AI image generation
Generate 2D images from a text prompt and turn the one you like into a 3D model, so you control the look before you generate in 3D.
Generate an image first when you have an idea but no reference. Then turn it into a model with Image to 3D. If you already have a reference image, use Image to 3D directly. To brainstorm and explore concepts in a chat, use 3D Agent.
Generate an image and turn it into 3D
To generate an image and turn it into a 3D model, follow these steps:
-
In the left sidebar, click Image.
-
In AI Model, select a model. Each model has its own cost, from 3 to 12 credits per image.
-
In Prompt, describe the image in up to 800 characters, for example "a low-poly cartoon fox sitting, front view, white background". You can write the prompt in your own language.
-
In Number of images, select how many images to generate, from 1 to 8. On the Free and Starter plans, you can generate up to 4 images a day.
-
Optional: Adjust the other settings:
-
Refine Prompt: Meshy refines your prompt for 3D generation by default, with clearer structure, defined angles, and simpler backgrounds. This can make the style more cartoonish. To use your prompt as written, click the wand icon in the prompt box to turn it off.
-
Image References: Add up to 5 images to guide the result.
-
Styles: Select a preset style, such as Japanese Anime, Western Realism, or Chibi Blind Box. Styles is available with the Nano Banana Pro and Nano Banana 2 models. When you select a style, Multi-view and Pose turn off.
-
Aspect Ratio: Select the image shape. The default is 1:1.
-
Multi-view: Turn on to generate consistent front, side, and back views.
-
Pose: Select A-Pose or T-Pose for a character you plan to rig.
-
-
Click Generate. The panel shows the time and credit cost before you generate.
-
In the list of generated images, hover over the one you want to use.
-
Optional: To change the 3D generation settings, click the settings icon next to Image to 3D.
-
Click Image to 3D. Meshy starts generating a 3D model from the image and opens the Image to 3D tab in Model.

Prompt tips for 3D-ready images
These tips help the image convert cleanly to 3D:
-
Ask for a "front view" or "3/4 view", which are the best angles for 3D conversion.
-
Add "white background" or "clean background" so the background doesn't interfere.
-
Describe the complete form. Describe the entire object, not only the head.
-
Avoid complex scenes. A single subject converts best.
FAQ
- How do I generate an image from text in Meshy?
- In the left sidebar, click Image, enter a prompt, and click Generate. To turn an image into a 3D model, hover over it and click Image to 3D.
- How do I turn an idea into a 3D model?
- Describe it in AI image generation to get a reference image, then open it in Image to 3D. Generating the image first lets you check the style and details before you spend credits on a model.
- How do I write prompts for 3D-ready images?
- Ask for a front or 3/4 view and a white background, describe the whole object, and keep to a single subject.
- What image styles can Meshy generate?
- Preset styles include Japanese Anime, Western Realism, and Chibi Blind Box. Styles are available with the Nano Banana Pro and Nano Banana 2 models.
- Can I send a generated image directly to Image to 3D?
- Yes. Hover over the image and click Image to 3D. Meshy starts generating a 3D model from that image.
Next steps
-
To set more options for the 3D model, see Image to 3D.
-
To write prompts that produce better results, see Prompting best practices.
-
To explore concepts in a conversation before you generate images, see 3D Agent.