How to Plan Your Own Brand Photo Shoot With AI

Most people plan a brand shoot by saving forty Pinterest images, buying an outfit the week before, and showing up hoping the photographer has a plan.

Then they get 200 photos back, use four of them, and book another shoot eight months later.

The shoot is not usually the problem. The plan is. And planning is the part AI is genuinely excellent at, because most of the work is asking yourself questions you would not think to ask.

This is how to plan your own, whether you have hired a photographer, roped in a friend with a good camera, or you are doing it yourself with a tripod and a window.

Step 1: Start from where the photos are going

Everyone starts with what they want to look like. Start instead with where these images have to work, because that decides everything else.

Make the list. Actually write it out.

Website hero, which usually needs you off to one side with empty space for text
About page, which wants something warmer and less posed
Instagram grid and stories, which are vertical and crop brutally
LinkedIn headshot, which is a small circle, so anything busy disappears
A sales page or a lead magnet cover
Podcast or speaker one-sheet, if you pitch yourself anywhere
Email header, press, a book jacket, whatever is real for you

Then have AI turn that list into requirements:

Here is everywhere I'll use these photos: [paste your list]. For each one, tell me the orientation, how much empty space the image needs around me and on which side, and whether I should be looking at the camera or away. Then give me the smallest set of distinct shots that covers all of it without repeating.

What comes back is a shot count. That number is the actual scope of your shoot, and it is almost always smaller than people fear and more varied than what they would have gotten.

The most common expensive mistake in brand photography is 200 versions of one photo.

Step 2: Let AI interview you instead of staring at Pinterest

You cannot brief a shoot you have not thought about, and staring at other people's photos is a way of avoiding the thinking.

So get interviewed. Dictate this if you can, because talking gets you further than typing here:

I'm planning a brand photo shoot for my business. Act as a creative director and ask me 15 questions, one at a time, that you'd ask before planning it. Cover what I sell, who buys it, how I want to come across, what I've hated about past photos of myself, and where these images will be used. Wait for my answer before asking the next one. Don't summarize until the end.

One at a time matters. A wall of fifteen questions gets you fifteen short answers. One at a time gets you talking, and the useful material shows up around question nine when you stop performing.

Pay attention to the question about past photos. "I always look stiff," "I hate my smile in these," "everyone shoots me like a CEO and I'm not that" are the most actionable things you will say all day, and they are the things nobody asks you.

At the end, ask for the summary and keep it. That document is your brief.

Step 3: Have AI make the calls you would otherwise guess at

Looks, wardrobe, and location are three decisions people make badly at 11pm the night before. Each one is a single prompt, and the reason to run them through AI is that you already gave it your interview in step 2, so it is deciding with context instead of guessing.

Looks. A look is one outfit plus one setting. Changing your earrings is not a look, and every real change costs thirty to forty-five minutes of your shooting day.

Based on my answers, propose four looks for a one-day shoot. For each one, tell me who it's for, where it'll be used, the wardrobe direction, the setting, and the feeling. Make them different enough that nobody scrolling my website could mistake one for another. Then tell me which to cut if the day runs short.

Wardrobe. Do not ask what to wear. Ask what will fail.

For each look, give me specific wardrobe guidance, and tell me what could go wrong on camera and why. I'm [your coloring, build, and what you actually feel good in]. I want to come across as [the words from my interview]. Flag anything that photographs badly, anything that behaves differently seated, and anything I should steam or tailor first.

Location. Describe the room honestly and let it tell you what the camera will see.

I'm shooting at [location]. It has [the windows, the light, the size, the wall color]. What will this look like on camera at [time of day]? What should I watch out for, and what would improve it without renting anything?

Run all three, then photograph your shortlisted outfits on your phone, standing and seated. That is the one step AI cannot do for you, and it is the one that catches the collar that will not sit flat.

Step 4: Have it write the shot list

The shot list is the difference between a productive day and a nice hangout with a camera.

