Elysia Tools
Navigation
Workflow Playbook
Turn a background-removed image into a checked target-size canvas by choosing a suitable edge treatment and reviewing the boundary before delivery.
Hubs
Work from a copy of the original. Run image-background-remover only when the source still has a background, then inspect the transparent PNG at the subject boundary. Hair, glass, thin straps, and low-contrast edges need a human check because an extension step cannot restore pixels that were removed incorrectly.
Record the final width and height, then calculate which sides need space. Use image-extend to establish that canvas and keep the transparent checkpoint beside it. A square product tile, a wide banner, and a portrait card can need very different left and right padding even when the subject is the same.
Use image-extend-blur-background when a soft photographic context is acceptable. Use image-extend-gradient when the added field should follow specified colors. Use image-extend-mirror when reflected edge texture is less distracting than a flat fill. Treat these as separate candidates, not a mandatory chain.
Check every join and corner at 100%, then view the image at its intended display size. Confirm the subject is not clipped, the added area has no obvious seam or repetition, and the file dimensions and alpha behavior match the delivery brief. If none of the candidates passes, retain the cutout and revise the canvas or background decision instead of presenting an unchecked result.
Workflow playbook
Run image-background-remover on a working copy and retain the transparent PNG as an intermediate checkpoint. Zoom into hair, thin handles, product corners, and translucent materials; do not extend a mask that already has an obvious halo or missing subject pixels.
Use image-extend with the recorded target dimensions or side-specific padding. Compare the proposed canvas with the subject bounds so the new space creates the intended composition instead of clipping the cutout or adding arbitrary empty margins.
Produce a candidate with exactly one visual strategy: image-extend-blur-background for a soft photographic continuation, image-extend-gradient for deliberate brand colors, or image-extend-mirror for a texture that can tolerate reflected edges. Keep the alternatives separate so the final result has a clear cause.
Compare the chosen candidate with the transparent checkpoint and the source. Check all four joins, corners, subject clearance, dimensions, alpha behavior, and the selected format; keep only the candidate that passes the acceptance checks and record the settings used.
Use image-background-remover when the original still contains a background, then inspect hair, fine product details, and semi-transparent areas. If a cutout already exists, treat it as a checkpoint and do not replace a clean mask without a reason.
Use image-extend to establish the canvas and padding. Choose image-extend-blur-background for a photographic context, image-extend-gradient for a controlled graphic backdrop, or image-extend-mirror when edge continuity matters; these are alternatives, not steps to stack.
Keep transparency when the next layout will supply its own background. When the extension is visible in the final image, inspect the transition at 100% and at the intended display size before accepting it.