Tool PantryFree browser tools for everyday files
← Back to all browser tools
MOSS-IMAGE / LAYOUT EDITOR

Draw a MOSS-Image layout.
Copy the JSON prompt.

Place a heading, a subject, and the small print. Draw boxes, enter the exact words, and export the layout for your MOSS-Image workflow.

Layout planning only. The canvas previews your boxes and text; generate the image separately in a MOSS-Image workflow that accepts layout prompts. No account or API key needed here.

Your composition

Layout preview, not a generated image. Drag empty space to draw; drag a box to move it. Arrow keys move a focused box by 1 pixel, or 10 with Shift.

Generate at 768 × 1024 pixels. Dimensions are separate from the model prompt.

Change the width or height, then choose Apply size. Boxes scale with the canvas. Use multiples of 16; 8192 px is this editor’s limit, not a tested model limit.

Elements

0 elements

No elements. Add a box to enter its position without drawing.

Your model prompt

Centered MOSS-Image layout JSON

Paste the model prompt into your generation workflow and set the same width and height there. The editable draft includes dimensions and top-left pixel boxes; reopen it here to continue editing.

From boxes to a MOSS-Image prompt

  1. Set the output dimensions. The example uses 768 × 1024. Use the same size when generating.
  2. Place the elements. Draw a box or choose Add box, then type its pixel edges. Overlap is allowed.
  3. Separate wording from appearance. Text elements have an exact-text field and an appearance description. The scene field describes the whole composition.
  4. Export and keep a draft. Copy the model prompt for generation. Save the editable draft to retain dimensions and reopen it later.

Why the exported boxes contain negative numbers

MOSS-Image’s documented experimental layout format uses abs_centralized coordinates, ordered [top, left, bottom, right]. The origin is the center of the requested image. The editor uses top-left pixel edges for editing, then converts them for export.

For a 768 × 1024 canvas, left 64, top 80, right 704, bottom 224 becomes [-432, -320, -288, 320]. Width and height are different, so each axis has its own offset.

See the poster walkthrough with downloadable input and output for a checked example. Export is a single-line JSON prompt; the file labelled editable draft is for restoring your work, not for pasting directly into generation.

What this editor checks

It checks positive box area, whole-pixel edges inside the canvas, required descriptions and text, and canvas dimensions. It does not run MOSS-Image or predict its output. The layout feature is experimental; an image may place objects differently or render words incorrectly.

Independent helper. Format checked 17 October 2026 against the official layout guide and official prompt compiler. The official enhancer is another option if you prefer an agent and a local script.

Keeping your work

The layout stays in this tab and is not saved between visits. This editor does not send your draft to a model service. Download it before leaving, then use Open draft to restore it. For the different FLUX layout format, use the FLUX 3 bounding box generator.