By Ned Walsch · Last updated:
Updated 8 October 2026: added Copyseeker (freemium); Bing now opens in paste mode because its search-by-URL link redirected to the Bing Images home page in our tests; face engines are labelled paid or freemium.
Reverse Image Search Hub
Search one picture on every reverse-image engine from a single page. Paste an image URL and get pre-filled links for Google Lens, Bing, Yandex, TinEye, SauceNAO, Baidu and more; drop a local file and get its SHA-256, perceptual hashes and a metadata-free copy, plus each engine’s upload page with paste instructions. Everything runs in your browser — nothing is uploaded.
Nothing is uploaded. The file is read in this tab to show a preview, compute hashes and build a cleaned copy. A search engine can only match an image it can fetch, so for a local file the launcher below opens each engine’s own upload page and you paste or drop the picture there yourself.
The address is handed to each engine inside the link you click; this page stores nothing. The preview is loaded by your browser straight from the image host. Use a direct link to the image file (ends in .jpg, .png, .webp …), not the web page that shows it.
Results
The cleaned copy is re-encoded from the decoded pixels, so EXIF, GPS, XMP, IPTC and embedded thumbnails are gone and the orientation is baked in — but the pixels change slightly and its SHA-256 will differ. Share the cleaned copy; keep the original for forensics, hashing and evidence.
- Click Copy image to clipboard above (or keep the file handy in a folder).
- On the engine’s page, click its camera / “search by image” control.
- Paste with Ctrl+V / ⌘V, or drag the file into the drop area. Google Lens, Bing, Yandex and TinEye all accept a pasted image.
Badges: auto = the engine opens with your image URL already loaded. paste = the engine has no URL endpoint we can rely on (or you are working from a local file), so it opens the upload page and you paste or drop the image there. Max Intel tools run in your browser; drop the same file there.
Which reverse image search engine finds what?
Short answer: start with Google Lens for general identification, add Yandex when a face or Eastern-European content is involved, use TinEye to find the oldest copy, Bing for products, and SauceNAO / IQDB / trace.moe for anime and fan art. No engine indexes everything, so run at least three and compare. The table below is drawn from each engine’s own documentation (vendor claims) and from Bellingcat’s reverse-image-search guide. Bellingcat (2019-12-26)
| Engine | Best for | Search by URL | Notes |
|---|---|---|---|
| Google Lens | Broadest general index; identifies objects, landmarks, products, text | Yes — lens.google.com/uploadbyurl?url= | Replaced the classic “Search by image”; results open in Google Search. Google (vendor) |
| Bing Visual Search | Products and shopping matches, visually similar images | Paste only here | In our 8 Oct 2026 tests the old q=imgurl: search link redirected to the Bing Images home page, so we open Bing Images: click the camera icon and paste the image or its URL. Microsoft (vendor) |
| Yandex Images | Faces, look-alike scenes, content from Russia and Eastern Europe | Yes — yandex.com/images/search?rpt=imageview&url= | Widely recommended by investigators for face-similar results. Bellingcat (2019-12-26) |
| TinEye | Exact and edited copies; sort by oldest to find first appearance | Yes — tineye.com/search?url= | Does not do face or “similar object” matching by design. TinEye FAQ (vendor) |
| SauceNAO | Anime, manga, fan art, pixiv / booru sources | Yes — saucenao.com/search.php?url= | Rate-limited for anonymous users. SauceNAO (vendor) |
| IQDB | Anime / booru image boards | Yes — iqdb.org/?url= | Multi-service booru lookup. IQDB (vendor) |
| trace.moe | Which anime episode and timestamp a frame comes from | Yes — trace.moe/?url= | Scene search, not artwork search. trace.moe (vendor) |
| Baidu | Chinese-language web | Yes — image.baidu.com/n/pc_search?queryImageUrl= | Redirects to Baidu’s 识图 (shitu) results. Baidu (vendor) |
| Sogou / 360 | Chinese-language web, second opinions to Baidu | Yes — pic.sogou.com/ris?query=, st.so.com/stu?imgurl= | Interfaces are in Chinese. Sogou (vendor) |
| Karma Decay | Reddit reposts | Yes — karmadecay.com/search?q= | Reddit-only index; could not be reached from our build environment, so treat as best-effort. Karma Decay (vendor) |
| Copyseeker | Source tracing plus an AI-image check in one upload | Upload page only here | Freemium: free searches with ads, paid plans for unlimited use. Copyseeker pricing (vendor) |
| ascii2d | Japanese illustration sources (Twitter, pixiv) | Upload page only here | Has a URL form on its site; automated links are refused, so we open the page. ascii2d (vendor) |
| PimEyes, FaceCheck.ID, Lenso.ai, Social Catfish | Matching a specific face across the web | No — upload on their site | Full source lists are usually behind a paid plan: PimEyes and Social Catfish are paid, FaceCheck.ID and Lenso.ai are freemium. Only search faces you are authorised to look up. PimEyes (vendor) |
How does this page fingerprint an image without uploading it?
