Hunyuan3D v3.1 Rapid
Runs Pro in the Image to STL tab and Fast in Text to 3D. It returns an untextured mesh, which is all a single color print needs.
Turn a photo or a text prompt into a 3D model with Hunyuan3D or TRELLIS.2, check it in the 3D preview, then download STL, OBJ or GLB for your printer. AI 3D models start at 25 credits.
AI builds a full 3D shape from your photo. Untextured, made for printing.
Credits come back automatically if a generation fails on our side.
AI 3D modeling without CAD: three AI models build a full object, sides and back included, from a photo or a few words.
Runs Pro in the Image to STL tab and Fast in Text to 3D. It returns an untextured mesh, which is all a single color print needs.
Runs Ultra and Text to 3D Pro. It returns a model with color textures, can take back and side photos on Ultra, and can add PBR materials.
An open image to 3D model in the Trellis-2 tab. You set the resolution and texture size, and it always returns a textured model. Read the TRELLIS.2 guide.
A photo fixes the shape and proportions of something that exists. A prompt can describe anything, but you control less of the result.
Upload one photo, add back and side views on Ultra, or switch to Text to 3D and describe one object.
The Generate button shows the credits. They are taken when the job starts, and most models finish in 1 to 3 minutes.
Spin the result in the 3D preview, type the print height in millimeters and save STL, OBJ or GLB with no extra charge.
Credits are charged once per generation. Every download of the result is included.
Hunyuan3D v3.1 Rapid from one photo. An untextured mesh for single color prints.
25 creditsHunyuan3D v3.1 Pro with color textures. Back and side photos and PBR materials cost 15 credits each.
35 credits, add-ons +15Hunyuan3D v3.1 Rapid from a prompt of up to 200 characters. Untextured, ready to slice.
25 creditsHunyuan3D v3.1 Pro from a prompt of up to 1024 characters. Textured, with optional PBR materials.
40 credits, PBR +15TRELLIS.2 from one photo at 512, 1024 or 1536 resolution, with textures up to 4096 px.
25, 30 or 35 creditsA raised relief of a picture, built in your browser with a free 3D preview. For logos and signs, not full objects.
2 credits per STLText to 3D builds what you describe, so describe something a printer can make.
Ask for a single thing, such as a sitting fox, not a fox and a rabbit in a forest. Scenes and backgrounds are hard to print.
Ask for a chunky, stylized or rounded look. Thick forms print more reliably than realistic fur, hair strands or lace.
Swords, antennae, whiskers and open fingers snap or fail. Ask for arms close to the body or a tail wrapped around the feet.
Say how it stands and what it stands on. Sitting upright on a round base gives a flat bottom that holds the bed.
A chunky owl figurine sitting upright on a round base, wings folded against the body, stylized, no thin parts.
For a single color print, Fast is enough. Pro adds color textures for painting reference, renders and viewers, and allows longer prompts.
Every result downloads as an STL sized in millimeters, so this works as an AI STL generator and an AI 3D print generator in one step. A few checks turn it into a clean print.
Most print problems start with a cluttered photo or a prompt full of thin parts. Prompt tips are above. Photo tips are on the TRELLIS.2 page.
AI models have no real scale. Type the print height, 10 to 300 mm, before you save; STL and OBJ follow it and the GLB keeps its original size.
A rounded bottom touches the bed at one point. Cut 1 to 2 mm off in your slicer, or ask for a round base in the prompt.
Keep walls and small parts at least 0.8 mm thick, two lines of a 0.4 mm nozzle. If details are thinner, print the model larger.
AI meshes can have small holes or stray floating pieces. If your slicer flags mesh errors, run its repair and delete loose bits.
Turn the model so supports land on its back, not its face. Tree or organic supports suit the curved shapes of figurines.
Credits, refunds, file formats, timing, commercial use, back and side photos, PBR, and text versus photo.