ChatGPT Prompts for Photographers (2026): 36 Prompts by Stage

Key Takeaways

• A useful photography prompt carries five parts: who you are, the client and brief, the limits, the output format, and where the result goes. Drop any one and you get mood words instead of a plan.
• The 36 prompts below follow a commission in order: inquiry, brief, references, shot list, the day, client review, delivery and the next job. Each one says what you get in the chat and, where it fits, what lands on a Storyflow board.
• Give ChatGPT your numbers (rates, call times, sunset, deliverable counts). A prompt that leaves a number out gets an invented one.
• Ask for one stage per prompt. "Make the moodboard and the shot list" skips the client's approval, which is the step the rest of the plan depends on.
• Through the Storyflow plugin, ChatGPT builds and reads the board, but a photo attached in the chat does not reach it, and it cannot read board comments. Ask for an upload link, and paste comments into the chat.
• Keep each client in its own ChatGPT project, so its files, instructions and chats stay together and out of the next client's work.

Why Photography Prompts Come Back Generic

You ask ChatGPT for "a moodboard for a brand shoot" and get eight adjectives and a list of things to search on Pinterest. The model is not lazy. It answered the prompt it was given, and the prompt had no client, no deliverables, no format and nowhere to put the result. Four gaps cause almost every weak answer.

No client in the prompt

"Moody, elevated, editorial" is what a model writes when it does not know who will see the pictures or where. Name the brand, the audience and the placement (a 4:5 Instagram post, a 16:9 website hero, a printed menu), and the light, palette and crop follow from them.

No deliverables, so no shape

A shoot plan is built backwards from the list of files the client is paying for. Without counts and crops, ChatGPT guesses, and its guess is usually ten nice ideas rather than 24 specific setups.

No format, so you rebuild it by hand

If you do not say "a table with these columns" or "one shot card per setup", you get paragraphs, and every paragraph is retyping work before it reaches a client.

No destination, so the answer scrolls away

The best shot list you ever got from ChatGPT is six screens up in a chat the client cannot open. A prompt that ends with where the result goes ("put it on my Saltmarsh board") is worth more than a perfect prompt that ends in the chat.

The Prompt Pattern Behind Every Prompt Here

OpenAI's own prompting guide splits a good prompt into identity, instructions, examples and context, and suggests clear sections so the model can tell them apart (OpenAI prompt engineering guide). For photographers, that becomes five parts.

You do not need all five every time. Inside a ChatGPT project, the role and context can live in the project instructions once, and each prompt carries only the constraints, the format and the destination.

How to Run These Prompts

Every prompt works in plain ChatGPT. The ones that mention a plugin need it installed first.

In plain ChatGPT

Paste the prompt, replace the bracketed parts, and copy the result where it belongs. This is fine for client emails, captions and one-off checklists.

With the Storyflow plugin

Install Storyflow from its plugin listing (select Install plugin, sign in to Storyflow, choose the projects ChatGPT may use, and allow changes if you want it to build), then start prompts with @Storyflow or pick it from the + menu, two of the ways OpenAI describes for calling a plugin (Plugins in ChatGPT). Availability depends on your ChatGPT plan, workspace and region. The full steps, and what to do when the button does not appear, are in our guide to installing plugins in ChatGPT. Plugin actions run on your ChatGPT plan and use no Storyflow AI credits.

With the business plugins

A few prompts hand off to HoneyBook, Dropbox, Calendly, Adobe or Metricool. Our roundup of the best ChatGPT plugins for photographers covers what each one can and cannot do inside ChatGPT, with prices. Set each plugin that reaches clients (proposals, invoices, shared links) to Always ask, so nothing goes out without your approval (Managing app permissions in ChatGPT).

One ChatGPT project per client

OpenAI's Projects keep related chats, files and instructions together, and project instructions override your global custom instructions inside that project; connected apps work in project chats too (Projects in ChatGPT). Make one project per client, paste the brief and your house rules into its instructions, and every prompt below gets shorter. It also keeps an embargoed product out of another client's chat.

