Text-to-3D

Text-to-3D generates a 3D model from a written description, no image required. You describe an object in words, TripoSR generates it on a cloud GPU, and you download the result as a GLB file. Use this tool when you have an idea in mind but no reference photo.

Prompting for 3D

A 3D prompt is not the same as an image prompt. You are describing a single physical object, not a scene or a mood, so keep it concrete:

• Describe one object. "A wooden rocking chair" reconstructs far more cleanly than "a cozy living room", which has many objects and no single subject.
• Name shape, material, and color. Details like "a matte black cast-iron kettle with a curved spout" give the generator clear geometry and surface cues.
• Skip scene and camera language. Words about lighting, background, camera angle, and atmosphere describe a photo, not a 3D object, and add little here.
• Keep it focused. A short, specific noun phrase usually beats a long paragraph. Add only the details that define the object's form.

What to expect

Text-to-3D produces a textured mesh that matches your description's overall shape and material. Simple, well-defined objects (props, furniture, tools, toys) work best. Highly detailed or ambiguous prompts, scenes with multiple objects, and fine text or logos are harder and less predictable. As with Image-to-3D, you can rotate and zoom the preview before downloading.

Steps

Related