
Marking a crop in Cloudinary's Upload Widget records a region; the delivery workflow must actually use that region. The documentation distinguishes selected coordinates, delivery gravity and an incoming crop that changes stored pixels. Decide which contract your client-photo form promises before calling the selection complete.
Key Takeaways
- A visible crop rectangle can represent coordinate data rather than discarded pixels.
- Interactive cropping requires a supported single-file route.
- The delivered proof must be compared with the client's selected framing.
Describe what the crop should control
Cloudinary's Upload Widget reference says interactive cropping sends selected dimensions as customCoordinates or faceCoordinates, depending on the coordinate mode. Delivery uses the appropriate custom or faces gravity to obtain the selected crop.
The reference separately describes incoming cropping in the upload preset when the intention is to crop before storage. Those are different stages. Preserving the full source plus a selected delivery crop is not the same as storing only the cropped pixels.
Trace the client's choice through the pipeline
- Retain the authorized source and a private record of the intended framing.
- Confirm the intake uses single-file mode when interactive cropping is required.
- Inspect which coordinate mode the integration submits.
- Have the implementation owner verify that the intended delivery transformation consumes that region.
- Check whether any incoming transformation changes the stored source.
- Retrieve the proof through the real delivery route.
- Compare subject placement and edge content with the accepted selection.
A screenshot of the widget selection can support the human review record. It does not by itself prove that the server stored or used the same region.
Handle routes where cropping is unavailable
The widget documentation says the interactive step is skipped for video and browser-unsupported image formats such as PSD or SVG. It also notes restrictions on preview/crop options for large local files. Do not design a form that silently treats a skipped crop as client approval.
Provide a documented alternate preparation or review route for those inputs. Keep a missing selection separate from an upload error or a deliberately full-frame submission.
A hypothetical portrait-card intake
A client selects a face-safe square from a wide source. The upload succeeds, but the website still shows the full-width image.
The proposed diagnostic checks whether the selection became coordinate data and whether the public transformation used it. Regenerating the face cannot repair a delivery pipeline that ignores the selected region.
The preview-coordinate guide addresses mapping boxes into original-image coordinates. This task concerns carrying an intake crop into its intended stored or delivered representation.
Acceptance and limits
Accept when the actual proof matches the approved framing and the source-retention policy is explicit. Inspect the stored master separately if preserving uncropped content matters. No widget submission, crop transformation or product test is claimed here.
Related tasks
- Why Does My Image Edit Land in the Wrong Place When I Copy Preview Coordinates?
- How Do I Crop a Set of Portraits Consistently When Everyone Is a Different Height?
- Why Can I Recover Image Content Outside a Photoshop Crop?
Sources and scope
- Cloudinary: Upload Widget reference: Source-specific local format/byte/dimension checks, separate server Usage Limits, reject-versus-scale behavior, crop coordinate/delivery contract and eligibility, plus camera HTTPS/browser/mobile conditions.
Documentation-based editorial synthesis with AI-assisted research and drafting. Scenarios, ledgers and acceptance checks are hypothetical or proposed. No hands-on test, account change, measured demand, AI-citation rate, benchmark or client acceptance is claimed. Availability and plan limits remain within each cited source's scope.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance with proposed examples, not a claim of hands-on testing or an independent product review. Report a correction.
Cover: AI-generated conceptual editorial illustration; not a real application screenshot, observed output or before-and-after test.