VISURAL 1.1.0 · by Sherlock Forensics
Point it at one messy folder. That is the whole pitch.
VISURAL reads every photo and video on your computer and makes them searchable by what is actually in them. Ask in plain language and it finds it: a person, a pet, a place, the text on a sign, even something someone said on camera. All on your own Windows PC. No cloud, no account. Nothing is uploaded unless you choose to share it.
Runs 100% on your PC · Windows 10/11 · One-time purchase, no subscription
New in 1.1.0
New in 1.1.0: from searching to doing
1.0.0 let you find any photo. 1.1.0 lets you do something with what you find: send it, group it and walk back through it by date.
Share any size
Send one photo, a selection or a whole album in one press. Uploads stream straight from disk with no practical size limit, so a heavy album goes in one share where the old build would refuse it. Watch the progress, cancel mid-send, protect it with a password and give it a name so your recipient sees Iceland instead of a code.
Albums
Group photos into albums, set a cover and search inside an album instead of the whole library. Taking a photo out of an album leaves the photo itself untouched.
On this day
Jump straight to any month in your library. Open On this day to see the same date from earlier years. Photos from today are left out, because a photo taken this morning is not yet a memory.
More like this
Search with a photo instead of words, for the things there are no words for. It uses the photo's own fingerprint as the query, so it is the one search that still works while the AI engine is stopped or still loading.
Sharing is free up to 50 files or 1 GB per share. Larger shares use an EasySend key or a Premium upgrade.
Thousands of photos and you still cannot find the one you want
You have thousands of photos and hours of video across drives, folders and old phones. You cannot find the one you want. Filenames and folders do not help. VISURAL reads every photo and video and lets you just ask for what you remember.
Where it comes from
Built from the tools we use to find evidence
Sherlock Forensics spends its days searching suspect systems, cloud accounts and mobile devices for the one detail that matters in an investigation: a face, a place, a phrase, a single photo buried among hundreds of thousands. The hard part was never storage. It was finding the needle. So we built search that works on what is actually inside a photo or video: the content, the faces, the text and the place it was taken.
VISURAL takes that same search engineering out of the lab and puts it on your own PC, as a private consumer app rather than a forensic tool. That is why it holds up on a serious library. Point it at hundreds of thousands of photos and videos spanning decades, on a drive, an external disk or a NAS. Give it a day or two to read and index everything once. After that you can search any moment in plain language, find and group every face, name a person once and see them across the whole collection and jump straight to where a photo was taken on the map.
Private by design
Every part of the AI runs on your machine: captioning, search, face recognition, text reading and transcription. Nothing is uploaded for processing. Nothing leaves your computer unless you choose to share it.
Finds everything
People, pets, places, the text on a sign or receipt and even what someone says on camera. Search "us driving red sports cars in Vegas" and it finds the Ferraris.
Yours forever
One-time purchase, no subscription. Works fully offline after a one-time setup. An expired free trial still lets you browse and search everything already indexed.
What you actually do
Point VISURAL at a folder on your drive, an external disk or a network share. It reads every photo and video and builds a searchable description of each. Then you search the way you remember it: "birthday cake", "the whiteboard from that meeting", "Grandma and the dog at the lake". Search is hybrid, so it matches your words literally and by meaning.
- Finds people: faces are detected and grouped. You name someone once and VISURAL recognises them across your whole library.
- Finds pets and objects the same way, kept separate so your dog never gets confused with a person.
- Reads text inside images (signs, receipts, whiteboards, screenshots) and makes it searchable.
- Transcribes speech in videos and audio, with a read-along view that follows the words.
- Maps your photos by where they were taken.
Video, done properly
Most photo tools treat video as an afterthought. VISURAL actually watches it.
- Describes a frame every couple of seconds to build a timestamped timeline of what happens in a clip.
- Finds a person who appears for three seconds in a two-hour video, with the timestamp.
- Transcribes the audio so you can search what was said.
