Bildwerk
visual similarity
Every image is embedded with OpenAI CLIP ViT-B/32 (MIT, fp16 ONNX, SHA-256 pinned; EXIF-rotated, shorter side to 224, centre crop) and compared by cosine similarity. Pairs are labelled near_duplicate (cosine >= 0.95, or >= 0.9 with dHash distance <= 10 of 64 bits), similar (>= threshold) or different; groups are the connected components at the threshold. A 64-bit difference hash (9x8 grey) is returned per pair as an independent pixel-level check. In our tests different photos stayed below.
POST https://bild.halowerk.com/v1/visual-similarityInputs
| Field | Type | Required | Description |
|---|---|---|---|
| images | array | Yes | Two to twelve images, each as image_base64 or image_url (each at most 10 MB, together at most 30 MB). |
| threshold | number | No | Cosine similarity from which two images count as similar and are grouped. |
Show full input schema
{
"type": "object",
"additionalProperties": false,
"required": [
"images"
],
"properties": {
"images": {
"type": "array",
"minItems": 2,
"maxItems": 12,
"items": {
"type": "object",
"additionalProperties": false,
"properties": {
"id": {
"type": "string",
"minLength": 1,
"maxLength": 64,
"description": "Optional label echoed in the result."
},
"image_base64": {
"type": "string",
"minLength": 16,
"maxLength": 14000000,
"description": "The image (PNG, JPEG, WebP, GIF, TIFF or AVIF) as base64 or data: URL, at most 10 MB decoded. Give either image_base64 or image_url."
},
"image_url": {
"type": "string",
"format": "uri",
"maxLength": 2048,
"description": "Public http(s) URL of the image, fetched once (at most 10 MB). Private and local addresses are refused."
}
}
},
"description": "Two to twelve images, each as image_base64 or image_url (each at most 10 MB, together at most 30 MB)."
},
"threshold": {
"type": "number",
"minimum": 0.5,
"maximum": 0.99,
"default": 0.85,
"description": "Cosine similarity from which two images count as similar and are grouped."
}
}
}Published input example
{
"images": [
{
"id": "a",
"image_base64": "iVBORw0KGgoAAAANSUhEUgAAAEAAAAAwBAMAAABTZWEsAAAAJ1BMVEV6uOjo16j/003jzW9+ueX50lTmzmuIu9eyw6fmzmqzxKWzw6b40VZx3hZPAAAACXBIWXMAAAsTAAALEwEAmpwYAAAAV0lEQVQ4y2NgGLzAyzR4Cz559klKSmoFeBR4KgFBCh4Fi0AKNPAoMAIpUMajIAikQJWQAh1KrNhEyJFuhLzJnkQgoBg8TINbGEbBsAOCBMCoglEFJCkAAMKYNKuC2Gj3AAAAAElFTkSuQmCC"
},
{
"id": "b",
"image_base64": "iVBORw0KGgoAAAANSUhEUgAAAEAAAAAwBAMAAABTZWEsAAAAMFBMVEU/j0wvazpaOx4+jUswbzw2ekE0dj89ikk6hEc5gUVXRSNHdj9DgkVKbTlOXTE3fUPlyye/AAAACXBIWXMAAAsTAAALEwEAmpwYAAAAlklEQVQ4y2MQJAAY6KRAJKLYvNURtwKRCQxAwOGIU0EyAxiY4VIgygAFgTgUJMAUsGFXIGIAU8DsiFWBJAMcTMSqwAGhgAWrggCEAlasChoQCjiwKihAKGDHqsAAoYCZPAUErSDoSILeJBhQBIOaYGQRjG7CCYZgkhMUuQyS58WdaAW1Xh+v2bcIT75QAoNRBaMKBoUCAA+rbbd55OrkAAAAAElFTkSuQmCC"
}
]
}Result & documentation
Machine-readable documentation of this hub
Source of this page: the published HALOWERK catalog. Updated: 2026-11-17 UTC.
More endpoints in this hub
- Bildwerk0.01 USDC
address plate blur
POSThttps://bild.halowerk.com/v1/address-plate-blurThe photo is read with Tesseract 5.3.4 (German + English, word boxes) at two scales and in sparse-text and block mode; small bright plaques with 1–5 dark characters (typical house-number signs) are additionally cut out and read one by one,
- Bildwerk0.003 USDC
ai disclosure metadata
POSThttps://bild.halowerk.com/v1/ai-disclosure-metadataChecks JPEG, PNG and WebP files for AI disclosure metadata: XMP DigitalSourceType and the IPTC Photo Metadata 2025.1 AI fields, C2PA/JUMBF stores (signatures are not validated), PNG text chunks and EXIF Software/UserComment, and returns a s
- Bildwerk0.008 USDC
animation to video
POSThttps://bild.halowerk.com/v1/animation-to-videoReads frame count, size, loop count and total duration from the file structure, then encodes with ffmpeg 6.1 in a child process under prlimit (no shell, empty environment, only the file protocol): GIF and APNG are demuxed natively, animated