How to make AI images that look the way you imagined
Go from your first AI image to precise edits, consistent characters and product photos you can actually use. You'll also learn what you're allowed to sell and how to spot fakes.
Who it’s for: Beginners, creators and small businesses who need visuals.

Making an AI image is easy. Making the image you actually had in mind is the hard part. Type 'a dog on a beach' and you'll get something generic that looks like a stock photo. This course teaches you to describe the subject, setting, style and mood, so the picture matches your idea.
You'll make your first image, then learn to direct the light and camera like a photographer. That means using terms such as backlit, overhead and low angle. You'll match a look from a reference image without copying an artist's work. You'll change one object while keeping everything else the same, and get short, readable text onto a poster. You'll also turn a phone snap of a product into a clean catalogue photo, keeping its label and colour accurate.
Later lessons cover keeping one character consistent across a series, upscaling for print and writing a reusable style block for a brand. You'll also learn what you're allowed to sell, and why purely AI-made images may not be protected by copyright. Then you'll learn how to spot an AI image and how to label your own. The course suits beginners, creators and small businesses who need visuals.
By the end you’ll be able to…
- Write an image prompt that sets subject, style, lighting and framing
- Edit one region of an image without regenerating the whole thing
- Keep a character or product consistent across a set of images
- Check licence terms and provenance before using an image commercially
Short on time? Start with the essentials
These 4 lessons cover the most useful ideas in this course. Each one takes about a minute.
Make your first AI image in under a minute
Generate an image from a one-sentence description and ask for a variation.
Fix the generic stock-photo look
Add subject detail, setting, style and mood to turn a bland result into a specific one.
Change one thing without ruining the rest
Use a selection or a targeted edit instruction to replace one object while keeping everything else identical.
Check whether you can sell what you generated
Read a tool's terms, spot trademark and likeness risks, and explain why purely AI output may not get copyright protection.
Every lesson in this course
This course has 12 video lessons of about a minute each, from the basics to expert habits. Each one ends with a question, three takeaways and its sources.
Foundations 2 lessons
Make your first AI image in under a minuteEssential
Generate an image from a one-sentence description and ask for a variation.
“Type one sentence, get a picture. Here's exactly what to type first.”
Fix the generic stock-photo lookEssential
Add subject detail, setting, style and mood to turn a bland result into a specific one.
“'A dog in a park' vs 'a wet terrier mid-shake, golden hour, 35mm film'. Watch.”
Building skills 6 lessons
Direct the light and camera like a photographer
Use lighting, lens and angle terms such as backlit, overhead, 85mm and low angle to control mood and composition.
“Same prompt, five lighting words. Each one tells a completely different story.”
Match a look using a reference image
Upload a reference and name which qualities to borrow, such as palette, texture or framing, without copying an artist's work.
“Stop typing 'in the style of'. Show it a picture and say what to borrow.”
Change one thing without ruining the restEssential
Use a selection or a targeted edit instruction to replace one object while keeping everything else identical.
“Perfect photo, wrong mug on the table. Fix only the mug, keep everything else.”
Get readable text and simple logos
Put exact wording in quotes, keep text short, and know when to finish a logo in a vector tool.
“AI used to spell 'Bakery' as 'Bakrey'. Here's how to get the words right.”
Turn a phone snap into a catalogue photo
Place a real product photo into a styled scene while keeping its label, shape and colour accurate.
“Kitchen-table phone photo to catalogue shot. Then check the label didn't change.”
Upscale an image for print without blur
Choose the right output size, upscale cleanly, and check pixel dimensions against print size and DPI.
“Sharp on your phone, blurry on an A3 poster. The pixel maths takes 10 seconds.”
Advanced 3 lessons
Keep the same character across six scenes
Build a character sheet and reuse reference images so face, outfit and proportions stay consistent across a series.
“Scene 1: red scarf, round glasses. Scene 4: a stranger. Here's the fix.”
Check whether you can sell what you generatedEssential
Read a tool's terms, spot trademark and likeness risks, and explain why purely AI output may not get copyright protection.
