{"course":{"slug":"ai-images","title":"Image Generation, in Practice","url":"https://addaly.com/learn/ai-images","level":"beginner","track":"create"},"lesson":{"slug":"writing-the-spec","title":"The one page that ends the revision argument","url":"https://addaly.com/learn/ai-images/writing-the-spec","position":20,"of":69,"minutes":7,"answer":"A signed one-page spec that names dimensions, ratio, forbidden content, the reference pack and what will be photographed rather than generated converts later disputes from matters of taste into matters of record.","keyIdea":"A signed one-page spec that names dimensions, ratio, forbidden content, the reference pack and what will be photographed rather than generated converts later disputes from matters of taste into matters of record."},"body":{"format":"markdown","text":"## What the document is for\n\nNot bureaucracy. The spec exists so that the sentence \"this isn't what we wanted\" has somewhere to be resolved. Without it, every review is a negotiation about remembered intentions. With it, a review is a comparison between an image and a page you both agreed to.\n\nIt fits on one side. If it runs longer, it has become a proposal and something else has gone wrong.\n\n## The template\n\n```markdown\n# Image spec — [client] — [date]\n\n## Deliverables\n- Hero: 1200 × 1500 px, JPEG, sRGB\n- Crops from the same frame: 1080 × 1080, 1080 × 1920\n- Print version: A4 at 300 dpi, CMYK, 3 mm bleed\n\n## Generation plan\n- Native render ratio: 896 × 1152 (SDXL bucket), upscaled and cropped\n- Crop-safe core: central square; headline sits in the upper third\n\n## The picture\nOne sentence describing the image. Not a prompt — a sentence a person understands.\n\n## Reference pack\n1. light — ref-01.jpg — the hard side light, not the styling\n2. palette — ref-02.jpg — the muted greens\n3. composition — ref-03.jpg — subject low and left, lots of sky\n4. product — ref-04.jpg — our actual bottle, photographed 14 Mar\n5. texture — ref-05.jpg — matte, fine grain, not glossy\n6. anti-reference — ref-06.jpg — too clean, too corporate\n\n## Constraints\n- Brand red #E4002B, exact, on the label only\n- Headline zone: upper third must be quiet, contrast ratio at least 4.5:1\n\n## Must not appear\n- Legible brand names or logos of any kind\n- Text other than type we set ourselves\n- Hands\n\n## Photographed, not generated\n- The bottle (composited into a generated environment)\n\n## Rounds\nTwo rounds of revisions included. Structural changes — different composition,\ndifferent concept — are a new round.\n\n## Disclosure\nClient is informed the environment is AI-generated. Delivery includes a note\nfor their records.\n```\n\n## The four lines that do the work\n\nMost of that template is useful. Four lines are load-bearing.\n\n**\"Photographed, not generated\"** sets the expectation that some elements are real before anyone has an opinion about the picture. It is far easier to state on day one than to explain on day six.\n\n**\"Must not appear\"** is the list a person checks against the finished file. It is the difference between an embarrassment caught in QA and one caught by the client's legal team.\n\n**The rounds clause** distinguishes a tweak from a rebuild. \"Can we make it warmer\" is a tweak. \"Can she be facing the other way\" is a regeneration that loses everything downstream of the original frame, and the spec is where that gets priced.\n\n**The disclosure line** means nobody is surprised. The rights block later covers what you must say and where; the point here is that it is agreed at the start rather than discovered at the end.\n\n## Getting it signed without ceremony\n\nYou do not need a legal process. An email saying \"here's the spec: confirm and I'll start\" and a reply saying \"confirmed\" is enough for the purpose, which is a shared record rather than an enforceable instrument.\n\nWhat matters is the timing: before generating. A spec written after the first review is a defence, and everyone can tell.\n\n## Its real limit\n\nA spec cannot make a client know what they want. Some do not until they see something, and insisting on precision before any image exists can stall a project that a quick sketch would unblock.\n\nSo use it proportionately. For a small job, three lines in an email (dimensions, what must not appear, how many rounds) carries most of the value. For anything with print, legal exposure or more than a few images, write the page. And if a client cannot answer the questions yet, generate three cheap, deliberately rough concepts to make the conversation concrete, then write the spec from whichever one they point at. The spec is a tool for removing ambiguity, not a gate that ambiguity has to pass first.\n\n## Keep it alive\n\nA spec written once and never touched becomes fiction by the third day, because briefs move. The version that stays useful is edited as decisions are made, with a short note of what changed and when.