
Different darktable export presets can still write to the same destination. When using multi-preset export, the manual explicitly requires a unique file path for each preset so that outputs do not overwrite one another. Naming presets “web” and “review” does not make their filename templates different.
Inspect destinations as complete paths
Open the current export documentation and inspect the filename template stored in every selected preset. Compare the resulting folder, basename and extension. A different size or quality value can still produce the same path when the output format is shared.
Read the on conflict choice as part of this inspection. The documented options include creating a unique filename, overwriting, overwriting if changed and skipping an existing destination. These choices change what happens at a collision; they do not identify which preset’s output a recipient was meant to receive.
The recorded batch-filename guide illustrates another source of repeated paths, a fixed action filename. Here the cause is several valid preset branches sharing one destination namespace, so correcting an unrelated recorded action cannot solve it.
Draw the source-by-preset matrix
Consider a hypothetical delivery with three product views and two JPEG renditions: a store preview and a larger client review. The expected set contains six source-and-preset combinations. File count alone is insufficient if a uniquely renamed collision hides which specification produced each file.
| Source | Preview branch | Review branch |
|---|---|---|
| Front | Preview folder, front filename | Review folder, front filename |
| Back | Preview folder, back filename | Review folder, back filename |
| Detail | Preview folder, detail filename | Review folder, detail filename |
Use separate destination folders or another explicit, documented path distinction for each preset. Keep source identity recognizable within each branch. Avoid depending on an automatically appended integer as the only indicator of output purpose.
Run a small, isolated export first
Prepare a fresh test destination, preserving all approved deliverables. Select one source and the two intended presets. Expand multi-preset export, choose those presets and start the small run. Reopen both expected outputs and check subject identity, pixel dimensions and branch location against the matrix.
Only then test all three sources. Reconcile each matrix cell with an actual file. Inspect the image content as well as the filename; several correctly named files can still contain an unintended source or rendition.
The approved-asset handoff guide helps keep this candidate run separate from the accepted delivery. These are proposed acceptance checks, not a reported export benchmark or a claim that six particular files have been generated.
Diagnose the result before retrying
Missing output can reflect a collision handled by Skip. Unexpected extra suffixes can reflect unique-name conflict handling. A replaced file can reflect an overwrite policy. Read the actual stored preset settings rather than guessing from the last output’s appearance.
The manual notes that overwrite confirmation is once per job, not once per file. Keep protections enabled and do not treat one confirmation as an itemized review. Preserve the editable master separately, correct the conflicting paths and rerun into an isolated destination. Accept the final set when every required source-and-preset pair has one identifiable output with the right specification.
Related tasks
- Why Does My Photoshop Batch Action Save Files Under the Same Name?
- How Do I Keep the Approved AI Image or Video from Getting Lost Among Revisions?
- What Is the Difference Between an Editable Artwork Master and a Flat PNG?
Sources and scope
- darktable user manual: export: Multi-preset export needs unique paths per preset. Conflict choices determine collision handling; overwrite confirmation is per job.
Based on the current online darktable manual reviewed October 31, 2026. Check the installed release. Examples and acceptance checks are proposed, not hands-on test results.
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.