Vi3W
Skip to content

Guides

Animate a picture in 3D: turn a photo or drawing into a moving model

Turn a photo or drawing into an animated 3D model: image to 3D, auto-rig, then walk, wave or dance. Picture tips, steps and limits.

5 min read

A cartoon fox mid-jump with ghosted poses trailing behind it along a glowing arc
Illustration. A picture becomes a model, the model gets bones, the bones get a move.

Quick answer

How do you animate a picture in 3D?

Turn the picture into a 3D model first, then rig and animate the model. An image-to-3D generator builds a textured mesh from the photo or drawing, an auto-rig adds bones, and a move like walk, wave or dance is applied to the bones. The result is a 3D file you can view from any angle, unlike a photo-to-video clip.

Video or 3D: decide first

"Animate a picture" usually means one of two things, and they're built differently:

  • Photo to video. A video model takes your photo and makes a few seconds of motion: a blink, a head turn, a wave. It's quick and it looks great on social media. But it's flat. The camera angle is fixed, and you can't use it in a game, an AR filter or a 3D scene.
  • Photo to animated 3D model. The picture becomes a mesh with textures, the mesh gets a skeleton, and the skeleton gets motion. The result is a GLB or FBX you can turn, light, put in Unity, put on a website, or render from any angle.

This guide is about the second. If a video clip is all you need, any photo animation app will do it. If you want the character to exist in 3D, read on.

The three-stage pipeline

  1. Picture to model. An image-to-3D generator builds a textured mesh from the picture.
  2. Model to rig. An auto-rigger adds bones and skin weights.
  3. Rig to motion. A motion clip (walk, wave, dance) plays on the bones.

In Vi3W those are: generate from your image, Add bones, then Make it move. You can do it in the app, or from a Claude or ChatGPT chat with the Vi3W MCP server connected, where it's three short messages.

Three platforms: a sketch card, a grey model and a finished mushroom house
Illustration. Each stage builds on the one before, so a weak picture weakens everything after it.

Stage 1: pick the right picture

The picture decides most of the outcome. For something you plan to animate, it needs more than a good look:

  • Full body, nothing cropped. Feet and hands in frame. Generators invent what they can't see, and invented feet don't walk well.
  • Neutral pose. Arms away from the body, legs apart. A character with crossed arms becomes a mesh with arms fused to the chest, and no rig can separate them.
  • One subject. Your dog, not your dog on the sofa with a cushion.
  • Plain background and even light. Shadows on a wall can become geometry.
  • Front or three-quarter view.

If your photo doesn't meet these, don't fight the generator. Make a better reference image first. In Vi3W you can turn a drawing into a clean reference image before building 3D, and ChatGPT, which can generate images, will take a request like: "Redraw this character standing in an A-pose on a plain grey background, keep the outfit and colours." Approve that image, then build from it. Best photos for image to 3D has the full checklist.

Drawings and kids' art

Drawings work well, often better than photos, because the style is clear and the background is usually blank. A child's drawing of a monster can become a chunky, charming 3D monster. Expect the generator to interpret, not trace: proportions get evened out, and the back of the character is invented.

Stage 2: generate and check the model

"Make a 3D model from this picture." Then open the model in the 3D view and turn it around. Check:

  • Is the back plausible?
  • Are the limbs separate from the body?
  • Are there floating bits or holes?

If the mesh is very dense, rebuild it lighter before rigging: "Rebuild it at about 8,000 faces." Clean shape does this (500 to 20,000 faces) and bakes the textures across. A lighter mesh rigs more cleanly.

Want a different look before animating? "Repaint it in a clay style" or "make it look like painted wood" uses New paint job, which changes textures without touching the shape. Do it now: every move you add later rigs the model again, so changes after rigging cost more.

Stage 3: bones and a move

"Add bones. It's a person." Or "It's an animal" for pets, creatures and anything on four legs.

Then: "Make it wave." Or walk, run, jump, dance or idle. You can describe a move too ("a happy hop", "a kick"), and Vi3W matches it to its library of motion clips. Walk and Run come with the rig.

Download the result as GLB for the web and viewers, or FBX for Unity and Unreal.

Where to use an animated 3D picture

  • A game character from concept art, ready for blocking out gameplay.
  • A pet or mascot that waves on a website, using a GLB in a web viewer.
  • AR stickers and filters, where the model needs to be seen from all sides.
  • A child's drawing come to life, as a gift or a print-and-animate project.
  • A Roblox or VRChat avatar prototype (check each platform's rig requirements first).

What it can't do

  • Keep a realistic likeness. A photo of a real person becomes a convincing figure, not a recognisable portrait. Stylised subjects survive the trip far better.
  • Animate faces. No lip-sync or expressions. Bodies move, faces stay put.
  • Handle odd anatomy. A snake, an octopus or a winged dragon is beyond standard humanoid and four-legged rigs.
  • Move cloth and hair naturally. A long dress or long hair is weighted to the body and moves with it, stiffly. For flowing cloth you need physics in your engine or hand animation.
  • Invent new choreography. Moves come from a library. A specific TikTok dance won't be in there.

For more on the animation side, see can ChatGPT animate a 3D model and AI walk cycles from a prompt. Vi3W also covers the image to 3D stage on its own if you only need a static model.

Bring one picture to life

Pick a full-body drawing or photo, generate the model, Add bones, then Make it move with a wave. It's the quickest way to see the whole pipeline end to end. Open Vi3W.

Frequently asked questions

image to 3d to animation

The pipeline has three stages: image to 3D model, model to rig, rig to animation. In Vi3W that's generate from your image, Add bones, then Make it move, and you can run all three from one Claude or ChatGPT chat with the MCP server connected.

ai 3d animation from image

Use a full-body image in a neutral pose on a plain background. The generator builds the mesh, and the rig and moves come after. Expect a good result for stylised characters and animals, and an approximate one for realistic faces.

animate a photo with chatgpt

ChatGPT on its own can turn a photo into a short video with its video tools. For a 3D animated model, connect a 3D app like Vi3W over MCP, so ChatGPT can generate the model from the photo, rig it and apply a move.

Last reviewed