ETCH — Coming
AI images of you that still look like you.
ETCH is an image engine built around a single hard problem: generative models will happily produce a beautiful photograph of somebody who is almost you. ETCH holds a structured identity and treats scene, wardrobe, expression and camera as modifiers layered on top of it.
What it is
ETCH makes images of a specific real person — the founder, the coach, the face of the brand — that stay recognisably that person across every shot.
It works from a structured identity blueprint rather than a prompt. The person is captured once, properly; after that, a scene, an outfit, an expression and a camera angle are modifiers applied on top of an identity that does not drift.
The intended outcome is unglamorous and very valuable: a business that needs forty on-brand photographs a month stops booking shoots it cannot afford and stops posting stock images of somebody else.
Why it was built
Content platforms are hungry. A founder posting consistently needs more images of themselves than any realistic photography budget covers, and the gap gets filled with stock, with screenshots, or with nothing.
General image models do not close that gap. Ask one for the same person twice and you get siblings — close enough to notice, wrong enough to be unusable as a brand asset.
ETCH exists because the identity problem is the whole problem. Solve that and the rest — scenes, wardrobe, angles — is layering.
How it was built
Strategy, product, build, launch.
Strategy
Scoped from a real bottleneck inside another product rather than from a market gap on a slide. ECHO can write a founder a fortnight of posts; the pictures were the part still missing.
Architecture first
Written and reviewed before the build: the identity model, the modifier layers, the generation pipeline and the phases they land in. The foundation and identity phases are complete; the generation pipeline is next.
Its own foundations
ETCH runs on its own database and its own infrastructure rather than inside ECHO. Products that will eventually be sold separately should not share a schema — that decision is cheap now and expensive later.
Where it actually is
Foundation and identity are built and tested. The generation pipeline has not run against a real database and has not produced a real image yet. That is the honest status, and it is why this page has no outcome section.
The outcome
Not yet.
ETCH has not launched. There are no results to report, and there will be no invented ones here. This section fills in when the generation pipeline runs.
A holding page for now — ETCH’s own home at createmyetch.com is not live yet.
Questions
What people ask about ETCH.
Can AI generate images of me that actually look like me?
Consistently, yes — but not from a prompt. The reliable approach is to capture the person once as structured identity data, then treat every scene, outfit and angle as a modifier applied on top of it.
Asking a general model for "a photo of me" repeatedly is what produces the near-miss effect, where each image is plausible and no two are the same person.
Can I have my own AI image tool built rather than paying per image?
Yes, and where the volume is high it is usually the cheaper answer over a couple of years — as well as the only way to get an image system that knows your brand rather than one that starts from zero every time.
The honest caveat is that it is a build, not a subscription. It is the same job as any custom app: specify it, build it, run it.
When will ETCH be available?
No date. Foundation and identity are complete; the generation pipeline is the next phase. When it can produce a real image against real data, that will be said here and not before.