Modelling must be one of the first things to be done when it comes to 3D art and it must be one of the most intuitive stuff that depends more on practise. But, after 2yrs with Blender, when I finally came down to animation and it appears that there is more to it than that. Books and references on the internet suggest modelling for animation should be done very differently than how it is done for still imaging. Why is it so? What should be done to successfully optimize one's skills to model for animation? If there are any resources that would help us climb up the stairs(that is from very basic principles to advanced level), it would be very helpful if links were provided.

  • $\begingroup$ This is a very large question... for instance, what do you want to animate ? A human character ? or a mechanical device (gears of a watch as example) ? $\endgroup$ – lemon Jun 27 '16 at 11:16
  • $\begingroup$ I heard the 12 principles of animation is a good place to start a checklist vimeo.com/93206523 $\endgroup$ – eromod Jun 27 '16 at 12:18
  • $\begingroup$ @lemon, a complete animation may compromise both, so I'd like to start with the basics. Would it be a mechanical device or a human character. $\endgroup$ – bzal Jun 27 '16 at 14:51

There's no hard rules for making good deforming mesh. Here are some general guidelines:

  • minimize n-gons.
  • minimize vertices with more than 5 edges.
  • maximize quads.

As far as tutorials go, Blenderella is one of the best Blender-based resources I can think of. Steven Stahlberg used to make some good tutorials too, but I can't find them.

Just make sure to test a lot. Even make your deforming rig before all your geometry is added, and start animating it, adding in loops wherever there's no volume preservation. Blender is remarkably good at interpolating weights. As a bonus, you'd be surprised how little geometry you can get away with and still make desirable deformations.

| improve this answer | |
  • $\begingroup$ Is maximize quads only for animation or is it for anything possible(like still imaging)? $\endgroup$ – bzal Jun 27 '16 at 14:50
  • $\begingroup$ It's generally for animation (that is, deforming mesh). That said, quads are often a bit more predictable when it comes to third-party rendering as well, so modelling with quads are a 'safe bet' for still imaging too. The only time quads aren't favourable are in any large, inorganic structures such as buildings. In that case, having extra quads starts to slow down rendering times. $\endgroup$ – Mike Belanger Jun 27 '16 at 14:57
  • $\begingroup$ In blender 1 plane or 2 triangles at the top it will always say 2 tris for both.. so triangles render faster than quads ? even when it's the same amount of vertices connected differently ? $\endgroup$ – user2816 Jun 27 '16 at 15:56
  • $\begingroup$ @MikeBelanger, what about the process that involves also retopo? $\endgroup$ – bzal Jun 29 '16 at 8:30
  • 1
    $\begingroup$ @bzal: I honestly haven't tried retopo, but it looks pretty good. My modelling skills are still stuck in pre-2.5. Maybe its me who needs to check out Blenderella videos :P Jonathan Williamson has a lot of good stuff about topology too. $\endgroup$ – Mike Belanger Jun 29 '16 at 13:06

When people say "modelling for animation", they are referring to making a mesh that deforms well. A model that is used in a single image is easy to position and make look good. A model that gets animated will get deformed and twisted in ways you probably never thought of and needs to look good in the process.

You need to consider how the model will be animated, think of the arm being lifted and swung around, think of extremes like swinging from an overhead bar or doing a high roundhouse kick, your model needs to be deformed by the armature without the shoulders or hips getting kinked and looking unnatural. Some rigs also allow a character's arms to be stretched and bent which also needs to be considered.

Often shapekeys are used to clean up a joint during animation but you still need a mesh that looks good deformed before you make a shapekey that adds a little polish for certain extreme positions.

Not sure of any tutorials that explain this in detail, I think it is more a matter of experience from watching your mesh deform and working out a topology that works for each position.

You can find many rigged models online from sites such as blender cloud or blendswap. Download some of the characters, pose them and look at how the mesh deforms. Enable wireframe display so you can see the topology as it is deformed. Compare how some look better deformed than others. Compare them to your characters.

| improve this answer | |

Few practical suggestions that are more a complement to the other answers (but cannot be an exhaustive list).

Learn the anatomy of your model. Examples : how a shoulder moves when the arm moves, how the leg moves at the hip level, how the skin stretches or folds.

Study separately bending meshes. Examples : front and back of a knee, the inner and outer parts of the articulation of a finger.

Look at how make or inset edge rings. Useful for eyes, mouth, nostril...

Retopology : possibly your model looks fine but some parts need to be redefined because they dont fit the wanted curves for animation. So retopo it (or at least the needed parts).

Use quads essentially, but quads lines need to be aligned to the bending curves (don't go through a quad or inner tris will appear).

Take into account the stretch of the textures.

Rest pose : build your model in an average rest pose, the relaxed pose. Extreme or not natural positions will be harder to rig and skin.

Work with mirror modifier : never make the job twice in particular for the weight painting part (mirror also helps to avoid to paint on the wrong part). But sometimes mirror does not work (example : weighting a long dress because weight influence may need to go over the mirror center).

Facial expressions : test your model by simply moving a vertex with proportional editing mode on. That can allow to test your mesh reaction when deformed before doing any rigging or shape keys.

Look and study the others realizations. But be critic and keep in mind that your mesh could need a bit different approach because of its particular shape.


| improve this answer | |
  • $\begingroup$ is retopo done differently for animation and still models? I saw a face being retoped in youtube that wasn't aimed at it's clean topology. $\endgroup$ – bzal Jun 28 '16 at 8:02
  • 1
    $\begingroup$ @bzal, as far as I know, the principles are the same. But the aim is both to have a clean mesh and curves adapted for bending/folding $\endgroup$ – lemon Jun 28 '16 at 8:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.