Image AI Tools

These are one-click AI operations you can run on an existing image node. Select the node and open Node actions — the chevron on its side toolbar — where they sit under Create, Edit and Extract. The same menu holds the prompt presets, and its search box filters both at once.

Upscale Image

Menu path: Select image node → Node actions → Edit → Upscale Image
Model: Topaz High Fidelity V2
Options: 2x, 3x, 4x scaling

Choose your upscale factor. The price depends on how big the result is, which is your image's size times the factor:

  • up to 24 megapixels — 10 credits (most 2x and 3x upscales)
  • up to 48 megapixels — 20 credits
  • up to 96 megapixels — 40 credits

Each factor in the menu shows its price for the image you picked. A factor that would make a result larger than 96 megapixels is greyed out.

The upscaled image is a new image placed beside the original, which is left as it was. It preserves detail using Topaz's high-fidelity algorithm.

Remove Background

Menu path: Select image node → Node actions → Edit → Remove Background
Also on a layer: in the 2D editor, right-click an image layer → Remove Background
Model: Bria
Cost: 10 credits

Removes the background from your image, leaving the subject isolated. Works best with clear subject edges and is safe for commercial use (trained on licensed data).

Run from a layer, it reads that layer's own pixels — not the visible composite — so whatever sits under it is not mistaken for the background. The cut-out arrives as a new layer and the original is left alone. The row is offered only on an image layer that is visible and unlocked.

Extract Color Palette

Menu path: Select image node → Node actions → Create → Color Palette
Cost: 5 credits

Analyzes your image and generates a set of dominant-color swatches. Each swatch appears as a new image node on the board, showing the extracted color palette.

Crop

Menu path: Select image node → Node actions → Edit → Crop
Cost: Free (creates new nodes from selection)

Opens an interactive crop dialog where you can:

  • Draw freely — drag a rectangle anywhere to crop a custom area
  • Use grid presets — split your image into grids (2×2, 1×2, 2×1, 1×3, 3×1, or Custom up to 8 rows by 8 columns) and select which cells to keep
  • Drag handles — resize by edges or corners
  • Move the selection — drag inside the selected area to reposition it
  • Adjust interior splits — drag the grid lines to adjust split positions
  • Right-click a split — delete it

Each selected cell becomes a separate image node on the board.

New Angle

Menu path: Select image node → Node actions → Create → New Angle
Model: Qwen (Multiple Angles)
Cost: shown on the Generate button's tooltip, as the most it can cost. You are charged for the size of the image it makes — 8 credits for a typical one — and the rest comes back to your balance.

Opens the New Angle dialog: the image stands in a 3D view with a small camera pointed at it. Set where the new picture is taken from:

  • Horizontal Angle: where the camera stands around the subject. Click a label under the view (Back, −135°, Left, −45°, Front, 45°, Right, 135°, Back) or drag the camera left and right.
    • 0° = front, 90° = right, 180° = back, 270° = left
  • Vertical Angle: −30° to 90°. Click a label at the left of the view or drag the camera up and down.
    • −30° = low, looking up; 0° = eye level; 90° = straight down
  • Zoom: 0 (wide) to 10 (close-up)

Each number can also be dragged across to change it, or double-clicked to type a value. The reset button returns all three to their starting values. Click the round Generate button to create the new view; it lands beside the original.

View Matrix

Menu path: Select image node → Node actions → Create → View Matrix
Model: Nano Banana 2.1 Edit, at 2K
Cost: shown on the Generate button. It is one generation whatever the grid size.

Draws your subject from a whole grid of cameras in one picture, then lets you keep the tiles you like as images on the board. It works on one image at a time, on the board only.

  1. Pick the grid. Grid offers 2×2, 3×3, 4×4 (the default) and Custom. Custom shows Rows and Columns fields, from 1 to 4 each, for at least two cells. Under the grid, "Canvas" names the shape of the picture the grid asks for. When the cells come out clearly tall or wide, a tip by Generate says which subjects suit them.
  2. Pin views, if you want. Click a cell to choose its view: Auto, Front, Rear, Left side, Right side, the four ¾ views, Top down, Bird's eye, Low angle or Close-up. Cells left on Auto are the model's choice. A pin is a request: the view appears in the grid, but the model may put it in another cell. Choose Auto to unpin a cell. A pin outside a smaller grid is hidden, and comes back when the grid grows again.
  3. Generate. You can close the window while it runs. The run waits as a View Matrix card in the tray at the top of the board; click the card to come back to it. A grid finished before a page reload has no card any more. It is still in your library, where Crop can cut its tiles.
  4. Keep tiles. The grid opens cut into tiles. Click each tile to keep (Select all picks every one), drag a line between tiles to move it if the grid came out uneven, then Add to board. Each tile becomes its own image beside the original, with any thin bar along its edge trimmed off. Adding tiles costs nothing more.

Change views goes back to the setup with your pins kept. Regenerate runs the same setup again for a fresh grid, at the same price again; the grid it replaces stays in your library.

New Perspective

Menu path: Select image node → Node actions → Create → New Perspective
Model: Nano Banana Lite Edit
Cost: 11 credits

Generate one of these predefined camera views in one click:

  • Front View — Straight-on face perspective
  • Rear View — Back at 0° angle
  • Left Side View — 90° profile (left side)
  • Right Side View — 90° profile (right side)
  • 3/4 Front Left — 45° angle from front left
  • 3/4 Front Right — 45° angle from front right
  • 3/4 Rear Left — 45° angle from back left
  • 3/4 Rear Right — 45° angle from back right

Each perspective generates a new angle of the subject while keeping it in its place—only the camera viewpoint changes, so backgrounds reflect what's naturally visible from that angle.

Image-to-3D

Menu path: Select one or more images on the board → chat bar → switch the category to 3D

Converts a 2D image into a 3D model (GLB file). The model lands on the board as a 3D model node and is listed under Home's 3D Models.

Auto-selected models (first available in your plan):

  1. Rodin v2.5
  2. Rodin v2
  3. Trellis 2
  4. Hunyuan 3D Pro

General process:

  1. Switch the chat category tab to 3D (marked in red)
  2. Select your preferred 3D model from the model dropdown
  3. Select the image(s) on the board; the chat bar shows them as numbered or named slots (Front, Back, …)
  4. (Optional) Adjust model-specific settings:
    • Rodin v2.5: Quality tier, mesh detail, PBR materials, HD textures, HighPack add-on
    • Trellis 2: Resolution (512px/1024px/1536px), texture size
    • Hunyuan 3D Pro: Generation type (textured or geometry-only), PBR materials, face count, up to 8 reference views
  5. Click Generate — the 3D model lands on the board as a 3D model node; select it to orbit, frame and light it from its toolbar

Credit cost varies by model and settings (typically 63–168 credits, or up to 336 with Rodin v2.5 Extreme-High tier + HighPack).

Note: All of these use your plan's credit balance. Insufficient credits will prevent generation. Use these operations to enhance, modify, and expand your designs without leaving the board.