The 36 Prompts at a Glance

Stage 1: Inquiries and Discovery

The first reply decides whether you quote the job or lose it to someone faster. These five prompts get you from inquiry to a brief you can price.

Prompt 1: Reply to an inquiry in your own voice

• Prompt: "Here are three replies I sent past clients, and a new inquiry. Write a reply in the same voice: thank them, confirm [date] is open, ask the three things I need before quoting (deliverables, usage, location) and offer two call slots. Under 150 words, no exclamation marks."
• In chat: a reply you paste into your inbox, with the three questions as a short list.
• With a plugin: "@HoneyBook Create a project for this inquiry with the date, shoot type and contact." HoneyBook's plugin creates and updates contacts and projects (HoneyBook in ChatGPT).

Prompt 2: Discovery questions for this kind of shoot

• Prompt: "I shoot [product / brand portraits / weddings]. List the 12 questions I should ask before quoting a [shoot type], grouped into deliverables, usage, logistics and approvals. Mark the four that change the price."
• In chat: a grouped list you can run the call from. Our photography client questionnaire page covers the written version.
• On the board: "...and add them as to-dos in a Discovery wall on my [client] board, so I can tick them off during the call."

Prompt 3: Turn call notes into a first brief

• Prompt: "Here are my notes from the discovery call. Turn them into: one sentence on who the pictures are for and what they must make that person do, a deliverables list with counts and crops, the fixed facts (dates, locations, talent, usage), and every open question. Do not fill gaps with guesses; list them as open questions."
• In chat: a four-part brief with an open-questions list at the end.
• On the board: "@Storyflow Put this brief in a document in a Brief wall on my [client] board."

Prompt 4: Quote options from your own rates

• Prompt: "My day rate is [X], half day [Y], retouching [Z] per image, and 12 months of paid-social usage is [W]. Build three options for this brief (lean, standard, full) as a table: deliverables, days, retouched images, usage, total. Use only my numbers; if something needs a number I did not give, ask."
• In chat: a three-row table with totals you can check.
• With a plugin: "@HoneyBook Draft a proposal from my [template] with the standard option." HoneyBook generates proposals from your templates and sends nothing until you approve. Our photography proposal page covers what the proposal itself should say.

Prompt 5: Follow up after silence

• Prompt: "The client has not replied to my proposal in six days. Write a two-line follow-up that restates the date hold, says when I release it, and asks one yes-or-no question. No guilt, no discount."
• In chat: two lines, ready to send.

Stage 2: The Brief and Creative Direction

A client approves a direction, not a brief. These prompts turn what they said into choices they can say yes to.

Prompt 6: Restate the brief to prove you listened

• Prompt: "Read this brief and restate it in five bullets using the client's own words. Then list what they did not say that I need to know, and flag anything that contradicts itself."
• In chat: five bullets plus a list of gaps and contradictions, which is often the most useful part.

Prompt 7: Three creative directions that really differ

• Prompt: "Give me three distinct creative directions for this shoot. For each: a name, the idea in one sentence, light, palette, set or location, styling, and the one risk. Make them differ in light and location, not just in mood words."
• In chat: three short blocks you can compare.
• On the board: "@Storyflow Put each direction on its own wall on my [client] board, side by side, titled with its name."

Prompt 8: A one-page treatment for the chosen direction

• Prompt: "Write a one-page treatment for direction [2]: the idea, how it looks, how it serves each deliverable, and what the day looks like. 250 words, plain language, no adjective the client cannot picture."
• In chat: a 250-word treatment.
• On the board: "@Storyflow Add it as a document in that direction's wall." For a pitch built around treatments, see how photographers pitch a shoot with ChatGPT.