Turn these four looks into a shot list for a six-hour day, with time blocked per look including changes. For each look, list the specific frames: wide, medium, close, seated, standing, working, hands, and any props. Note which frames need horizontal space for text. Flag anything that needs a second person.

Print it. Bring it. Hand a copy to whoever is shooting.

A photographer who gets a shot list will do better work, because you have removed the part where they guess what you want and you nod politely.

Step 5: Generate the entire shoot before you shoot it

Here is the part that changes how the day goes.

Before anyone picks up a camera, generate the whole plan as images. Not to use them. To find out what you got wrong while it is still free to change.

Build yourself as a consistent character. Most AI image tools have a feature that holds one person across every image, usually called a character. I use Magnific, which lets you tap into multiple models. Upload ten to twenty reference photos of yourself, varied angles and lighting, and create the character. Every image after that is recognizably you rather than a stranger with your haircut.

Connect the tool to Claude if you can. Copying prompts into a browser tab and downloading files one at a time works, and it is about ninety rounds of paste, wait, download, rename, drag. Connecting the image tool over MCP (Model Context Protocol, the standard plug that lets an AI tool talk directly to another app) means you describe the shots in the same conversation where you planned them, and the files come back already named and filed. The MCP starter guide covers that setup.

Describe photographs, not vibes. This is where most people lose. "Confident, warm, professional" produces the exact stock photo you were avoiding. Describe the frame: how close, what lens feeling, where the light comes from, the specific clothes, the setting, what the face is doing.

Using my character, generate the shot list below. Three variations per frame. File them into folders by look name. Skin should read natural rather than retouched. Expression natural rather than neutral.

Two settings worth locking, because you will not notice them going wrong until the whole set is done:

Natural, not neutral. Neutral expressions read as passport photos.
Say no to glassy skin. Image models drift toward an airbrushed look on their own. Push back in the prompt and keep pushing.

Then walk away. It runs while you do something else, which is the whole point.

Step 6: Review it looking for what is wrong

Go through the generated set hunting for problems, not admiring it. You are trying to spend all of your objections now instead of on the day.

Useful questions to ask yourself:

If I only had time for three looks, which one goes?
Does this outfit still work at full length, or only cropped?
Is any of this just my last shoot again?
What is missing that I will want in six months and will not have?
Would I actually put this on my homepage?

Change the plan. Generate again. Two or three rounds is normal and each one costs you minutes instead of a rebooking.

This is also the point where you find out that the outfit you were most excited about looks like everything else you own.

Step 7: Put it on one page you actually bring

Schedule, looks, wardrobe per look, location notes, the shot list, and the generated previews. One document.

I build these as a simple web page rather than a deck, because everybody opens things on their phone and a deck on a phone is a pinch-and-zoom exercise. A page scrolls. Same information, and it gets read.

If you are working with a photographer, send it three days ahead. If you are shooting yourself, put it on a tablet and prop it up.

If you do this for clients

Everything above works the same, with two changes.

Replace the self-interview in step 2 by having Claude read every meeting note you have with them and pull out what they have already told you about how they want to look. Half the brief is usually sitting in notes nobody has reopened since the call.

And get written permission before you build a client as a character in an AI tool. They should know their likeness is going into it, what it is used for, and that the previews are internal planning rather than anything published. One sentence in the project agreement prevents the worst version of that conversation.

The mistakes that cost the most

Booking before planning. The date is the last decision, not the first.
Four variations of one look. Range beats volume every time.
Deciding wardrobe the night before. This is the single most common reason people dislike their photos.
No shot list. You will remember the shots you wanted about a week later.
Treating the AI previews as final images. They are planning tools with AI hands and AI text. They exist so you can decide, not publish.
Skipping the review. Generating previews and not interrogating them is the same as not making them.

Generate the shoot before the shoot. Look at it properly. Change your mind while changing your mind is still free.