Input
Text prompt — no image required
Generative · 4 Credits
No artwork yet? Describe it.
The Relief Generator skips the source-image step entirely. Describe a subject — an owl, a stag, a crest, a portrait bust — and a diffusion model trained with a relief-style component carves a sculpted, medallion-style depth map directly from the prompt.
Built for dimensional, carved detail
The Relief Generator creates depth from nothing rather than reading it off an existing photo — it's aimed at generative, sculptural medallion-style output rather than reproducing a real object's geometry. Switch on the AI Depth refinement pass afterward and the same cloud depth-estimation model used elsewhere in Depth2Beam turns the raw generation into depth that reads like anatomy rather than a flat brightness map. That combination makes it a good fit for coin and medallion-style engraving where you want deliberate, carved relief detail. The dedicated Coin Depth Maps workflow is this same generator, locked to a round medallion composition and exporting nothing but the coin-cropped height map.
A Background setting controls whether the subject sits on a flat, deeply recessed backdrop or a textured one (stone, wood, metal) that carries its own real depth, with any lettering or panels stepped above it. Generated output drops straight into the same console as every other engine — preview it as a Height Map or Lit Relief, crop it to a panel shape, and export it exactly the way you would any other relief.
Background
Flat or Layered (textured)
Cost
4 credits / generation, +1 with AI Depth refinement
Why does generating a relief cost credits?
Generating a relief from a prompt is a real cloud AI job, same as the optional AI Depth refinement pass on top of it — see exactly how the shared credit allowance works.
Read Why Credits →