Conjure · text to 3D · extension
Type it. Watch it get sculpted.
Conjure will turn a text prompt into a 3D asset: AI that models in voxels the way an artist does — sculpting, comparing, refining — landing the result as engine-ready glTF in the godspeed engine. It is not on sale yet: the apparatus runs end to end, the trained models are not good enough, and we would rather say so than sell it. When it needs to know more about what you asked for, it researches live through alice — which is why every conjure subscription includes alice at no extra charge.
Coming soon
Conjure’s models are still training, so it is not on sale yet. Every other combination is — and BigBrain Complete includes conjure the day it ships, at no extra cost, at whatever price you locked in.
What it does
Prompt to sculpture
Describe the thing; conjure sculpts it in voxels, renders its own work from 62 viewpoints under varying light to check it against what you asked for, and keeps the edits that scored better. No lookup table maps a word to a shape, so two runs of one prompt genuinely diverge.
Live research refinement
Unsure what a “fennec fox” looks like? Conjure asks alice, studies the references, and refines — the research loop is part of the product, not a demo trick.
Engine-ready output
Results are designed to land as standard glTF 2.0 through godspeed’s byte-preserving pipeline — assets you can immediately sculpt, paint, and ship like anything else in the engine. This is the seam still being finished, and it is the main reason conjure is not yet for sale.
Runs on your machine, not our cloud
Conjure’s models ship as versioned artifacts baked into the product — no cloud dependency for generation, no per-asset fees, no metered credits, and no ash purchase required.
Deeper dive: the conjure documentation.