
Editing a Cloudinary named transformation does not automatically regenerate images already derived with its earlier definition. Old and newly generated renditions can therefore differ even though their URLs use the same preset name. Treat the preset change as a rollout that includes existing derivatives, rather than replacing source photographs to force an update.
Key Takeaways
- A named definition and previously generated outputs are separate states.
- Existing derivatives are not automatically invalidated after an edit.
- Deleting a preset can break dependent deliveries.
Identify what changed
Cloudinary's named-transformation guide explicitly states that updating a named transformation does not automatically invalidate and regenerate existing derived assets. It documents invalidation or another deliberate delivery-parameter change to trigger re-derivation.
Keep the source file unchanged while checking this mechanism. If the master and preset both change during diagnosis, a visible difference cannot be assigned confidently to either one.
Prepare a small rollout matrix
- Record the current named definition and its production uses.
- Select one existing derivative and one source that has not yet generated that rendition.
- Write the expected new crop, size or visual effect.
- Have the authorized owner prepare the revised definition and supported regeneration plan.
- Request and inspect both representative outputs after the planned update.
- Check that the application still emits the intended transformation name and complete chain.
- Record the definition revision alongside the approved output.
Cloudinary documents the regen_derived CLI capability, but this guide does not execute it or prescribe a command without reviewing the intended assets and account.
A hypothetical portfolio crop update
An imagined portfolio changes its square portrait preset to preserve more headroom. Recently requested portraits use the new definition while previously generated cards remain tightly cropped.
The proposed diagnostic compares the preset definition, source identity and actual delivered files. It does not conclude that the portrait generator ignored the instruction or that the master needs another edit.
The generic cache-stage guide helps distinguish a direct asset response from a later social preview. This problem begins with a changed processing definition and existing derivative dependencies.
Do not use deletion as a refresh button
The same Cloudinary page warns that deleting a named transformation can invalidate dependent derived assets and cause 404 responses. It also documents dependency-count conditions under which deletion fails. Inspect usage before deciding whether the preset is genuinely retired.
Acceptance and boundaries
Accept the rollout when representative old and new sources both produce the approved rendition and their active references remain valid. Keep any unresolved variants out of a claim that the entire portfolio was refreshed.
No regeneration, cache invalidation or account test is claimed. A working updated sample establishes only that sample; broader rollout needs its own inventory and observations.
Related tasks
- Why Does a Shared Link Still Show My Old Image After I Updated the Page?
- How Do I Keep the Approved AI Image or Video from Getting Lost Among Revisions?
- Why Does Google Still Show My Old Product Image After I Upload a New One?
Sources and scope
- Cloudinary: Named transformations: Updating a named definition does not automatically invalidate/regenerate existing derived assets; documented regeneration options and deletion/dependency consequences.
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.