“The US Copyright Office says a prompt alone doesn't make you the author.”
Spot an AI image and prove where yours came from
Check images using provenance tools, visual tells and reverse image search, and publish your own with clear disclosure.
“Extra fingers are mostly gone. These are the checks that still work.”
Expert 1 lessons
Lock your brand look into one reusable style block
Write one style block of palette, lighting, framing, reference images and banned elements that anyone on a team can reuse to get on-brand images.
“Fifty images, three people prompting, one consistent look. It starts with one block.”
Apps we recommend
These are the tools this course uses, and what each one is best for. We have no partnership with any of them.
ChatGPT
Best for: Conversational edits and readable textDescribe changes in plain English and refine the same image over several turns; strong at signs, labels and posters.
FreemiumGemini
Best for: Editing your own photos consistentlyGoogle's Nano Banana image models edit real photos and keep faces and products consistent; images carry a SynthID watermark.
FreemiumMidjourney
Best for: Distinctive artistic styleThe art director's favourite: style references, moodboards and a strong default look for illustration and concept art.
PaidMake a brand-consistent image set for a real use
Pick a real need: social posts for a small business, club flyers or visuals for a class project. Write a style block, generate 4 to 6 images that share one look, fix flaws with targeted edits, and check the tool's terms for your use. Use your own products or photos, and no real person's likeness without consent.
What you’ll make: A set of 4 to 6 on-brand images, your one-page style block, and a note of the tool, plan and licence terms you relied on.
Check your own work
- Every image matches the palette, lighting and framing written in the style block
- Any text in the images is spelled exactly as intended
- No image shows a real person, logo or artwork you lack permission to use
- The tool's terms allow your intended use on your plan, noted with the date you checked
- Each image has the pixel size needed for where it will be printed or posted
Try a ready-made prompt
Copy a better prompt for a real task, and see why it works better than the version most people type.
Further reading and viewing
We’ve chosen these articles, research papers, videos and guides from other trusted sources.
Copyright and Artificial Intelligence, Part 2: Copyrightability
The official US position on when AI-assisted images can and cannot be copyrighted.
Nano Banana image generation
Google's own prompt patterns for photoreal scenes, edits, text in images and multi-image composition.
Verify Content Credentials
Drop in any image to see whether it carries a record of how it was made or edited.
SynthID
How Google's invisible watermark for AI images works and where it can be detected.
Questions about this course
Is there a free AI image generator?
Yes. Several big assistants, including ChatGPT and Gemini, can make images on their free plans, usually with limits on how many. Specialist tools such as Midjourney focus on illustration and concept art. Free limits change often, so check each tool's current terms. The skills in this course work in any of them, because good image prompts follow the same pattern.
How do I write a good AI image prompt?
Describe four things: the subject, the setting, the style and the mood. Then add details about light and framing, such as soft morning light or a close-up from a low angle. If text should appear, put the exact words in quotes and keep them short. Rather than starting over, ask for one change at a time until it's right.
Can I sell images I make with AI?
Sometimes, but check first. Each tool's terms say what you may do with its images, and they can differ between free and paid plans. In many countries, an image made purely by AI may not be protected by copyright, so others could copy it. Your own creative edits may be protected. Avoid trademarks and real people's likenesses without permission.
How can I tell if an image is AI-generated?
Start with the context. Ask who shared it first, and use a reverse image search to find the original. Look for visual tells such as odd text or mismatched details, though these are getting rarer. Some tools add hidden marks or Content Credentials that show an AI was involved. When you publish your own AI images, label them clearly.
Do I need design skills for this course?
No. The first lesson starts with a one-sentence description, and each lesson adds one new technique. Every lesson is a short video of about a minute, followed by one question. Later lessons go further, into consistent characters and brand styles, for people who need images for work. Your first five lessons in Tau are free.
Get early access to this course
Tau opens soon. Join the waitlist and we’ll email you as soon as you can start this course.
We’ll only email you about Tau, and you can unsubscribe any time. Privacy policy.
We’ll email you as soon as Tau opens.


