Make a consistent set of product photos
Use one saved recipe for canvas size, background, compression and filenames, while checking each product before exporting the batch.
Try the workflow
Download a transparent product illustration ↓Examples describe a practice workflow. Your file size, recognition accuracy and processing time depend on the source and your device.
Choose one canvas for the catalog
A square canvas is useful when product cards use a fixed shape. Try 1600 × 1600 pixels for this exercise, then adjust to your store's requirements. Keep whole image fits the entire source inside the canvas and adds margins; Fill and center crop fills the canvas and can remove part of a wide or tall product.
Select the images that belong in the batch. A recipe uses the current selected images, so check the selection count before running it. Test one representative product first, especially a tall bottle or wide box, before applying the same settings to a large set.
Understand the background choice
A background color fills transparent pixels and the new canvas margins. It does not erase an existing opaque wall behind a product. For the supplied transparent illustration, a white background is enough. For a photograph, enable subject isolation only if you want to remove its original background.
Automatic isolation can struggle with fine hair, glass and similar foreground/background colors. The first use downloads a model of approximately 183 MB. Inspect the edges and use the dedicated background tool for manual refinement before producing your final catalog files.
Name, export and verify
Set a meaningful filename prefix, a maximum JPG quality and a target file size. The recipe assigns distinct numbered names and offers a ZIP with the completed results. Saving a recipe stores its settings in this browser; it does not save the original image files.
Review dimensions, file sizes and above-target warnings for each output. Use Undo on an individual image if the composition changed unexpectedly. A batch can contain mixed successes and errors, so compare the result count with the selected count and retain your originals until the store displays the uploads correctly.