Sculpt first or last?

KRakarth

As I have been reading over the last few days I have got myself a question. The last time I was involved in game art (yes it was a long time ago) the low poly would be built first and then the high poly would be sculpted, normal mapping would occur etc. etc.. It would appear that now sculpting is done first and then the low poly is generated (retopologised) from the sculpt. Anyone care to weigh in on which way round they prefer to do things and why? Is it different for different assets - organics, hard edged, environments?

GarageBay9

When the high poly is important as a centerpiece or hero asset, I prioritize time for a detailed sculpt first and derive a low poly from it for a bake.

When an asset just needs to have some normal map details so it doesn't look like I stole it from 1997, I build the low poly first and then buff it a little with nDo or some quick kitbashing, and move on.

Musetatron

I think cars and guns are easier to poly model first.

ExcessiveZero

Depends but 99% of the time I will make a low poly model in max/maya/Blender even if it is an organic sculpt as a base mesh, I can get the head and shoulders and other pre sculpting forms out quicker than using dynamesh or Zspheres.

but rarely is this my low poly model ever returned to its just a base to work off, then I proceed once I have the high poly to retopo or poly reduction whatever is needed, sometimes I will go Base-Medium-High-Strip the Medium to Low then bake High to Low.

Then more Organic heads it seems easier to go, Base-Sculpt-Retopo to low then bake.
But there are many different workflows and it all falls to personal preference and what works quickest and gets the best result.

pior

The problem with this question is that you will get many answers from Zbrush aficionados claiming that sculpting first is the way to go (no offense intended here ! )

But the problem with that is that since you come from a lowpoly background, you are very likely used to working with extremely responsive scenes (running at over 60fps at all time, allowing you to easily cut and merge models, applying materials to faces to test out color schemes, and so on). All these operations are usually very slow to perform in sculpting packages (example : no way to select two "subtools" at the same time in Zbrush), thus most likely causing frustration.

Regular, non-sculpted highpoly is another beast altogether. By nature it is much more flexible than straight up sculpting, and has many of the advantages of lowpoly modeling : fast response time, a flexible 3d environment to work with, and so on.

My advice would be : 

- If the design is 100% locked, then I can see cases where one can jump straight to highpoly, even sculpting. But concept artists are not machines, and if any edit is required down the line you might have to roll back quite a bit.

- If there is *any* loose end on the design being worked on, use the most flexible technique you are confortable with. It could be straight up box modeling, primitives mashing, or even some very rough sculpting to validate the design before going forward.

At the end of the day, the way I see it is not so much about which tool pushes the more polygons, but rather, which tool/approach allows for the fastest response time and maximum flexibility. The ideal environment is the one allowing to work non-linearly, on all fronts, at any time, without slowdowns. Giving priority to these factors will allow you to pick the tool that fits you best.

A traditional 3d modelling package with a module allowing you to also sculpt freely is probably the way to go in your case.

Good luck !

KRakarth

Thanks all for your replies, especially Pior.  I have access to Autodesk's suite so it will be Max and Mudbox for me. I have been recently working on some straight up high poly for 3d printing so it's going to be an interesting journey.

EtotheRic

I would drop mudbox. Unfortunately no one uses it these days especially for games, just zbrush. Not really unfortunately it is just pretty out of date. 

pior

@EtotheRic : it might be unwise to advise someone against a specific tool (perfectly capable of performing the task at hand) without letting him/her the chance to evaluate it first. Zbrush has a huge market share, but it is far from perfect and has deep rooted issues.

Ruz

 there is other work out there apart from sulpting you know. it may seem like the in thing, but there are other ways to go. I do a lot
of subd work these days

Burpee

EtotheRic said:

I would drop mudbox. Unfortunately no one uses it these days especially for games, just zbrush. Not really unfortunately it is just pretty out of date. 

Dan Roarty

JacqueChoi

I think in an actual production environment, it's easier to start with a low poly (at least a proxy) to ensure joint locations and such are properly placed (Most characters in games share rigs), along with all kinds of other technical type constraints. Seams, shared vertex order,  etc. etc..

gsokol

Always depends on the situation.  If you are doing some hard surface stuff, sometimes easier to tough out forms for low poly first, and subd it up.  Sometimes sculpting is a more organic process and its easier to retopo.  Just things both ways so you can judge when to do which.

kanga

I started off box modelling, and when zbrush was new I used to use it for detailing. Zbrush has changed a lot and it is far quicker to concept in it and also do quite a bit of modelling too. In general I dont see how building something and deciding where the best place for geometry is without having a template to work on top of. I am sure there are props and specialized requirements for game assets where this isnt the case, but generally I find starting an object in zbrush faster and easier. I use 3dsmax and still swing between both apps but I am doing more in z lately.