\n\n```markdown\n## Changes\n- 14 Mar: headline moved to lower third — upper third now free\n- 16 Mar: bottle to be photographed, not generated (label unreadable in tests)\n- 18 Mar: second size added, 1080 × 1920\n```\n\nThree lines like that resolve more disputes than the entire rest of the document, because the arguments that actually happen are almost never about the original brief. They are about a change somebody remembers differently.\n\nA last, unglamorous point: keep the spec beside the work, in the project folder, not in a chat thread. Chat histories are searched badly and scroll away. A file called `brief.md` sitting next to the images is still there in a year, when someone asks why the composition has all that empty space at the top and nobody remembers that a headline was supposed to go there."},"check":{"question":"What does an image spec actually accomplish at review time?","options":[{"text":"It transfers responsibility for any content problem in the delivered file from the person who made the image to the client who approved the specification document.","correct":false,"tempting":"Misreads a shared working record as a liability shield; the forbidden list still has to be checked by whoever produces the file."},{"text":"It documents the exact technical parameters in enough detail that the approved image could be regenerated later from the specification alone.","correct":false,"tempting":"Reproducibility comes from recorded model hashes and saved files, not from a brief; a spec describes intent rather than settings."},{"text":"It turns a disagreement about taste into a comparison against a record both sides agreed to beforehand.","correct":true},{"text":"It allows the number of revision rounds to be enforced, which is the main commercial protection the document provides on a fixed-price job.","correct":false,"tempting":"The rounds clause is genuinely useful, but it is one line of a document whose broader function is to make every later disagreement checkable."}]},"prev":{"slug":"what-to-shoot-instead","title":"The frames where a phone camera wins","url":"https://addaly.com/learn/ai-images/what-to-shoot-instead"},"next":{"slug":"rights-and-disclosure","title":"What you can sell, and what you must disclose","url":"https://addaly.com/learn/ai-images/rights-and-disclosure"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'The one page that ends the revision argument', in Image Generation, in Practice. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-images/writing-the-spec (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"The one page that ends the revision argument","courseTitle":"Image Generation, in Practice","module":{"index":2,"slug":"the-brief","title":"From a request to a specification you can generate against"},"answer":null,"keyIdea":"A signed one-page spec that names dimensions, ratio, forbidden content, the reference pack and what will be photographed rather than generated converts later disputes from matters of taste into matters of record.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"what-the-document-is-for","runs":[{"text":"What the document is for"}]},{"t":"p","runs":[{"text":"Not bureaucracy. The spec exists so that the sentence “this isn’t what we wanted” has somewhere to be resolved. Without it, every review is a negotiation about remembered intentions. With it, a review is a comparison between an image and a page you both agreed to."}]},{"t":"p","runs":[{"text":"It fits on one side. If it runs longer, it has become a proposal and something else has gone wrong."}]},{"t":"h2","id":"the-template","runs":[{"text":"The template"}]},{"t":"code","lang":"markdown","label":"Markdown","text":"# Image spec — [client] — [date]\n\n## Deliverables\n- Hero: 1200 × 1500 px, JPEG, sRGB\n- Crops from the same frame: 1080 × 1080, 1080 × 1920\n- Print version: A4 at 300 dpi, CMYK, 3 mm bleed\n\n## Generation plan\n- Native render ratio: 896 × 1152 (SDXL bucket), upscaled and cropped\n- Crop-safe core: central square; headline sits in the upper third\n\n## The picture\nOne sentence describing the image. Not a prompt — a sentence a person understands.\n\n## Reference pack\n1. light — ref-01.jpg — the hard side light, not the styling\n2. palette — ref-02.jpg — the muted greens\n3. composition — ref-03.jpg — subject low and left, lots of sky\n4. product — ref-04.jpg — our actual bottle, photographed 14 Mar\n5. texture — ref-05.jpg — matte, fine grain, not glossy\n6. anti-reference — ref-06.jpg — too clean, too corporate\n\n## Constraints\n- Brand red #E4002B, exact, on the label only\n- Headline zone: upper third must be quiet, contrast ratio at least 4.5:1\n\n## Must not appear\n- Legible brand names or logos of any kind\n- Text other than type we set ourselves\n- Hands\n\n## Photographed, not generated\n- The bottle (composited into a generated environment)\n\n## Rounds\nTwo rounds of revisions included. Structural changes — different composition,\ndifferent concept — are a new round.\n\n## Disclosure\nClient is informed the environment is AI-generated. Delivery includes a note\nfor their records."},{"t":"h2","id":"the-four-lines-that-do-the-work","runs":[{"text":"The four lines that do the work"}]},{"t":"p","runs":[{"text":"Most of that template is useful. Four lines are load-bearing."