One paragraph about a half-elf ranger became thirteen assets of one character, for 72.44 cr in one run.
A character in a game is more than one picture: a turnaround to model from, a portrait and expressions for dialogue, an icon for the interface, a 3D model that walks, a voice.
The question this run answers: can one written brief produce all of them, reading as the same character, in a single run?
Two images are made from the brief — the identity and the portrait. Every other image is image-to-image from one of them, so the costume, palette and face carry across.
Aelin, a half-elf ranger in weathered brown leather with a green hooded cloak, dark hair with one white streak, a scar through the left eyebrow, sharp green eyes, a longbow on her back.
Identity image
The reference everything else follows, made from the brief and a style.
FLUX.1 [dev] on Fal
Turnaround — front, side, back
Image-to-image from the identity.
FLUX.1 [dev] on Fal
Portrait, head and shoulders
From the brief again, with the same seed and style.
FLUX.1 [dev] on Fal
Expressions — neutral, angry, happy
Image-to-image from the portrait.
FLUX.1 [dev] on Fal
Icon on a clear background
Image-to-image from the portrait, then a background cutout.
FLUX.1 [dev], then BiRefNet on Fal
3D mesh, rig and walk animation
A mesh from the identity image, rigged, then given a walk cycle.
Tripo
Voice line
One spoken line in the character's voice.
ElevenLabs
| Step | cr |
|---|---|
| Nine images, 3.25 eachFLUX.1 [dev] on Fal | 29.25 |
| Icon cutoutBiRefNet on Fal | 0.65 |
| 3D meshTripo | 19.50 |
| RigTripo | 16.25 |
| Walk animationTripo | 6.50 |
| Voice lineElevenLabs | 0.29 |
| Total, one run | 72.44 |
A credit is a cent, so the run came to about $0.72. Figures are the charged step costs of the third run — the kit on Discover.
All thirteen are on Discover, each with the model, provider and cost of the step that made it, and the workflow opens on the canvas from there.
The first run finished every step and still failed as a kit: the back view faced the front, the portrait was a full body, the three expressions were one face. The fix was not in the prompts.
Each run cost the same 72.44 cr; the lesson cost two of them. Run three is the kit on Discover: a head-and-shoulders portrait, an angry face that reads as angry, an icon cut clean.
The kit is live on Discover — every asset, with the step that made it. Request an invite to run the character kit on your own brief — for now the 3D and voice steps need your own Tripo and ElevenLabs keys.
← All case studies