Prompt 9: The usage and rights checklist

• Prompt: "List the usage questions I must settle before this shoot: media and placements, territory, term, exclusivity, and whether the client may edit or composite the images. For each, give the default I usually propose for a [regional brand] and the question to ask. This is a checklist, not legal advice."
• In chat: a checklist of five to seven questions with your defaults. ASMP's licensing guide shares terms and licenses from real jobs; one photographer's terms keep every right the invoice does not grant and grant none until full payment. Keep your own lawyer-approved wording for the license itself.
• On the board: "@Storyflow Add this checklist as to-dos in the Brief wall."

Prompt 10: Check the brief for gaps before you quote

• Prompt: "@Storyflow Read the Brief wall on my [client] board and list every deliverable without a count, a crop or a usage term, and every open question still unanswered."
• In chat: a short gap list, read from the board rather than from ChatGPT's memory of the conversation.

Stage 3: Moodboard and References

References are where most shoots are won or lost: the client says yes to pictures, and then the day has to reproduce the right part of each one. These prompts build the moodboard on a board the client can open, which our ChatGPT moodboard page covers in full.

Prompt 11: A search plan before the search

• Prompt: "Before you search, list the six facets my references must cover for this brief (light, palette, texture, pose, set, crop) and two search phrases for each that avoid stock-photo clichés."
• In chat: twelve search phrases mapped to what each reference has to prove.

Prompt 12: Put references on the board

• Prompt: "@Storyflow Search the web for 12 reference photos that match direction [2], add them to a References wall on my [client] board, and add a short note under each with its source."
• On the board: a References wall with 12 images and their sources. Twelve is the most ChatGPT can add in one request; ask again for more. Storyflow downloads each picture and keeps its own copy, so a reference survives the original page going down. Check licensing before anything from the open web goes in front of a client; our guide to moodboard images for client work covers it.

Prompt 13: Your own photos, from your phone

• Prompt: "@Storyflow Give me an upload link for the References wall on my [client] board."
• In chat: a link. Tap it, pick test shots or past work, and each picture lands on the board as it uploads. The link lasts 15 minutes and takes up to 200 photos. A photo you attach in the chat itself does not reach the board.

A fashion shoot moodboard: portrait references around one heading, locations around another, with folders for mood, materials and UGC between them.

Prompt 14: Keep and Ignore notes on every reference

• Prompt: "@Storyflow For each reference you added, put a note beside it with one Keep line (what to reproduce: the light, the crop, the texture) and one Ignore line (what is in the picture but not wanted)."
• On the board: a two-line note beside every tile. For your own uploads, write those two lines yourself; you know what each picture is there for. A reference without an Ignore line is an argument waiting to happen on set.

Prompt 15: A palette from the brand and the references

• Prompt: "@Storyflow Build a five-colour palette for direction [2] from the brand's two hex codes plus three supporting tones, place it beside the References wall, and add a one-line use for each colour (backdrop, props, wardrobe, accent, avoid)."
• On the board: a palette element next to the references it serves. Our ChatGPT colour palette page shows the same move for brand work.

Stage 4: Shot List and Coverage

A shot list is one entry per setup, built from the deliverables and the approved references, not from a list of nice ideas. These prompts make it checkable. The ChatGPT shot list page covers shot cards in more depth.

Prompt 16: Shot cards from the deliverables

• Prompt: "@Storyflow From the deliverables in the Brief and the approved references, build shot cards on my [client] board, one per setup. In each: the subject or product, crop, light, and the reference it comes from. Do not order them yet."
• On the board: a row of shot cards, each with a frame for its reference, a short description and a checklist.

Shot cards on a board: one per setup, with the frame on top, the shots required as a checklist, and the gear underneath.

Prompt 17: The coverage check

• Prompt: "@Storyflow Compare the shot cards with the deliverables list in the Brief. List any deliverable no setup produces, any crop no setup is framed for, and any setup that produces nothing on the list."
• In chat: three short lists. Run this after every change to the brief.

