Text to Image allows you to generate an AI image using a text prompt, which you can then use as the input for Meshy's Image to 3D feature. Choose from multiple AI models to match your style and budget, iterate with artistic control, and convert your final image into a 3D model.
Plan Requirements
Access to Text to Image depends on your subscription plan:
Free users — Text to Image is not available. The Generate button shows an upgrade prompt.
Starter users — Nano Banana Pro with Multi-View enabled requires a plan upgrade.
Pro and above — Full access to all models and features.
Choosing an AI Model
Text to Image supports four AI models. Each produces different results and uses a different number of credits per image. Select your model from the dropdown in the Image Workspace before generating.
Model | Credits per image |
Nano Banana Pro | 9 |
Nano Banana 2 | 6 |
Nano Banana | 3 |
GPT Image 2 | 9 |
Generating an Image
Enter a text prompt describing what you want to create — for example: "Stylized Low-Poly Dragon Sword, with Ruby Gems." You can write in your native language.
Stylization
Apply an art style to shape the visual direction of your generation:
None — Default (no style applied)
Japanese Anime
Western Realism
Chibi Blind Box
Create Style — Define and save a custom style
Aspect Ratio
Select the shape of your generated image:
1:1 — Square (recommended for Image to 3D workflows)
16:9 — Landscape
9:16 — Portrait
4:3
3:4
Note: Enabling Multi-View automatically locks the Aspect Ratio to 16:9.
Number of Images
Choose how many variations to generate (1–8). More variations help you compare results, pick the strongest composition, and refine your direction. Each image costs credits based on the selected model.
Multi-View
Toggle Multi-View to capture multiple angles of the same subject in a single generation.
Produces 1 composite image and 3 individual angle images
Automatically sets Aspect Ratio to 16:9 — other ratios are disabled while Multi-View is active
Useful for character design, object modeling, and improving 3D reconstruction accuracy.
Pose
For character subjects, choose a starting pose:
None — Default generation
A-Pose — Recommended for 3D modeling and rigging
T-Pose — Recommended for 3D modeling and rigging
For environments or props, leave Pose set to None.
Image References (Optional)
Add up to 5 reference images to guide the generation — upload from your device or send a previously generated image using its three-dot menu.
Upload methods: Click, Drag & Drop, or Paste
Supported formats: PNG, JPG/JPEG, WEBP — max 20MB per image
Credit Cost & Generation Time
Before generating, the panel shows:
Estimated time — e.g., 15 seconds
Credit cost — credits per image × number of images (e.g., Nano Banana Pro at 4 images = 9 × 4 = 36 credits)
Both values update automatically when you change the model, image count, or settings.
Click Generate
Press Generate to create your images. Once complete, select your preferred result, refine further, or send it directly to Image to 3D.
Working with Generated Images
After generation, each image supports the following actions via its menu or toolbar:
Remove Background
Strip the background from a generated image in one click. Useful before sending to Image to 3D or using the image as a reference for the next generation.
Download
Save any generated image to your device as a PNG file. Files are named Meshy_AI_xx by default.
Full-Screen Preview
Open any image in full-screen to review detail. Navigate between images using the left/right arrow buttons (or your keyboard arrow keys) and the thumbnail strip at the bottom.
Image Helper
The Image Helper is a built-in chat interface that helps you craft, expand, and batch prompts before generating. Open it from the prompt panel.
Key Capabilities
Prompt expansion — Describe a rough idea and Image Helper rewrites it as a detailed generation prompt
Prompt variation — Generate alternative versions of an existing prompt
Batch prompts — Create multiple prompts at once for varied generation runs
Hashtag system — Tag style keywords with # to lock them across iterative generations
Create Image mode — Trigger a generation directly from within the chat
Chat history — Stores up to 8 recent chats for reference
Using Reference Images to Modify or Combine
Modify a Reference Image
Upload a reference image and instruct the model how to change it:
"Make this low-poly."
"Simplify for 3D printing."
"Change the material to gold with cracks in it."
"Remove background and change to a front-facing angle."
Tip: You can also send a previously generated AI image as a reference by selecting the three-dot menu below it.
Combine Multiple Reference Images
Upload multiple images and guide how they merge:
"Use the shape from Image 1 and the style from Image 2."
"Combine this armor design with that pose."
"Blend these into one cohesive prop."
This enables controlled concept merging and design iteration across multiple sources.
If you find interesting workflows or tips for the AI Image feature, share them in the Meshy Community Discord!