Image to 3D infers shape and texture from a reference image and builds a 3D model from it. Your own drawings work just as well as images you generated in PicoBerry.
Generate from an image
- Choose Model in the tool bar on the left. The panel opens on the Image to 3D tab.
- Pick an AI model — the list matches Text to 3D, plus the image-only Meshy 7.
- Choose an input mode (Single / Batch) and add your image(s) (see below).
- Set options such as texture, texture resolution, target polygon, and license. They work the same as in Text to 3D.
- Press Generate.
Ways to add an image
- Click the image box — the Select Media window lets you upload a new image (Your Uploads) or pick one from Your Generations or Collections. PNG, JPG, or WebP, up to 20 MB.
- Drag & drop or paste — drop a file on the image box, or paste with ⌘V / Ctrl+V.
- Drop it on the 3D viewer — with the Model tool open, drag or paste an image onto the 3D viewer in the middle and it becomes the input for Image to 3D (Single). Generation starts only when you press Generate.
- Send it from the image feed — click the Image to 3D icon on any generated image to bring it into this panel.
The image box has a built-in Remove Background button, so you can cut out the background right there (it runs in your browser, no credits). Images larger than a model accepts are resized automatically before generating.
Input modes
Single — one image
The fastest, simplest route: one front or three-quarter view becomes a 3D model.
Multi-view — four directions
Under the image box in Single mode, turn on the Multi-view switch and Left · Back · Right slots appear. Adding views of the same subject from other directions lets the model see the sides it couldn't, so the shape comes out more accurate.
- Multi-view needs the front image and all three of Left, Back, and Right. If any of them is missing, only the front image is used.
- Pressing Send to Multi-view 3D → 3D Mesh on a 4-View image set fills all four slots at once.
- The side slots are cleared after a multi-view generation.
Models that support multi-view:
- Supported — Tripo P1 · Tripo 3.1 · Tripo 3.1 Ultra · Tripo 3.0 · Meshy 6 · Meshy 7 · Hunyuan 3.1 · Hunyuan 3.1 Lite
- Not supported — Tripo P2 · Meshy T2 · Meshy 6 - Low Poly
Pick a model that doesn't support it and the switch locks with "The selected model does not support multi-view." If you'd already added side images, a notice tells you only the front image will be used.
Batch — many images, one model each
The Batch tab takes several images at once and turns each one into its own 3D model. It's handy when you want the same model and options applied to a whole set.
- How many images you can add at once follows your plan's 3D concurrent-generation limit (one on the Free plan).
- The Generate button shows the total as "credits per image × number of images".
- Pressing Generate all as 3D in Parts Breakdown opens this Batch tab already filled with the split part images.
Ultra Mode (Meshy 7)
Choose Meshy 7 and an Ultra Mode switch appears. It adds a refinement pass for crisper geometry and finer surface detail, at the cost of a little fidelity to your input image, a longer wait, and extra credits.
- Ultra Mode works only with a single image; turning on Multi-view switches it off.
Getting the best results
- Put one subject, centered, on a plain background. If the background is busy, use Remove Background first.
- Front or three-quarter views work best; extreme angles make reconstruction harder.
- Clear contrast and even lighting reconstruct more reliably than dark or cluttered images.
- For multi-view, keep the subject the same size and pose in all four directions.
After generating
Just like Text to 3D, you can carry on to Remesh & Retexture, UV Unwrap, Animation, or Export. Press Reuse prompt in the asset info panel to put the source image (all four for multi-view), model, and options back into the panel.