Prompt 18: A, B and C priorities

• Prompt: "Mark every setup A, B or C. A means a deliverable fails without it, B makes the set better, C only if we are ahead. Show the A list as a short table I can send the client for sign-off."
• On the board: "@Storyflow Add the letter to the start of each shot card's title."

Prompt 19: Wedding family formals in the fastest order

• Prompt: "Here are the family names and relationships the couple sent. Build the family-formals list in the order that moves the fewest people: largest group first, then peel off one side at a time, grandparents early. Flag anyone I have noted with mobility needs."
• On the board: "@Storyflow Put it as a table on the wedding board: group, names, who to find, notes, rows kept in shooting order."

Prompt 20: A product matrix, one row per SKU

• Prompt: "@Storyflow Make a table on my [client] board with one row per product ([9] rows): product, angles (front, back, detail), background, props, retouch notes, status. Rows in the order products arrive at the studio."
• On the board: a table the studio assistant updates during the day.

Stage 5: Logistics and the Shoot Day

The day fails on time, not on ideas. Give ChatGPT the fixed facts and make it show the slack.

Prompt 21: A timed schedule with the slack shown

• Prompt: "The day runs 08:30 to 17:30, lunch is 13:00 for 45 minutes, sunset is [time], and the talent leaves at 15:00. Order the setups: fixed windows first, then location, then lighting setup, then wide to tight. Give each block a start time and a one-line reason, and tell me the total slack."
• In chat: a schedule and a slack figure you can check against your own sense of the day.
• On the board: "@Storyflow Put the schedule in a table on a Day Plan wall, rows kept in time order."

Prompt 22: Gear by setup, then one pack list

• Prompt: "From the shot cards, list the gear each setup needs (body, lens, light, modifier, grip), then a combined pack list with quantities and the two items I should bring a spare of."
• On the board: "@Storyflow Add the pack list as to-dos on a Gear wall."

Prompt 23: A location recce checklist

• Prompt: "I am scouting [location] for [shoot]. Give me a recce checklist: light direction by hour, power, parking and load-in, noise, permits, toilets, phone signal, and the five photos I must take for the client."
• On the board: "@Storyflow Add it as to-dos on a Location wall, and give me an upload link for my recce photos." The ChatGPT location scouting page covers a multi-location version.

Prompt 24: A styling brief for the talent

• Prompt: "Write a styling brief for the talent: three outfits per look, colours to avoid against the palette, fabrics that read well under soft light, what to bring and what to arrive wearing. Under 200 words."
• On the board: "@Storyflow Add it as a document beside the palette."

Prompt 25: The night-before crew message

• Prompt: "Draft the message I send the crew the night before: call time, address, parking, first setup, wet-weather plan and the board link. Plain text, under 120 words."
• In chat: a message you send yourself. Storyflow does not send messages or call sheets; it holds the plan the message points to.

Stage 6: Client Review and the Lookbook

The review round is where a plan either gets signed off or quietly drifts. Keep the decisions on the board. The ChatGPT lookbook page covers the deck in more detail.

Prompt 26: A lookbook deck from the board

• Prompt: "@Storyflow Build a six-slide lookbook deck on my [client] board in the palette's colours: the idea, three looks (each with its name, a one-line description and two labelled picture spaces, such as 'Look 2, hero'), and a closing slide with the deliverables and dates."
• On the board: the deck's words and layout, with every picture space empty. ChatGPT cannot put photos into slides through the plugin, so you drag each approved reference from the board onto its slide in Storyflow. The deck presents fullscreen and exports as PDF; there is no PowerPoint export or speaker notes.

Prompt 27: The review email

• Prompt: "Write the email that sends the board link for review: what to look at first, the three decisions I need by [day], how to comment (they sign in to comment; viewing needs no account) and what happens if I hear nothing. Under 150 words."
• In chat: an email you send yourself.

