Selection criteria
Which of these templates to pick
The thirteen split into three jobs. The realistic display figures — collectible toy, action figure, miniature figurines — keep normal body proportions and sell the effect through material and product lighting; pick these if you want something that looks like it came out of a box. The six chibi templates break proportion deliberately, with an oversized head and a short body, and read as blind-box or gashapon product; they are more forgiving of an imperfect source photo because the face is simplified anyway. The diorama and turnaround templates put the figure into a small scene or rotate it on its base, which is what you want if the result is going somewhere it will be watched rather than glanced at.
What you actually get back
Stills come out at high resolution with no watermark, and the video templates return a short loop — a rotation on the base or a small movement inside a diorama — rather than a long clip. Video costs meaningfully more credits than a still because it renders many frames, so it is worth settling on the style with a cheap still first and only then spending on the animated version. Every template here works from a single uploaded photo; none of them need a prompt written, and none of them need more than one image.
Usage guidance
- Pick a lane first: realistic display figure, chibi proportions, or an animated turnaround.
- Upload one photo — single subject, face and upper body visible, even light, plain background.
- Generate a still version first, even if you want video, so you can judge the style cheaply.
- If the face reads soft, re-shoot or re-crop the source rather than regenerating with the same input.
- Once the style is right, run the animated template for a rotation or diorama loop.
- Download at full resolution — there is no watermark on either stills or video.
Related models
Where it goes wrong, and why
Three failure modes account for almost everything. A group photo gives each face too few pixels to survive being shrunk to figure scale, so faces go soft — use one subject. Strong backlighting in the source leaves the model guessing at facial structure, and it guesses badly. And a busy background competes with the figure for attention in a frame that is supposed to read as product photography; a plain wall behind you in the source photo materially improves the result. None of these are model limitations you can prompt around — they are input problems.
How this differs from a 3D print or a real figure
Worth being clear: this produces an image of a figurine, not a file you can print. There is no mesh, no STL, no geometry — the output is a photograph-style render that looks like a manufactured object. If you want something physically made, this is the step before that, useful for deciding what the thing should look like. Several people use these renders as reference for a commission rather than as the end product.
