Guides to AI image generation and detection
Plain-language explanations of the ideas behind open-weight vision models, with sources.
- Open-weight vs open-source AIWhat each term promises, what it leaves out, and a checklist before you build on an open model.
- How AI image detection worksPixel forensics, provenance and watermarks: what each can tell you, where each fails, and how to read a score.
- C2PA and Content CredentialsHow a signed history travels with an image, what it proves, and why a missing credential isn't evidence of a fake.
- Reproducible image generationWhat a seed does, what else has to match to get the same image twice, why results still drift, and what to log.
Why we write these
People who have to decide whether to trust an image, or whether to build on a model, shouldn't need a research background to understand the trade-offs.
Each guide states what is known, says where methods fail, and links to its sources. If we can't back a claim up, we leave it out.
Get early access
Join the waitlist and we'll email you when the models and their licenses are released. One email, no spam, remove yourself any time.