The browser reads the file into memory and does four things locally. It computes the SHA-256 of the exact bytes with the Web Crypto API, so you can prove later that an evidence file is unchanged. It draws the image onto small canvases and derives two perceptual hashes: a 64-bit dHash (9×8 greyscale, each bit says whether a pixel is brighter than its right-hand neighbour) and a 64-bit pHash (32×32 greyscale, 2-D DCT, the 8×8 low-frequency block thresholded at its median — the same convention as the Python imagehash library and as our Near-Duplicate Finder, so the hex values are directly comparable). Wikipedia: Perceptual hashing It scans the container for metadata blocks (JPEG APP1/APP13 segments, PNG eXIf/text chunks, WebP EXIF/XMP chunks) to tell you whether anything is there to strip. And on request it re-encodes the decoded pixels with canvas.toBlob(), which produces a file with no metadata at all. MDN: HTMLCanvasElement.toBlob() Browsers apply the EXIF orientation tag when they decode a photo, so the cleaned copy comes out upright even though the tag is gone. MDN: image-orientation
What is a sensible reverse-image workflow for an investigation?
- Preserve first. Record the SHA-256 of the original before you do anything else, and keep that file untouched.
- Read the metadata. Open the file in the EXIF Viewer: camera, timestamps, software and GPS can answer the question before any engine does.
- Search broadly. Run Google Lens, Yandex and TinEye at minimum; add Bing for products and Baidu for Chinese-language leads. Sort TinEye by oldest to find the earliest copy.
- Pivot on hits. For each page that hosts the image, check the Wayback Machine for earlier versions and captions that have since changed.
- Check manipulation. Run suspicious hits through Photo Forensics Studio (error-level analysis, clone detection) and compare perceptual hashes in the Near-Duplicate Finder to see which copies are crops or edits of which.
- People last, lawfully. Only move to face engines when you have a legitimate reason and authorisation, and document that reasoning in your notes.
Frequently asked questions
Can I reverse image search a photo that is only on my phone or computer?
Only after the engine has the pixels. Every engine either accepts a public image URL or wants you to upload the file on its own site. This page never uploads anything, so for a local file it opens each engine’s upload page and tells you where to paste or drop the image. If you need a URL instead (for example to share one link that works on every engine), the legacy helper at image.php will host the file on our server for 24 hours and then delete it.
Which reverse image search engine is best?
No single engine wins. Google Lens has the broadest general index and identifies objects, places and products. Yandex is widely regarded as the strongest free engine for faces and for content from Russia and Eastern Europe. TinEye finds exact and edited copies and can sort by oldest, which is how you find where an image first appeared. Bing is strong on products. SauceNAO, IQDB and trace.moe cover anime and fan art. Baidu, Sogou and 360 cover the Chinese web. Run several and compare.
Is anything uploaded when I use this page?
No. The file is read inside your browser to show a preview, compute the SHA-256 and perceptual hashes, and produce a cleaned copy. Engine links are built from an image URL you supply. The only way an image leaves your device is if you click an engine and paste or drop the file there yourself, or if you deliberately use the separate image.php helper, which says so before you upload.
Why do the engines need a public URL?
A search engine can only compare an image it can fetch. Google Lens, Yandex, TinEye and most of the others expose a search-by-URL endpoint that takes the address of an image on the web, downloads it and matches it against their index. A file on your disk has no address, so the engine has to receive it some other way — an upload on their site, or a temporary host that gives it a URL.
Does removing EXIF change the picture?
Slightly. The cleaned copy is produced by redrawing the decoded image onto a canvas and re-encoding it, which drops every metadata block (EXIF, GPS, XMP, IPTC, thumbnails) but also re-compresses the pixels. Use the cleaned copy for sharing and the untouched original for forensics, hashing or evidence, because the SHA-256 of the two files will never match.
What is the difference between SHA-256, dHash and pHash?
SHA-256 is a cryptographic digest of the exact bytes: change one byte and the hash changes completely, so it proves two files are identical. dHash and pHash are perceptual fingerprints of what the image looks like: the picture is shrunk to greyscale and the hash encodes brightness gradients (dHash) or low-frequency DCT coefficients (pHash), so resized or re-compressed copies produce the same or a nearby hash. Use SHA-256 for integrity and perceptual hashes to find look-alikes.
Is it legal to run a face search on someone?
It depends on where you and the subject are, and why you are searching. Face templates are biometric data, which the GDPR treats as a special category requiring an explicit legal basis, and some US states regulate biometric identifiers as well. Searching your own face, or a face you are authorised to look up as part of a lawful investigation, is the normal use case. Only search faces you are authorised to look up, and check the engine’s terms.
Can I link straight to a search from another tool?
Yes. Append ?url= or #url= followed by the URL-encoded image address, for example image.html?url=https%3A%2F%2Fexample.com%2Fphoto.jpg. The page validates that the link is http or https and no longer than 2048 characters, then shows the preview and every engine link ready to click. The Copy deep link button produces this form for the current image.