research
Photoshop: the history of an image no longer trusted
The software made the image open, layered, and reversible — and thereby suspect.
Photoshop matters not only as retouching software but as a cultural mechanism that made the image open, layered, reversible, and suspicious.
1. Research thesis
The software changed not the quality of processing but the presumption. Before it, an image was assumed to be evidence; after it, an image is assumed to be processed. What moved was not technique but the burden of proof.
2. Why this is more than a software history
Retouching existed from the nineteenth century and was always technically possible. What matters is different: the tool became widely available and the result stopped leaving visible traces of intervention.
3. Historical setup
The first version was released in 1990 for the personal computer. The moment was determined by technology: raster displays and sufficient memory made on-screen image work practical.
4. What came before
Analogue retouching required a workshop, brushes, dyes, and work on a negative or print. It was expensive, slow, and available to few — and its use was therefore limited.
5. The layer as the decisive invention
The arrival of layers in the mid-1990s changed the nature of the file. An image stopped being a single impression and became an assembly of separable parts, each editable independently.
6. The mask and reversibility
The mask allowed concealment rather than deletion. Together with an action history this created a fundamentally new situation: any change is reversible, and so a decision stopped being final.
7. Non-destructive editing as a discipline
A professional norm formed out of these technical possibilities: do not damage the original, work with adjustment layers, preserve reversibility. It is a craft code that formed around the tool.
8. The mutability of the image
When a file is open to editing at any stage, the moment of finality disappears. Publication becomes the only point of fixation — and so the question of version acquires significance.
9. Body retouching and editorial norms
Mass processing of figure and skin in advertising and magazines created a public problem beyond the profession. The response was editorial policy and, in some countries, labelling requirements.
10. Documentary photography and prohibitions
News agencies introduced strict limits on processing a news frame: correction is permitted, alteration of content is not. The boundary is drawn administratively because technically there is none.
11. Compositing: assembly instead of exposure
Combining several sources into one frame became a routine operation. A frame stopped corresponding to one moment and one viewpoint — and this concerns advertising more than documentary.
12. Colour as an authorial decision
Colour grading turned colour from a property of film stock into a matter of choice. This gave control and simultaneously deprived the image of an objective chromatic tie to an event.
13. The economy of form
The cost of processing fell so far that processing became cheaper than reshooting. This changed production logic: part of the decisions is deliberately deferred to post-production.
14. The word entered the language
The product name became a verb and an accusation. Few tools produce that effect; it shows that we are dealing with a cultural phenomenon and not only a product.
15. Canon and museum
The software and its source code entered museum collections as objects in the history of computing. This recognises its role as infrastructure of visual culture rather than as a product.
16. Archive and reproducibility
A file with preserved layers is itself an archive of process: it shows what was done and in what order. A flattened image loses that information irreversibly.
17. Where the myth begins
The myth begins with the idea that photography was honest before the software. Staging, frame selection, cropping, and darkroom printing were always instruments of interpretation.
18. Where the myth blocks analysis
Moralising obscures the real problem: not processing but the absence of a record of it. What is harmful is not the edit but the impossibility of knowing an edit occurred.
19. First key example
The layer and the mask deserve analysis as a new logic of trust: an image became a device with internal structure rather than a surface. This changes what can be asked about it at all.
20. Second key example
Provenance systems are worth analysing as the industry’s answer to a problem it created. A record of creation and alterations embedded in the file restores the possibility of verification.
21. Comparison with adjacent systems
Unlike generative tools, here an original always exists. That is a fundamental difference: processing can be proved by comparison, generation cannot.
22. Link to fashion
Fashion photography was the main field of application and the main source of public objection. It is here that body processing became normal earliest and most visibly.
23. Link to the image
The software standardised operations across the industry: curves, levels, masks, blending modes. The professional vocabulary of processing is largely the vocabulary of one product.
24. Link to architecture and space
Architectural visualisation depends entirely on post-production: sky, people, vegetation, and light are added after rendering. An image of a building is almost never a photograph of a building.
25. Link to technology
The software became a platform: formats, plug-ins, and workflows formed around it. Changing such a platform costs an industry more than changing cameras.
26. Link to materials
A screen image has no material, and this created a separate problem: matching screen and print required colour management as a discipline in its own right.
27. Market and logistics
The move to subscription changed the relationship with the user: access to one’s own files became contingent on an active contract. This is a new kind of dependency that did not exist in the boxed-software era.
28. Critique and limits
The critique is justified in two directions: the tool helped spread unattainable standards of appearance and made image verification difficult. Both objections concern use, but the use was predictable.
29. What has changed today
The trust problem has moved to generative images, where no original exists at all. Against that background, processing a photograph looks like a comparatively verifiable operation.
30. What to connect next
The material links to the entries on retouching, compositing, colour grading, authorship in the age of generative models, and digital photography.
31. Editorial conclusion
The image lost trust not because it became possible to alter it but because the record of alteration disappeared. Restoring trust is a task of documentation, not of prohibition.
32. How to read this case today
Ask not "was this processed" but "is there a record of what was done". The first question almost always has the answer yes and is therefore useless.
Sources
- Adobe — Nondestructive editing in Photoshop
- Adobe — Add layer masks
- Adobe — Understanding colour management
- Computer History Museum — Make Software: Photoshop
- Computer History Museum — Adobe Photoshop source code
- Adobe — Photoshop turns 25: Q&A with Thomas Knoll
- OpenAI — Advancing content provenance