- Skips near-identical frames and condenses the timeline so it does not read like the same sentence twelve times.
Features worth naming
People view
Everyone VISURAL has found, named once, then recognised everywhere.
Who is this?
Quick naming, one face at a time, yes or no, built for speed.
Recheck
Finds faces filed under the wrong person and shows how confident it is.
Progression
A morph of one person across the years, built from the best photo per period.
Map view
Every geotagged photo and video on one map.
Read-along transcripts
The words follow the audio as it plays.
Light and dark themes
Comfortable to use day or night.
Ctrl+Z undo
Undo naming actions in a keystroke.
Privacy is the whole point
This is the center and it is true
All of the AI runs locally. Your library, your search and every part of the AI stay on your machine, with no account and nothing stored on a server. VISURAL touches the network for only these things:
- A one-time model download from our CDN on first run (no identifiers sent).
- An update check: one small request at launch, which you can decline (no identifiers sent).
- A feature request or bug report, only when you type one and send it.
- Sharing, only when you choose it. If you share photos, those exact photos are uploaded to the sharing service so your recipient can open them. Nothing is uploaded until you pick what to send and press share.
- The map. Opening the map view fetches its tiles from OpenStreetMap, so those servers see which areas of the world you are viewing. None of your photos and none of your locations are uploaded.
After setup, everything except a share you send and the map you open works fully offline. If you are not comfortable uploading your family photos to a cloud to get this kind of search, that is exactly why VISURAL exists.
Requirements
- Windows 10 or 11, 64-bit. Installer about 62 MB.
- First run downloads AI models once and keeps them on the device: about 274 MB always, with a vision model sized to your GPU (about 1.5 GB on modest hardware, about 3.2 GB on a strong GPU), so roughly 1.8 GB or 3.5 GB total, one time.
- Runs on NVIDIA, AMD and Intel GPUs and falls back to CPU.
- A GPU makes it much faster: with one, a photo is described in a few seconds; without one, about six seconds on typical desktop hardware. Speed varies a lot by machine, and the app detects which you have and tells you.
- No internet needed after setup.
Simple pricing
Try it free on a real folder. Unlock everything with a one-time key. No subscription, ever.
Free
Full features, up to 500 items indexed. Try it on a real folder.
Pro
Unlimited items. Activated with a key. Already own a Sherlock Forensics licence? It unlocks Pro too.
The cap limits new indexing only. Even after a trial ends, you keep browsing and searching everything already indexed.
Who it is for
- Anyone with a big personal library who cannot find anything in it.
- People with 10+ years of phone photos across drives and folders.
- Families wanting every photo of one person or pet.
- Anyone who will not upload family photos to Google or Apple to get smart search.
- NAS owners: it works on network shares.
Honest limits
We would rather tell you up front than have you find out later.
- Windows only for now.
- First run needs an internet connection and a multi-gigabyte download.
- Face recognition needs a reasonably large, reasonably sharp face. Small or blurry video faces are discarded rather than guessed.
- Long videos are sampled at up to 120 frames, so they are summarised at intervals, not continuously.
- VISURAL describes what is visible. It does not identify strangers and does not know who anyone is until you name them.
- Video face timestamps are not yet saved between sessions.
- Captions are a search aid, not a record. Accuracy is good but not perfect.
Release History
Version 1.1.0 (2026-07-29). A library you search became one you act on. Sharing now carries a whole album at any size. Albums group your photos. Dates let you jump to any month or open On this day. More like this finds a photo by another photo, even while the AI engine is stopped. The privacy claims were also corrected to be exact: nothing is uploaded unless you choose to share it. The map view fetches its tiles from OpenStreetMap.
Version 1.0.0 (2026-07-27). The first release. Local plain-language search across photos and videos, face and pet recognition, text-in-image reading, speech transcription with read-along, video timelines, the Progression view and a map of where your photos were taken.
Point it at one messy folder
Download the free version, point it at one messy folder and search it. That is the whole pitch.