The OSINT Newsletter - Issue #88
Image OSINT that Goes Beyond Tools
đ Welcome to the 88th issue of The OSINT Newsletter. This issue contains OSINT news, community posts, tactics, techniques, and tools to help you become a better investigator.
đ¨ The last day for the Black Friday discount is December 8th, 2025. This is the largest discount of the year. Follow the link below or click the button to take your skillset to the next level.
đŞ If you missed the last newsletter, hereâs a link to catch up.
⥠Discovering and Tracking Data Breaches and Leaks
đď¸ If you prefer to listen, hereâs a link to the podcast instead.
Letâs get started. âŹď¸
From profile pictures to product shots, vacation selfies to newsreels, nearly everything on the internet comes with an image attached. So in this issue, weâre looking at the big picture. Weâll cover everything you need to know about image-based OSINT, with tips, tricks and techniques to make pictures work for you. Weâll cover:
What reverse image search is
Traditional vs. AI image search
The best tools for the job
Face and object recognition 101
And how to make your photos actually return results.
Letâs start searching smarter - with pictures.
What Is Reverse Image Search?
Reverse image search basically flips the script youâre used to with search engines. Instead of sticking in a text query and waiting for results to bounce back, you upload an image, and the algorithm lets you know where else it appears online.
In general, reverse image search works by looking for visual similarity across an entire picture: background colours, edges, textures etc. It does this by comparing with an existing index of crawled images. Meanwhile, face recognition locks on to a specific face - and knows itâs a face - even if the background, clothing, lighting or angles change.
Thatâs why reverse image search will only rarely find real matches for people⌠unless theyâre literally Donald Trump. For investigations that involve identifying people (who arenât the most famous living human) face recognition is always your best bet.
But what about AI?
AI makes it possible to extract even more from a photo than just faces. They can identify vehicles (make, model, even trim details), read signs, recognise brands, detect landmarks and even spot patterns across different photos. Each element of a photo becomes searchable, and a potential pivot. AI can:
Recognise faces (even in different lighting or angles)
Detect objects and landmarks
Track altered, cropped, or edited pictures
Match photos to social media profiles
And even reveal where a face, building, or background has appeared before
This opens a whole new world of possibilities for image OSINT. If you know how to use it, reverse image search can pivot your investigation from just one picture to a whole load of genuine leads.
The Full Paintbox: OSINT Image Search Tools
Like colours on an artistâs palette, every OSINT image search tool is different. The difference comes from how they index, as well as how and where they search. Just like Google versus Bing (remember those?), different platforms index different content and work on different materials, so youâll need to use the whole paintbox to get the full picture.
These are the âmust-knowâ tools for basic image OSINT:
đ Yandex: The Rembrandt of face-focused searching.
Excellent at recognising faces, even with changes in background
Strong at matching architectural and geographic landmarks
Supports search operators like site: for precision
Uses AI search for max accuracy
Great for: people finding, cross-platform identity checks, and location clues. Especially effective for Eastern Europe and Russia - where the platform is most popular.
đ PimEyes: One of the most advanced public facial recognition tools available.
High match accuracy on human faces
Works with low-resolution images
Paid versions reveal unblurred matches and direct source links
Great for: identity verification, tracking reused profile pictures. Particularly useful if your search material is poor - or even potato - quality.
đ Lenso.ai: A modern AI-powered facial recognition engine.
Handles altered, heavily edited, or angled photos
Can compare multiple images at once
Sends alerts when a new match appears
Great for: Long-term monitoring and tricky face matches. Could probably identify a face from a 0.5x (take one and test this).
đ TinEye: An oldie, but a goodie.
Excels at finding older duplicates of the same image
Useful for detecting edits (e.g., cropped, flipped, or filtered versions)
One of the oldest engines, so brings up historic results
Great for: tracking image origins, copyright checks, spotting reused photos. TinEye has been finding lost memes for aeons, and itâs just as good for more serious images.
đ Google Lens & Bing Visual Search: Your general-purpose engines.
Good for objects, products, and âwhatâs this thing?â searches
Weak for identifying individuals (unless theyâre public figures)
Great for: object identification, location clues, UI-based searches. Better with objects than faces, these engines are perfect for identifying products or landmarks,
Finishing Touches: How to Improve your Search Results
Of course, sometimes your search will come up blank. A poor artist blames their tools - but the issue is usually with the image, rather than with your OSINT skills. The best way to bring out better results is to prime your images; hereâs how.
Remove the background: Some image search services work with busy backgrounds, but all will search better if the backgroundâs uncluttered. Use Photoshop - or a free online service like remove.bg - to remove the background.
Poor crop, youâll starve: Always try the full version first, before cropping the image into smaller pieces. If youâre isolating and investigating a certain part of the image (like a logo or building), crop down to just that part before you search.
Image restoration: If your image is blurry or obscured, it doesnât have to stay potato. Use AI image restoration to reconstruct the image. Of course, use your judgement; make sure the AI is fairly accurately recreating the picture, and not hallucinating nonsense.
Example: Image OSINT in Action
Letâs go through an example now. This time, you are just a humble potato farmer, and youâve met the potential love of your life on farmersonly.com. Her name is Mary, and her profile photo shows a stunning blonde woman in her late-thirties. You want to know if Mary is your true love - or if youâve shifted from farming potatoes to catfish husbandry.
Step One: Identify the Face
First, you need to identify the face in âMaryââs photo. You run the profile pic through Yandex, only to find that the same face appears on multiple VK accounts. Of course, Mary could be Russian⌠but the accounts all use different names. Major red flag.
Step Two: Find the Original
Youâve identified that this isnât Maryâs photo, so whose is it? Time to run it through TinEye and check for historical matches. It brings up an original, from a 2017 photography blog. So the picture was stolen; this confirms the catfish didnât even take it.
Step Three: AI Check the Objects
In the picture, the girl is wearing a watch. You crop down this section of the image, and search the cropped version with Google Lens; it returns a distinctive watch brand. When you Google that brand, you find out itâs most popular in Belarus, not the US (where Mary claims to be).
Step Four: Search the Background
You zoom in on the background, and see a reflection of Cyrillic lettering in the window. AI identifies this as the Belarusian language.
Step Five: Pivot to Social Media
Using the region, watch brand and the original photographerâs social pages, you end up on the original modelâs legitimate Instagram. Time to let her know her photoâs been stolen⌠and return to the potato fields with a broken heart đ.
Key Takeaways
So now you know the basics of image OSINT. Hopefully, this paint-by-numbers guide helps you out and leaves you with some essential skills. So remember:
Use the full paintbox. AI and traditional tools are all useful, for their own reasons.
Search the whole picture. Nowadays, any part of an image is a potential pivot.
Keep it prime. No results? Prime your image and try it again.
Faces and objects take their own tools, but can both take your investigation places.
â Thatâs it for the free version of The OSINT Newsletter. Consider upgrading to a paid subscription to support this publication and independent research.
By upgrading to paid, youâll get access to the following:
đ All paid posts in the archive. Go back and see what youâve missed!
đ If you donât have a paid subscription already, donât worry. Thereâs a 7-day free trial. If you like what youâre reading, upgrade your subscription. If you canât, I totally understand. Be on the lookout for promotions throughout the year.
đ¨ The OSINT Newsletter offers a free premium subscription to all members of law enforcement. To upgrade your subscription, please reach out to LEA@osint.news from your official law enforcement email address.