. If you do low poly first you can do a MUCH better job with the UV map. Rather than dealing with folded over and extruded shapes that need to be carefully Relaxed painstakingly (such as nostrils, eyelids, ears and fingernails). If you UV map your low poly mesh first, you set up the seams early and have a WAY better UV layout for later on (which is vital later when applying normal and color maps).

2nd issue here, is form and proportion. The longer you spend forming the low poly and planning it out, the less vague the shapes will be and the better the high poly will sculpt itself when you up-rez the mesh. A good sense of proportion and a good silhouette is something that ONLY the low-poly can determine, no normal maps can fix those two issues, and you do NOT want to be adjusting the length of arms, legs, or silhouette using lumpy brush tools in Zbrush (unless your character is made of mashed potatoes).

3rd, TIME management. Your Low Poly model is THE ONLY model that you will use, it has to be perfect, it can't be an afterthought, so it needs THE MOST PLANNING. Fortunately, if you start on the Low Poly first, plan out the flow of the mesh carefully (smart edge loops, good anatomical shapes) a great deal of the high rez sculpt will be done for you just by applying a smooth.

The final detail sculpts are extremely fun, but you can do 90% or more of your high rez work in 2d programs like Crazy Bump or the Nvidia normal map plug in because most of your character is clothing (unless your character is naked ALL THE TIME...then you might want to revisit the concept).