Prompt 28: Apply the client's feedback, and log it

• Prompt: "Here are the client's comments from the board. Apply the clear ones to the shot cards, list any that conflict with the brief or with each other, and add a Decision log note: date, decision, who asked."
• On the board: updated shot cards and a Decision log. ChatGPT cannot read board comments through the plugin, so paste them into the chat.

Prompt 29: What changed this week

• Prompt: "@Storyflow What changed on my [client] board in the last week, and who changed it? Summarise it as a three-line update I can send the client."
• In chat: a change summary read from the board's history, not from memory.

Stage 7: Delivery, Follow-Up and the Next Job

Delivery is the last thing the client remembers and the first thing that brings them back. Seven prompts close the job and start the next one.

Prompt 30: The delivery email

• Prompt: "Write the delivery email: what is in the folder, the usage the license covers in one sentence, how long the link stays live, and how to ask for retouch changes. Under 140 words."
• With a plugin: "@Dropbox Create a shared link for Clients/[client]/Finals." Dropbox asks before it creates the link (Dropbox in ChatGPT).

Prompt 31: The usage renewal reminder

• Prompt: "Draft a short note for [month] reminding the client that their 12-month paid-social license ends on [date], with the renewal fee of [X] and one sentence on what renewing covers."
• In chat: a note you schedule in your own inbox.

Prompt 32: A testimonial request that gets answered

• Prompt: "Write a two-question testimonial request: what problem the shoot solved for them, and what they would tell a peer. Make it answerable in one minute, and offer to draft it from their answers for their approval."
• In chat: a four-line message.

Prompt 33: A portfolio case study from the board

• Prompt: "@Storyflow Read the Brief wall, the directions and the Decision log on my [client] board. Write a 300-word case study: the problem, the idea, the day, the result in the client's words. Leave a bracketed placeholder wherever I need a number I can verify."
• In chat: a draft for your site, with placeholders instead of invented results.

Prompt 34: Behind-the-scenes carousel pages

• Prompt: "@Storyflow Make carousel pages on my [client] board for a six-slide behind-the-scenes post: one idea per slide, a caption under 125 characters on each, and a closing slide that names the client only if the brief allows it."
• On the board: carousel pages you export and schedule elsewhere. Storyflow does not schedule or publish posts; Metricool's plugin does (Metricool in ChatGPT).

Prompt 35: A retro while you still remember

• Prompt: "Here is what went right and wrong on the day. Turn it into a retro: three things to repeat, three to change, and the one checklist item to add to my next shoot."
• On the board: "@Storyflow Add it as a note titled Retro on the board."

Prompt 36: A reusable outline for the next shoot

• Prompt: "@Storyflow List every wall, table and checklist on my [client] board as a reusable outline, with all client details removed, so I can rebuild the same structure for the next shoot in one prompt."
• In chat: an outline you save in your ChatGPT project instructions. Next time, paste it with the new brief: "Build this outline on my [new client] board."

A Prompt Chain Worked End to End: A Restaurant Menu Shoot

Okafor Studio is a fictional two-person food and drink studio. Saltmarsh, a 40-cover seafood restaurant, is relaunching its menu and needs 22 dishes, 6 drinks, 4 interiors and a team portrait, delivered as 4:5 crops for Instagram, 16:9 website heroes and 1:1 crops for delivery apps. The shoot is one Monday, 09:00 to 16:00, while the restaurant is closed. Here is the chain, with the shape of each output. The details are illustrative.

Step 1: The brief from the call (Prompt 3)

Ade pastes the call notes. ChatGPT returns the one-sentence brief ("Pictures for locals deciding where to eat on a Friday: make the food look caught this morning and the room look warm"), a deliverables list with 33 subjects and three crop sets, the fixed facts, and five open questions, including who supplies the props and whether the delivery-app crops need a plain background. Ade answers four by email and sends the brief to the Brief wall.

