What is the best way to model based on cartoon images?

Say, for example, I want to make Pokemon models. I can search the web and find pictures of squirtle, from all different angles. Maybe not ideal photos, because squirtle is a cartoon–these aren’t photographs of a real physical object. But assume that I’ve found pictures of the character from at least two different angles, and start from there.

What is the best method for generating the model?

I’ve found four options:

  1. Trace using bezier curves, and then extrude.

This only traces two dimensions of the subject, making it useless for character modeling. (Unless there’s a way to use this as an intermediate step.)

  1. Build a blocky model around the subject, and then subdivide and refine.

  2. Using planes, ctrl+click to grow out panels over the surface of the model, working in front and side views simultaneously.

  3. Glance at the reference now and then, but build the model basically from the ground up using whatever method suits you. (You might use the reference photos to size primitives for a basic construction. Say, a sphere for the body and a sphere for the head.)

My aims here are for accuracy and workflow efficiency.

Well, there’s always the method professional studios use: build a clay or sculpy model using the concept drawings, then digitize (or take reference photos of) the model.

Tom Williams has a tutorial over at 3d.sk showing how to use wrap around sketches as modeling references, it might be possible to extend his set up to drawings where the points of view are not all on the same plane.

Of your four choices, I’d recommend number 4.

use two and three. put the image planes in then add a cylinder or cube model the shape from the front, then the side. then refine from there.

You’ve asked a charged question, actually (one that starts flame wars).

There is the box method (your #2) which is preferred by most people I think.

There is the poly-by-poly, click-and-extrude method, which is your numbers one and three (the process is the same whether you are using a mesh or a curve). [This is my method of choice.]

Your #4 is not the same thing as the others. Whether you exactly trace a reference or just freehand it depends on your need and/or skill level given the subject.

In all cases, you should try each way and see what you get. You will prefer one over the other. Once proficient with your preferred method your workflow is sufficiently efficient.

Hope this helps.

So there’s no five, six, or seven that I don’t know about? It seems like there should be some “magic bullet” method, or at least that cartoons should be easier than real people.