}]},{"t":"p","runs":[{"text":"”Photographed, not generated”","b":true},{"text":" sets the expectation that some elements are real before anyone has an opinion about the picture. It is far easier to state on day one than to explain on day six."}]},{"t":"p","runs":[{"text":"”Must not appear”","b":true},{"text":" is the list a person checks against the finished file. It is the difference between an embarrassment caught in QA and one caught by the client’s legal team."}]},{"t":"p","runs":[{"text":"The rounds clause","b":true},{"text":" distinguishes a tweak from a rebuild. “Can we make it warmer” is a tweak. “Can she be facing the other way” is a regeneration that loses everything downstream of the original frame, and the spec is where that gets priced."}]},{"t":"p","runs":[{"text":"The disclosure line","b":true},{"text":" means nobody is surprised. The rights block later covers what you must say and where; the point here is that it is agreed at the start rather than discovered at the end."}]},{"t":"h2","id":"getting-it-signed-without-ceremony","runs":[{"text":"Getting it signed without ceremony"}]},{"t":"p","runs":[{"text":"You do not need a legal process. An email saying “here’s the spec: confirm and I’ll start” and a reply saying “confirmed” is enough for the purpose, which is a shared record rather than an enforceable instrument."}]},{"t":"p","runs":[{"text":"What matters is the timing: before generating. A spec written after the first review is a defence, and everyone can tell."}]},{"t":"h2","id":"its-real-limit","runs":[{"text":"Its real limit"}]},{"t":"p","runs":[{"text":"A spec cannot make a client know what they want. Some do not until they see something, and insisting on precision before any image exists can stall a project that a quick sketch would unblock."}]},{"t":"p","runs":[{"text":"So use it proportionately. For a small job, three lines in an email (dimensions, what must not appear, how many rounds) carries most of the value. For anything with print, legal exposure or more than a few images, write the page. And if a client cannot answer the questions yet, generate three cheap, deliberately rough concepts to make the conversation concrete, then write the spec from whichever one they point at. The spec is a tool for removing ambiguity, not a gate that ambiguity has to pass first."}]},{"t":"h2","id":"keep-it-alive","runs":[{"text":"Keep it alive"}]},{"t":"p","runs":[{"text":"A spec written once and never touched becomes fiction by the third day, because briefs move. The version that stays useful is edited as decisions are made, with a short note of what changed and when."}]},{"t":"code","lang":"markdown","label":"Markdown","text":"## Changes\n- 14 Mar: headline moved to lower third — upper third now free\n- 16 Mar: bottle to be photographed, not generated (label unreadable in tests)\n- 18 Mar: second size added, 1080 × 1920"},{"t":"p","runs":[{"text":"Three lines like that resolve more disputes than the entire rest of the document, because the arguments that actually happen are almost never about the original brief. They are about a change somebody remembers differently."}]},{"t":"p","runs":[{"text":"A last, unglamorous point: keep the spec beside the work, in the project folder, not in a chat thread. Chat histories are searched badly and scroll away. A file called "},{"text":"brief.md","code":true},{"text":" sitting next to the images is still there in a year, when someone asks why the composition has all that empty space at the top and nobody remembers that a headline was supposed to go there."}]}],"check":{"question":[{"text":"What does an image spec actually accomplish at review time?"}],"options":[{"text":[{"text":"It transfers responsibility for any content problem in the delivered file from the person who made the image to the client who approved the specification document."}],"correct":false,"tempting":[{"text":"Misreads a shared working record as a liability shield; the forbidden list still has to be checked by whoever produces the file."}]},{"text":[{"text":"It documents the exact technical parameters in enough detail that the approved image could be regenerated later from the specification alone."}],"correct":false,"tempting":[{"text":"Reproducibility comes from recorded model hashes and saved files, not from a brief; a spec describes intent rather than settings."}]},{"text":[{"text":"It turns a disagreement about taste into a comparison against a record both sides agreed to beforehand."}],"correct":true,"tempting":null},{"text":[{"text":"It allows the number of revision rounds to be enforced, which is the main commercial protection the document provides on a fixed-price job."}],"correct":false,"tempting":[{"text":"The rounds clause is genuinely useful, but it is one line of a document whose broader function is to make every later disagreement checkable."}]}],"order":[3,1,2,0],"why":null},"prev":{"slug":"what-to-shoot-instead","title":"The frames where a phone camera wins","minutes":7},"next":{"slug":"rights-and-disclosure","title":"What you can sell, and what you must disclose","minutes":10}}}