
AI image editor
Generate, transform, layer, and arrange images on an object-based infinite canvas.
Explore the image editor →AI creative canvas and image editor
Upload references, generate or transform images with an AI Agent, and arrange every source, variation, layer, and final composition in one visual workspace.
10 free credits every UTC day. No card or subscription required.
A real CubbieAI Studio workflow: select a fragrance reference, ask the Agent for a premium campaign image, follow the generation process, and open the AI-generated result on the canvas.
Built-in Studio example
The included project shows a real CubbieAI workflow: keep the original product reference on the canvas, ask the Agent for a premium campaign direction, review the generation process, and open the finished image beside the source material.
Unlike a one-shot image generator, the reference, Agent activity, generated result, and surrounding composition remain together in one project, ready to inspect or export.
Open the read-only project →The original fragrance image stays available as a reusable project asset.
The Agent turns the request and visible reference into a focused generation plan.
The generated campaign image returns as its own asset without replacing the source.
Start with an empty canvas or your own images. Generate, compare, annotate, and compose without moving every new result into another tool.

Generate, transform, layer, and arrange images on an object-based infinite canvas.
Explore the image editor →
Build connected campaign directions, variations, and channel-ready compositions.
Explore ad creative →
Turn product references into launch imagery, detail views, and styled scenes.
Explore product photos →More ways to use the canvas
CubbieAI is not limited to one prompt-and-result pattern. Each project can combine uploaded references, generated assets, layout objects, notes, and version history in a way that matches the work being delivered.
Begin with a pack shot that already has the correct shape, label, and proportions. Keep it visible on the canvas while exploring settings such as warm fabric, stone, water, or a simple studio surface. Because every generated image returns as a separate asset, the original reference remains available for comparison instead of disappearing into a prompt history.
Place the strongest hero image beside alternate backgrounds and detail directions, then add notes for retouching or later production. The result is a compact launch board that can guide a product page, marketplace listing, or campaign handoff.
Start with the product, campaign message, and any brand references that should remain consistent. Use the Agent to create a focused visual direction, then compare generated scenes without separating them from the brief. Frames, text, arrows, and shapes can mark headline zones, safe areas, or feedback directly around the images.
Once a direction is approved, compose square, portrait, or wide layouts from the same source material. This keeps channel adaptations connected to one master board while still allowing each placement to have its own crop, hierarchy, and supporting copy.
Collect photography references, color cues, typography samples, sketches, and written notes on an open canvas. Arrange them into groups that explain the direction before any image generation begins. When a specific cluster is ready, select only those references and ask the Agent for a new visual that follows their shared mood.
The generated result lands back beside the material that informed it, making the relationship easy to inspect. Continue refining the board, preserve earlier versions, or export the visible composition as a single PNG for review. The canvas remains useful even when the task is primarily organization rather than generation.
Open the existing project instead of rebuilding the brief in a new generation thread. Keep the approved composition visible, create a version checkpoint, and select only the image that needs another background, crop, or visual treatment. New alternatives can sit beside the accepted work, so reviewers see exactly what changed.
Notes, arrows, and frames can identify the requested revision while project history preserves an earlier state. If the new direction is rejected, restore the previous version rather than recreating it. This makes the canvas useful during approval rounds, not only during the first moment of image generation.
CubbieAI keeps references, Agent decisions, generated results, and the surrounding canvas together, so the next change starts with visible context.
Upload source images and place the references, notes, and assets you need on the canvas.
Select reference images, describe the result, and let the Agent prepare the generation.
Arrange generated images with text, shapes, frames, drawings, and editable layers.
Compare versions, restore earlier states, and export the visible composition as PNG.
Canvas editing is free. Image-generation quotes account for the chosen model and output settings.
Resets every day at 00:00 UTC.
Depends on model, resolution, and count.
Use them when your free daily balance runs out.
Clear expectations before you move image creation and visual composition into Studio.
CubbieAI is an AI creative canvas for generating, editing, arranging, and exporting images. It keeps references, layers, AI variations, and project history in one visual workspace.
You can combine uploaded and generated images with text, shapes, drawings, arrows, and frames, then organize the objects with the asset and layer panels.
No. Sign in with Google to receive 10 free credits each UTC day. When you need more, buy a one-time credit pack. Purchased credits do not expire.
Select or upload reference images, describe the change, and let the Agent prepare the generation. The new result is saved as a separate asset and automatically placed on the canvas when possible.
CubbieAI shows the credit cost before generation. Model, resolution, and image count can change the quote. Failed generations restore the reserved credits.
Sign in with Google and start with today’s free credits. No card required.
Open CubbieAI Studio