Searching by hand, and searching by sight
How long the same image takes to find, each way
Manual search requires recalling a filename or a folder; content search only needs what the picture looked like. The larger and messier the library, the wider the gap. For small, well-named libraries, doing it by hand really is better.
By hand is enough when
- The library is a few hundred images you can scan in one screen.
- Filenames and folders are consistent enough to tell you what is inside.
- You only need to find something once or twice a week.
By hand stops working when
- The library runs to tens of thousands and keeps growing.
- Much of it is camera exports, chat downloads, and client originals with meaningless names.
- A client sends an image asking whether you have a match, and you have to answer now.
- You need to pull one person entire set out of a full event shoot for delivery.
- Years of duplicates and near-duplicates need clearing out.
Kinds of clue you can use
| Capability | Assist Local Image Search | Searching by hand |
|---|---|---|
| The filename or folder | Yes | YesBy hand is fast in this case too |
| What the picture looked like | YesSimilar image or a sentence | PartialOpen them one by one |
| Roughly what color it was | YesFilter by color | PartialOpen them one by one |
| Who was in it | YesFace search with clustering | PartialRecognize them one by one |
| What was printed on it | YesOCR search | PartialOpen every image to read it |
| Only a fragment, such as a pattern | YesSelect a region and find the original | PartialCompare from memory |
| Images spread across several drives | YesIndex once, search across all | PartialRecall which drive first |
Everyday tasks
| Capability | Assist Local Image Search | Searching by hand |
|---|---|---|
| A client sends an image asking for a match | YesRanked results in seconds | PartialDig from memory and experience |
| Pull every photo of one person | YesCluster, then export at once | PartialIdentify and copy by hand |
| Clear duplicates and near-duplicates | YesDetect in bulk, then handle together | PartialCompare by eye, one by one |
| Turn a selection into a quotation list | YesExport images, filenames, or a table | PartialCopy and paste by hand |
Cost and setup
| Capability | Assist Local Image Search | Searching by hand |
|---|---|---|
| Upfront investment they win | PartialFirst indexing takes hours for tens of thousands of images | YesNothing to set up, available immediately |
| Money they win | Free for 7 days, then paid by usage time | Costs no money, costs time |
| When the library is only a few hundred images they win | PartialInstalling and indexing is slower than just looking | YesOne screen is enough; by hand wins |
| Machine resources they win | PartialRuns local AI models, so install size and memory use are notable | YesUses nothing at all |
You do not have to organize first
Many people stall on "I should tidy the folders first". You do not. It changes neither your directory structure nor your naming.
- 01Point it at the folders you haveMessy is fine. External drives and network shares can be added too.
- 02Let it indexThe first pass scales with library size and resumes if interrupted.
- 03Keep saving files the way you always didNew images are picked up incrementally; your workflow does not change.
FAQ
Do I have to tidy my folders first?
No. It changes neither your directory structure nor your naming, so point it at things as they are. Skipping that tidying step is the point.
How long does indexing take?
It depends on library size and machine specs; tens of thousands of images usually means hours. Try one folder first and extrapolate. It resumes if interrupted.
Do I have to reindex when I add images?
No, new images are picked up incrementally without a full rebuild.
Are my images uploaded?
No. Indexing and search run on your machine. The standard edition only goes online to validate licence time, and a fully offline edition is available.
Test it on your own library
Free for 7 days. Add a folder and search your own images before deciding.