Our technology
at a glance
Deterring image-based abuse, pixel by pixel.
We saw the harm
& vowed to make a change
Our software embeds a traceable fingerprint into images at the point of access, not upload. This enables platforms to act on unauthorised sharing with forensic confidence – while protecting user privacy and platform integrity.
Do I need a new overlay for every image or video resolution?
Yes. Each resolution requires its own overlay to maintain watermark fidelity. This applies to both images and videos.
Do I also need multiple overlays per file for watermark rotation?
For video or advanced protection, yes. We recommend 3–5 overlays per resolution to rotate during playback and reduce visibility.
Will this increase billing units?
No. As long as you use the same WM_ID, multiple overlays are counted as one payload for billing.
What’s the best way to handle unusual image dimensions?
Request a high-resolution overlay (e.g. 4K x 4K) and crop as needed client-side. This saves on generation and works for smaller images too.
What is the minimum supported image size?
For 13-bit watermarking: 640×360
For 32-bit payloads: at least double that (e.g. 1280×720)
Smaller sizes may result in unreliable extraction or visual distortion.
Can I scale a watermark to fit non-standard sizes?
Yes, scaling is supported, but avoid excessive downscaling or stretching, which may affect performance.
How does Image Angel combat pre-CDN scraping or Chrome-based plugins?
We offer a server-side FFM plugin to protect images from scraping before they hit the CDN. However, deploying it can impact user experience, so it’s an optional layer based on your platform’s risk profile.
What is a distributor mark, and how does it work?
A distributor mark is a platform-specific identifier layered into the watermark. For example, all images served through a specific dating app can carry that app’s ID invisibly. It’s applied server-side before the individual user mark.
Are distributor marks included by default?
No. Distributor marks must be explicitly configured. They are not automatically added to every payload.
How does watermark rotation work for video?
Overlays rotate on a timer basis (e.g. every 30–50ms). They’re not frame-synced, but this is effective for deterring visibility, especially during panning or motion.
What’s the “dirty window” effect, and how do we prevent it?
A static overlay can become visible when a video pans or moves behind it. To avoid this, pre-fetch several overlays and rotate them randomly during playback. Sample code is provided.
Can overlay rotation be done on the client side?
Yes. While overlays are generated server-side, the client can cycle through them dynamically during playback.
Can we generate overlays on the client side?
No. Overlay generation must be done server-side. However, client-side overlay application is supported using the images retrieved via API.
Is there a client-side (e.g. JavaScript or WASM) option to embed watermarks into media?
No. At present, watermark embedding requires server-side generation. The client can only apply existing overlays, not create them.
Does the system use perceptual metrics to preserve quality?
Not on the client side. Watermark quality is maintained through robust encoding practices, but no perceptual tuning is applied in the client.
How are payloads counted for billing?
Billing is based on unique WM_IDs. Multiple overlays for the same ID (e.g. for different sizes or rotation) only count as one payload.
How fast is image delivery with watermarking applied?
The delay is minimal — approximately 200ms from API request to image delivery.
Forensic proof
The development and operation of Image Angel’s on-demand protection system involves a combination of advanced technologies across several areas. These include image processing, cryptography, and cloud infrastructure.
What Image Angel Does
Embeds invisible, traceable watermarks
Every image viewed is fingerprinted at access — not upload.Identifies the source if content is leaked
Makes it possible to trace abuse back to a specific viewer.Triggers forensic investigation on violation
No noise. No guesswork. Just proof when it matters.Integrates at platform level via API
Built for infrastructure, not individuals.Works in under 200ms
Fast enough for seamless delivery. Strong enough for legal evidence.Enables discreet resolution
Survivors and platforms can act without public exposure.Supports regulatory compliance
Aligns with OSA, DSA, and AB-3211 — reducing risk for platforms.Backs up education and trust strategies
Lets platforms support ethical digital intimacy with real consequences for breaches.Protects revenue and platform reputation
Stops content misuse before it damages users — or the brand.Scales with platform growth
Pricing is usage-based and designed for B2B SaaS flexibility.
What Image Angel Doesn’t Do – By Design
No installation or opt-in required
Users are protected automatically. No downloads, permissions or tech knowledge needed.Doesn’t shame, restrict or moralise
We don’t punish digital intimacy. We protect it.No tracking, profiling or surveillance
We don’t follow user behaviour. We only act when abuse is reported.Doesn’t expose uninvolved users
Forensic tracing targets only the source; not the bystanders.No changes to user experience
The watermark is invisible and does not alter the image or interrupt interaction.Doesn’t require users to report content
Platforms receive proactive signals when protected content is leaked.No visible watermark or metadata
Protection is server-side. It cannot be removed, cropped, or detected.Doesn’t rely on AI flagging or content filters
It’s not guessing what might be harmful; it proves what happened, when, and by whom.Doesn’t store personal data
Fingerprints link to access logs already held by the platform. We don’t collect or retain identifying info.