VectorPaint photo-frame variants: mask one image for landscape, square and portrait

English translation of the Chinese original, translated on 2026-10-26. Original publication dates, attribution, code and version notes are retained.

The same image needs different compositions in different frames. Make landscape, square and portrait versions from an image you own or may use. Focus on masking and layout; a new aspect ratio does not invent content outside the source image.

Inspect source dimensions and crop options

Choose an adequately sized image with room around the subject and retain the original. List essential areas such as a head, product edge or hand. Prepare 1200 by 600, 800 by 800 and 600 by 900 pixel exercise canvases with clear names. Actual platforms may require different sizes; do not start from a small chat thumbnail.

Compose the landscape version first

Preserve the image ratio and place the subject to one side with room for text. Create the rectangular mask after checking selection and stacking. Move or scale the source while inspecting cropped edges instead of stretching it. Landscape compositions often lose important top/bottom content; check that before adding a title.

Recompose the square independently

Start the square from the full image or a copy retaining it, not an already cropped landscape preview. Reassess subject centre, title and margins. Leave looking space for a person or retain essential product edges. Compare visual prominence across formats rather than forcing identical numerical bounding widths.

Prioritise vertical edges and title space

Portrait crops remove more side content. If a wide group image cannot fit, use whitespace or separate image/text regions instead of cutting people awkwardly. Keep titles clear of faces and product identification. Background panels are exported objects. Masks do not automatically identify subjects or extend backgrounds; inspect the entire canvas after each adjustment.

Preserve proportions and assess raster quality

Inspect detail in all versions and distinguish cropping, scaling and source blur. A mask does not convert raster pixels into infinitely scalable paths. For excessive enlargement, use a larger source or allocate less image area. More output pixels do not restore missing detail. Assess normal display size and reopen delivery files to confirm embedded imagery remains present.

Audit wording and output across all variants

Check title, date, credit and image against one content list across all three versions. Save independent editable files and name exports with ratio, dimensions and revision. Transparency does not remove a photograph’s background. Open every output and inspect actual platform placement because one successful landscape version does not validate the others.

Recheck each crop against the subject list

Open landscape, square and portrait exports and audit only the original must-preserve subject list first. Choose whitespace or a new composition when a crop cannot retain essential areas. Then reverse the order and inspect wording independently. Separate image and text checks to avoid overlooking edges in an attractive layout, and start future ratios from the full source.

Diagnose missing content after mask edits

Identify whether selection targets the masked group, mask path or image; release a copy to inspect if needed. Avoid repeated whole-group scaling. Keep decorative borders separate from image transforms. For missing exports, inspect the reopened original and actual delivery file rather than thumbnails. Retain the full source and all editable variants.