THE DARKROOM NOTES / 03
Remember the frame.
Find the file.
A client remembers the red helmet. You remember the shot. The camera called it IMG_4827. Chimp adds descriptions and tags that help you find it again, then lets you add the names, event details and client language that make the catalog yours.
What comes back for each photo
Every completed review includes a short factual description and a set of visible-content keywords. Depending on the frame, the tags may cover the subject, activity, equipment, setting, and useful background details: cyclist, cornering, red helmet, road bicycle, crowd, or wooded background, for example.
The description gives those details context in one sentence. That is useful when six frames share similar tags but only one shows the exchange, reaction, or arrangement you remember. Chimp is describing visible content, not writing a caption for publication. Both the description and tags can be edited in the companion, so your language becomes the durable record.
Search the local shoot you are working on
Type a filename, keyword, phrase, or detail from a review into the companion search. It checks the current local description and tags, subject and focus notes, quality reasons, blur type, and path. Exact metadata matches rank first. A small, concrete synonym list can bridge ordinary wording differences such as bike, bicycle, and cyclist. Put quotation marks around a phrase when the words belong together.
Each result says which saved field matched and whether a related term helped. This remains metadata search on your computer; it does not inspect the pixels again, recognize a person, or infer an object the review omitted.
That search is especially useful after a long event. Try the visible thing you remember: bouquet, podium, yellow jersey, telephoto lens, dog, or rain. Combine the result with the focus filters: find the ceremony frames, then inspect the sharp group; find the crash sequence, then check yellow and red before deciding what survives.
Use the listing on the computer that holds the files
The result is tied to the original relative path, not just the basename. That keeps Card_A/DSC_0042.NEF separate from Card_B/DSC_0042.NEF and gives the companion an exact local match. Use the listing to propose sorting, moving, or deletion on the Mac or Windows computer where the originals live. Before anything changes, a dry run shows every affected file for approval.
| Export | Best for | What it carries |
|---|---|---|
| XMP sidecars | Bringing focus colors and keywords into a compatible photo editor | Sidecars in preserved folder paths, hierarchical Chimp keywords, focus assessment fields, and supplied unrelated XMP metadata |
| CSV | A quick spreadsheet, production handoff, or simple local script | One row per completed photo with path, focus, blur type, description, keywords, review notes, Perfect Shot details and source hashes |
| JSONL | The complete local record and reliable automation | One structured record per completed photo, first and final assessments, human edits, hashes, source paths and Top Quality status |
Sidecar XML files are exported separately under xmp/. The accompanying manifest.json maps each sidecar to its source photo.
If you supplied XMP sidecars, existing ratings and unrelated fields are preserved. Chimp does not invent a star rating. The standard XMP profile includes Chimp’s flat and hierarchical keyword entries. The Capture One profile removes Chimp-generated flat dc:subject entries because they can import poorly there; hierarchical keywords, custom assessment fields, JSONL, and CSV still retain the information. See the Capture One guide before a large merge.
A cleaner handoff to an editor
An editor can receive the photographs in their original folders plus a result file that already distinguishes likely focus misses and describes visible content. CSV works when the handoff needs to be readable at a glance. JSONL is the better source when a script or companion will create folders, build selects, or reconcile duplicate filenames. XMP is the route when the metadata should appear in a compatible catalog.
Because the folder path and source hash travel with the result, the listing is more dependable than a loose document of filenames. It is still wise to test XMP on a small batch before merging and to inspect any move or delete dry run before approving it.
Useful now, portable later
A description and a handful of good tags can turn an old delivery from “somewhere in the 2026 folders” into a much faster lookup. The companion stores the photos, previews, descriptions, tags, assessments, and exports locally; the online service does not keep a second library. Keep the CSV or JSONL beside the shoot and include that folder in your normal backup plan.
If you need a controlled vocabulary, client names, named locations, or a finished caption, edit the generated text locally or add those fields in your catalog. Chimp’s job is to give you a useful, factual first layer without pretending it knows more than the pixels show.