Step 2: Three directions, one chosen (Prompts 7 and 8)

ChatGPT proposes "Harbour Morning" (cool daylight, slate and linen, top-down), "Candlelit Counter" (warm practicals, close crops, the bar as set) and "Market Table" (a long shared table, hands in frame). Each lands on its own wall. The owner picks Harbour Morning for dishes and Candlelit Counter for drinks and interiors, which is a real answer the shot list now has to serve.

Step 3: References and palette (Prompts 12 to 15)

ChatGPT adds 12 web references to a References wall with sources. Ade's own test plates from last week's lighting check go up through the upload link. Keep and Ignore notes go beside each tile ("Keep: low side light on the fish skin. Ignore: the lemon wedge"). The palette holds slate, linen, sea green, brass and one "avoid" colour, the restaurant's old red.

Step 4: Shot cards, coverage and priorities (Prompts 16 to 18)

The board gets 33 shot cards. The coverage check finds two gaps: no setup produces a 1:1 for the drinks, and the team portrait has no crop at all. Two cards are added. Priorities mark 21 setups A, 9 B and 5 C, and the owner signs off the A list.

Step 5: The day (Prompts 21 and 25)

The schedule puts the interiors first, while the morning light comes through the front windows, then dishes in kitchen order, drinks at the bar after lunch, and the team portrait at 15:15 before service staff leave. ChatGPT reports the slack, and Ade moves two C setups to "only if ahead". The crew message goes out the night before with the board link.

Step 6: Review, lookbook and delivery (Prompts 26 to 30)

A six-slide lookbook goes to the owner before the shoot, so the room and the dishes are agreed in one place. After the edit, Ade sends the delivery email with a Dropbox link to the Finals folder, and the retro adds one checklist item for next time: confirm who brings the ice.

The chain is not one big prompt. It is ten small ones, each checked before the next, with every result on the board the client opens. Our post on the single-prompt fallacy explains why an assistant that reads the project beats a fresh prompt every time.

When the Output Is Weak: Edits That Fix It

Most bad answers trace back to one missing part of the pattern. Fix the prompt, not the answer.

Ask for the reasoning in one line

"Give each block a one-line reason" turns a schedule you have to trust into one you can argue with. The same works for priorities and directions.

Change one thing per retry

If the directions are too similar, change only the constraint ("differ in light and location"), not the whole prompt. Otherwise you cannot tell which edit fixed it.

Correct the board, not the chat

When ChatGPT gets a shot card wrong, say "change card 14 to a 1:1 crop" rather than asking for the list again. A full rebuild loses the client's edits.

Adjusting the Prompts by Kind of Shoot

The pattern stays the same; the context and the board pieces change.

What These Prompts Cannot Do

Prompts plan the work around the pictures. They do not take them, and a few limits are worth knowing before you rely on them.

• They will not know your numbers. Rates, sunset times, travel times and usage fees have to come from you or from a source you check. ChatGPT can search the web for a sunset time; confirm it before it goes on a call sheet.
• They are not legal advice. Use them to list usage questions and track agreements, and keep lawyer-approved wording for licenses and releases.
• Through the Storyflow plugin, ChatGPT cannot generate images, read board comments or set who owns a to-do. It can fill an Owner or Assignee column in a table with people already on the board; a name that matches nobody stays empty. Photos reach the board through an upload link, not as chat attachments.
• They do not cull, grade or deliver galleries. Your editor and gallery app still do that work.
• Storyflow runs in a browser. There is no native app; plan on a laptop and open the board on a phone browser on the day.

The Bottom Line

The prompts that save a photographer real hours are not clever. They are specific: the client, the deliverables, your numbers, an exact format, and a place for the answer to live. Run them one stage at a time, check each result before the next, and put the plan on a board the client opens instead of in a chat they cannot see. Start with Prompt 3 on your next inquiry, and build the rest of the chain on a Storyflow board from there.

This article was published in October 2026. Plugin features change; each mechanic here was checked against the vendor's own documentation on 28 October 2026.