Batch resizing: naming, sampling, and rerunnable failures
`batch-resize-photos` powers catalog refreshes and DAM syncs: normalizing longest-edge beats forcing one WxH on mixed orientations. Pipelines must honor EXIF orientation, surface corrupt inputs, and log skips—silent failures leave mystery holes on the site.
Batch resize playbook
- Define longest-edge or boxed resize, output format, and ICC handling inside `batch-resize-photos`.
- Sample landscape, portrait, and square sets for accidental stretch or unintended crops.
- Emit a CSV map from source to destination and reconcile file counts before upload; rerun only failed rows.
Batch resize Q&A
Ten thousand files include a few corrupt JPEGs—should the job halt?
Default to log-and-continue with a final error report; switch to fail-fast for regulated launches if needed.
Colors drift versus the single-file test?
Check whether one batch converts to sRGB while another preserves embedded profiles, or auto-rotate toggles differ.
How do you avoid overwriting masters?
Write into versioned folders or suffixed outputs, keep sources read-only, and diff counts plus total bytes before release.
Before publishing `batch-resize-photos` assets externally, which compliance checks are mandatory beyond visual quality?
Start with "prepare rollback versions", "enforce pre-release QA gates", and "document post-release reviews", then explicitly verify "color profile mismatch" and "upload rejection by size policy" before release approval.
Under deadline pressure, how should teams balance speed and stability in `batch-resize-photos` processing?
Start with "lock dimension tiers first", "normalize naming conventions", and "sample on real destinations", then explicitly verify "batch naming collisions" and "alpha transition artifacts" before release approval.