AI Pose Generator
Re-posing a photograph is image editing and needs an external engine key. Generating a posed 3D figure works today on the local engine — this page is clear about which is which.
Image generation runs through pluggable engines, exactly like the 3D studio. This deployment ships a local procedural engine that always works and produces real images. External image engines activate only when their API key is configured.
External image engines are not enabled on this deployment. No external image engine is configured, so generation runs on the built-in procedural engine. It produces real PNG output — patterns, gradients and surface textures — driven by your prompt.
Prompt
Pose generation covers two different jobs. The first is editing a photograph you already have: change the stance of the person in it — arm raised, body turned, weight shifted. The second is building a new subject: a 3D humanoid generated in a stance you can orbit, texture, export and print.
Only the second one runs on this deployment. Photo re-posing is generative image editing and needs an external image-editing engine; no external image engine is configured here, so there is no button that pretends to do it. The 3D route is the local engine in the studio, it needs no key, and it returns a real mesh in seconds.
How it works
Open the studio and describe the figure
Sign in and open the studio. Words like character, warrior, figurine, person or 手办 select the humanoid family, while colour words in the prompt and the style preset shape how it comes out.
Generate and inspect it
Run a text-to-3D generation on the local engine. You get geometry plus a thumbnail, and the viewport has ten shading modes — wireframe and normal are the fastest way to judge the silhouette before you commit to a print.
Export and print
Download the GLB, or re-export as STL, OBJ, PLY, glTF, 3MF, DAE or OFF from your assets. Models are normalised to a 100 mm longest edge, so scale deliberately in millimetres before you slice.
AI Pose Generator
A real humanoid, not a billboard
The figure family is assembled from solid primitives — body, head, arms, legs — unioned into one mesh, unwrapped and given procedural PBR maps. It has volume from every angle, which a photo cut-out never does.
Generated in a natural stance
The figure comes out standing with its arms at its sides and its weight on both feet, rather than as a raw T-pose file. That is what makes the thumbnail readable at a glance and the silhouette printable without re-modelling.
Style presets that change the build
Realistic, stylized, low-poly, toy and statue. Low-poly decimates the mesh for game-style assets, toy rounds the forms, statue keeps a heavier and smoother surface.
Texture you can switch off
Texture and HD texture are separate options with separate credit costs, so a silhouette test does not have to pay for maps you will never look at.
Every export format that matters
GLB for the web, STL for the slicer, OBJ and PLY for DCC tools, 3MF when units and colour have to survive the trip.
Honest about the gap
What this deployment cannot do is change the pose of your photograph. That is an external image-editing job, and the page says so instead of returning a failure you have to decode.
Image tool questions
Can I change the pose of a person in a photo?
Not on this deployment. Re-posing a photograph is generative image editing: an external engine has to repaint the body and the background around it. No external image engine is configured here, so the tools state that plainly instead of failing halfway through.
What works today, then?
Generating a 3D figure in a stance, in the studio, on the local engine. Type a prompt, run text-to-3D, and you get a real humanoid mesh you can orbit, texture, export and print — with no API key involved.
Does the prompt control the pose?
It controls the subject and the look: words like warrior, character or 手办 select the humanoid family, and the style preset changes how the forms are built. The figure is generated standing naturally. Per-joint control — 'raise the left arm 40 degrees' — is not part of this build.
Can I rig or animate the result?
The output is a static mesh. Rigging and animation belong in Blender, Maya or a game engine; the studio's job is to hand you clean geometry to start from.
How long does a figure take?
The local engine generates, unwraps and bakes in a few seconds. There is no external queue to wait in, because there is no external service in the path.
What does a generation cost?
Credits, like any studio run: 10 for a standard generation, 20 for quality, plus 4 if you want texture maps. The free plan starts with 100 welcome credits, which is enough to try a figure several times on the local engine.
Can I print the figure?
Yes — export STL and slice it. Models come out normalised to a 100 mm longest edge, so scale to the size you actually want first, and hollow anything taller than roughly 100 mm.
Will photo re-posing be added?
The slot is already in the codebase: the studio's engine picker lists external engines with their configuration state. But a slot is not a capability, and this page will only claim it once an operator configures a key and it returns an image.