What a Film Photograph Actually Is
A film photograph starts as light passing through a lens and hitting a strip of chemically-coated plastic — the film. Where light hits, silver halide crystals react. That reaction is developed, fixed, and made permanent through real chemistry, producing a negative: a physical, unchangeable record of that exact moment of light.
Everything downstream of that negative — a darkroom print, a digital scan — is a reproduction of something that verifiably happened. The negative itself doesn't update, doesn't regenerate, and can be reprinted identically decades later. That permanence is the entire value proposition of shooting film in the first place.
What a Digital Photograph Actually Is
A digital photograph also starts with light through a lens, but instead of chemistry, it hits a digital sensor that converts light into an electronic signal, which a processor turns into a file. That file is real — it represents an actual moment the camera captured — but it's also infinitely editable, copyable, and, critically, mutable after the fact in ways a negative structurally isn't.
Digital photography is faster, cheaper per frame, and gives instant feedback — real, practical advantages for a lot of work. What it gives up is the negative's built-in permanence. A digital file's "originality" depends entirely on whether anyone bothered to preserve it unedited, which in practice, almost never happens anymore.
What an AI-Generated "Photograph" Actually Is
This is where the category breaks down entirely. An AI image generator doesn't capture light through a lens at all. It produces a statistically plausible arrangement of pixels based on a prompt and training data — an image that resembles a photograph without any camera, any lens, or any real moment behind it. Calling that a "photograph" is, strictly, inaccurate.
The same applies to AI-edited photography — an AI tool that reshapes a face, invents background detail, or "enhances" a real photo past what the camera actually captured. At that point, some meaningful percentage of the final image is invention, not documentation, even if it started from a real photo.
Why the Difference Matters More Than It Used To
For most of photography's history, a photograph was reasonably trustworthy evidence that something happened — a moment, a person, a place, in front of a real lens at a real time. AI image generation breaks that assumption for the first time at real scale, which is exactly why we think the distinction between "real photograph" and "AI-generated image" is going to matter more, not less, going forward.
Film sits at the far end of that spectrum — the most physically verifiable, hardest-to-fake format available. Digital sits in the middle, real but fragile to invisible editing. AI-generated imagery sits outside the category of photography altogether, however convincingly it's styled to look like one.
What This Means When You're Hiring a Photographer
If what you want is a genuine, verifiable image of a real moment — a portrait, a wedding, a headshot that's actually you — ask directly what's shot on film versus digital, and whether any AI editing touches the final result. A photographer confident in their process will answer plainly. One that gets vague is telling you something.
At The Film Photographer of San Francisco, every session is shot on real 35mm black and white film, hand-developed and hand-printed, with no AI anywhere in the process — no AI retouching, no AI-generated backgrounds, none of it. Text (415) 573-FILM if you want to talk through what that actually looks like for your specific session.
The Cost Difference, Explained Honestly
Film costs real money per frame before a single photo is even taken — the film stock itself, then developer, stop bath, and fixer chemistry, then archival photo paper for every print. A digital sensor's marginal cost per frame is close to zero once the camera is paid for. An AI generator's marginal cost is smaller still. That's the honest, unglamorous reason film photography costs more, and it's worth saying plainly instead of hiding it in a vague "premium service" framing.
That cost buys something specific: a negative that exists independently of any software, service, or subscription, and a print made through a physical process that hasn't fundamentally changed in over a century because it still produces something nothing else does.
A Simple Test You Can Use
Here's a practical way to think about it: ask whether the image could be reprinted, identically, in fifty years using only the original artifact — no software, no cloud account, no company still needing to exist. A film negative passes that test trivially. A digital file passes only if someone actively maintained it through every device and format change along the way. An AI-generated image doesn't really pass at all, since regenerating "the same" image from a prompt produces a different result every time.
That test isn't about nostalgia — it's about which format actually behaves the way people assume "a photograph" behaves when they're trusting it with something that matters to them.