This is a lightweight browser-based layer editor for quick multi-image jobs: composites, simple banners, watermark stacking, and text labels on top of photos. Every operation runs on your own computer - open images are read into canvas layers in memory, never uploaded. Use Open image to drop a photo onto the canvas as a new layer, drag it with the Move tool, add a text layer with the Text tool, tune colors with the filters panel, then Export a PNG or JPG.
Each open image becomes a layer. Use the layer list to select, rename mentally by the order it was added, duplicate, delete, or move up/down in the stack. Opacity and blend mode (multiply, screen, overlay, soft light, difference, color, luminosity) apply to the active layer. Undo keeps up to 20 steps as snapshots, so you can freely experiment.
Pick the Text tool and click anywhere on the canvas to type a label; Enter commits, Esc cancels. The text panel lets you switch font, size, color, bold/italic and alignment, and double-clicking an existing text layer reopens it. Filters (brightness, contrast, saturate, hue-rotate, blur, sharpen, grayscale, sepia) preview live on the active layer; click Apply to bake them in, which also pushes a history snapshot.
Because everything is local, the tool works offline after load and never sends your photos anywhere. It is intentionally simple: no brush, eraser, selection masks or automatic object removal - for those, use the dedicated smart background remover or the business portrait maker. If you need a full photo retouch pipeline, the desktop-grade editor has more tools.
No. Open image reads the file into an in-memory canvas layer in your browser; nothing is sent to a server and the file handle is released immediately.
The history keeps up to 20 snapshots. Each layer change, filter apply, text commit or move pushes one snapshot.
PNG (with transparency) or JPG (flattened on a white background) from the top bar.
Filters bake onto raster (image) layers. If the active layer is a text layer, switch to an image layer in the list first.
The workbench is a lightweight browser-based layer editor for quick multi-image jobs: stacking a watermark on a photo, composing a simple banner, putting text labels on a collage, or combining a product shot with a background. You open one or more images as layers, drag them around with the Move tool, add a text layer with the Text tool, tune colours with the filters panel, and export a PNG or JPG. Every operation runs on your own computer - images are read into in-memory canvas layers and are never uploaded to a server.
It is deliberately small: no brush, no eraser, no selection masks, no automatic object removal. For those you would use the dedicated smart background remover or the business portrait maker. What the workbench is good at is the everyday layering job that a full editor over-kills: stack two photos, drop a caption on top, tweak the brightness, export. For the heavier desktop-grade editor, the photoshop-core page has the deeper tool set.
Each image you open becomes a layer stacked on the canvas. The layer list on the right shows them top to bottom; the topmost layer is what you see in front. Selecting a layer makes it the active one, and the opacity slider and blend mode dropdown then apply only to that layer. Blend modes matter most when you are compositing: Multiply darkens and removes white (so a black logo on a white sheet prints onto a photo), Screen lightens and removes black (so a glow or light leak overlays cleanly), Overlay and Soft Light contrast-match, and Color / Luminosity let you tint one image with the colours of another.
Opacity simply fades the layer; at 50% the layer is half transparent and the layers beneath show through. Duplicate layer is handy when you want to stack the same photo twice - once normal for detail, once on Multiply at low opacity to deepen shadows. The Move tool drags the active layer around; the history snapshot records the new position so you can undo it.
Pick the Text tool (or press T) and click anywhere on the canvas; a small inline editor appears. Type your caption and press Enter to commit, or Esc to cancel. The first commit creates a new text layer; double-clicking an existing text layer reopens the editor at that position. The text panel lets you switch between four fonts (system UI, Arial, Georgia, Courier), set size from 12 to 200 pixels, change colour, toggle bold and italic, and align left, center or right. Because text is a live layer, you can keep editing it after creation - the canvas repaints the text at the new size and position.
The eyedropper tool (press I) samples the colour under your cursor and sets it as the foreground colour, which is also the default colour for new text. That makes it easy to match a caption to a colour already in the photo: pick from the photo, click with the text tool, and the caption is instantly on-palette. For persistent watermarks that should sit above many photos, set the text layer to low opacity and Multiply or Overlay.
The filters panel on the right exposes brightness, contrast, saturate, hue-rotate, blur, sharpen, grayscale and sepia. Moving a slider previews the result live on the active layer - you see the canvas update without committing anything. When you press Apply, the tool bakes the current filter stack into a raster copy of the layer and pushes a history snapshot, so you can still undo it. Reset clears the sliders without baking.
Filters only bake onto raster (image) layers; if the active layer is a text layer, Apply prompts you to switch to an image layer first. Sharpen is a small convolution kernel that emphasises edges; use it lightly on a slightly soft photo. Blur is a Gaussian radius in pixels, useful for defocusing a background layer. Grayscale and sepia are one-click mood conversions that stack with the other sliders. For a global, whole-image enhancement pass instead, the enhance tool is the dedicated pipeline.
The history panel on the left keeps up to 20 snapshots. Every layer add, duplicate, delete, move, opacity change, blend change, text commit and filter apply pushes one snapshot. Ctrl+Z steps back, Ctrl+Shift+Z (or Ctrl+Y) steps forward, and clicking any history item jumps to that state directly. Ctrl+S exports the current document as if you pressed the Export button. The V, T and I keys switch between Move, Text and Eyedropper tools without reaching for the mouse.
Snapshots are stored as PNG blobs in memory, which is why the cap is 20: beyond that the memory use starts to matter. If you find yourself needing more history than 20, you are probably doing a job that belongs in the photoshop-core editor, which has a deeper history and more tools. For a one-off export after compositing, the convert tool changes the final file format.
Because every operation is local canvas work, the workbench works offline after the page loads and never sends your photos anywhere. Open image reads the file into memory via an object URL, which is released immediately after the frame is drawn; nothing is logged, nothing is uploaded, and closing the tab discards everything.
What it does not do is equally important. There is no brush, no eraser, no selection, no automatic object removal, no generative fill, no server-side anything. If you need to remove a person or object from the background, use the smart background remover. If you need to retouch a portrait, use the business portrait maker. If you need a full photo editor, use the photoshop-core. The workbench is the middle layer: simple compositing, simple text, simple filters, exported in a click.
No. Open image reads the file into an in-memory canvas layer in your browser tab. The object URL is released immediately after the frame is drawn, and nothing is transmitted. After the page loads you can go offline and keep working.
The history keeps up to 20 snapshots. Each layer change, filter apply, text commit, move, opacity or blend change pushes one snapshot. Ctrl+Z steps back, Ctrl+Shift+Z steps forward.
PNG, which keeps transparency, or JPG, which flattens onto a white background. Pick from the dropdown in the top bar and press Export (or Ctrl+S).
Filters bake onto raster (image) layers. If the active layer is text, switch to an image layer in the list and press Apply there. You can still change the text style from the text panel.
No. The workbench intentionally ships Move, Text and Eyedropper only, plus layer compositing and filters. For brush, eraser and selection tools, use the photoshop-core editor; for object removal use the smart background remover.
Yes. After the page loads, every operation - opening images, layering, text, filters, export - runs locally on your device. Disconnect from the network and the workbench keeps working.
A new document starts at 800 by 600 pixels. Opening an image when the canvas is blank sizes the document to that image; opening more images stacks them as layers on the current document size.