Image Generation Mastery — Nano Banana, DALL-E, Imagine, Canva
Visual AI
Photoshop cost $60k of training. This class costs 2 hours.
What used to cost a studio
Product photo shoot → now free
Logo design → 60 seconds
30-second brand film → one afternoon
Sign in to all 4 right now
- 01Nano Banana (Gemini 2.5 Flash Image) — fastest, best at editing existing images
- 02DALL·E 3 (inside ChatGPT Plus) — best at text-in-image and cinematic scenes
- 03Imagine Art — pro-quality photoreal & artistic, free tier is generous
- 04Canva AI (Magic Studio) — turn any prompt into a full designed post/deck
Not all image AIs are the same
Photoreal humans, characters, product shots. Best for hero imagery, moodboards, thumbnails. Nano Banana specifically WINS at editing an image you upload — 'remove background', 'put this shirt on that person', 'make it night'.
DALL·E for anything with text (menus, posters, book covers). Canva AI for finished, designed assets — Insta carousels, ads, pitch slides, resumes. Both bake typography in so you don't fight fonts.
S · S · M · D — memorize this
'A 16-year-old girl in a Karachi rooftop café, laptop open, sipping chai…'
'Shot on Fuji X100, 35mm, cinematic, warm color grade, film grain…'
'Golden hour, hopeful, quietly determined, slightly rebellious…'
'Steam from the chai, city lights blurring, 16:9, shallow depth of field.'
The founder portrait prompt
A [AGE] Pakistani [GENDER] founder in a [SPECIFIC PLACE], [ACTION]. Style: shot on 35mm film, cinematic composition, [WARM/COOL] color grade, subtle film grain. Mood: [3 ADJECTIVES]. Details: [1 SIGNATURE PROP], shallow depth of field, 16:9. No text in image, no distorted hands.
Run in Nano Banana AND Imagine Art. Compare. Winner becomes your LinkedIn header.
Stop doing these
- 01'Make it beautiful' — beautiful means nothing to AI
- 02Stacking 20 styles ('anime + oil painting + pixar + realistic') — pick ONE
- 03Ignoring aspect ratio — always name it (1:1 Insta, 9:16 reels, 16:9 hero)
- 04Not iterating — the first output is a draft, not the answer
- 05Skipping negative prompts — 'no distorted hands, no text, no extra fingers'
Nano Banana + image upload = your unfair advantage.
Upload YOUR product / YOUR selfie / YOUR sketch, then edit with words. Anyone can generate. Very few know how to edit.
Turn your selfie into a brand hero
[UPLOAD your selfie] → 'Keep my face exactly the same. Put me in a moody cyberpunk Karachi setting, neon reflections, wearing a black hoodie. Cinematic 35mm. Same haircut, same expression.'
This is how influencers make 30 posts from 1 photo. Learn it tonight.
Canva AI is the secret weapon
Give you a raw image. You still need to lay out fonts, sizes, CTAs. Great for hero art.
You describe the FINAL asset ('IG carousel: 5 slides teaching R.A.C.E. to Pakistani students, mint & black, bold headlines'). It designs. You tweak. Ship in 3 minutes.
Live activity: 20 assets in 20 minutes
Ship a full launch pack for a fake brand.
- 01Invent a Pakistani brand (Lahori Chai Co., Karachi Cricket AI, whatever).
- 02Nano Banana → 3 product photos.
- 03DALL·E → 1 poster with the brand name IN the image.
- 04Canva AI → a 5-slide IG carousel launch.
- 05Screenshot the whole pack. Whoever's is most on-brand wins.
Three tools. Three superpowers.
Today you get hands-on with the exact stack: DALL·E (inside ChatGPT), Nano Banana, and Imagine Art. Every other image AI is a variation on these three moves.
The 'just describe it' tool
Concept art, posters with real text baked in, storybook scenes, quick 'good enough' hero images. Because it lives inside ChatGPT, you can iterate in plain conversation — 'make it warmer', 'add a Pakistani flag on the wall', 'now 9:16 for reels'.
In ChatGPT: just type your prompt. Ask for variations in-thread. Ask it to rewrite YOUR prompt before generating — 'expand this into a full DALL·E prompt with style, lighting, camera'. That single move doubles quality.
The 'text-in-image' prompt
Create a vertical 9:16 poster for a Pakistani teen tech event called 'BUILD NIGHT'. Bold sans-serif type reading 'BUILD NIGHT' at the top, subtitle 'Karachi · Friday 8pm'. Style: risograph print, mint + midnight palette, subtle grain, high contrast. The typography must be crisp and legible — no gibberish letters.
Run in ChatGPT. Ask it to redo the text if any letter looks off. That's the loop.
The 'edit anything' tool
Editing an image you already have. Change the outfit, swap the background, keep the face identical, remove a person, add product on a shelf, change day to night. Faster than DALL·E and shockingly consistent with faces and objects.
Take ONE photo you like. Edit it 10 different ways with words. Suddenly you have 10 on-brand posts from one shoot. This is how top creators make a week of content in an hour.
One selfie → 3 worlds
[Upload your selfie] Keep my face, hair, and expression exactly the same. Change ONLY: (1) background to a moody Karachi rooftop at golden hour, neon signs blurred behind; (2) outfit to a black hoodie with a subtle mint logo; (3) lighting to cinematic 35mm, warm rim light from the right. Do not distort features. 4:5 aspect ratio.
Then run it again with a different background prompt. Same face, new world. Repeat.
The 'photoreal hero shot' tool
Studio-quality photoreal images — product shots, editorial portraits, cinematic environments. When you need something that looks like a real photographer took it, Imagine wins. Generous free tier, fast.
Use Imagine for the ONE hero image per project (your website hero, your pitch cover slide, your LinkedIn banner). Use Nano Banana for the 20 variations after. Different tools for different jobs.
The founder portrait — for real
Editorial portrait, 35mm film photograph, medium shot. A 16-year-old Pakistani founder standing in a sunlit Lahore rooftop workshop, wooden table with a laptop and a chai cup, morning light through a large window, dust particles visible in the light. Warm cinematic color grade, subtle film grain, shallow depth of field, natural skin, no retouching look. 3:2 aspect ratio. No text in the image.
Ship this to Imagine Art. Compare against DALL·E and Nano Banana running the SAME prompt. Notice the difference.
How pros actually chain them
- 011. Imagine Art → generate ONE hero image you love
- 022. Nano Banana → edit that image into 10 variations (outfits, scenes, moods)
- 033. DALL·E in ChatGPT → posters, thumbnails, and anything needing legible text
- 04Never fight a tool on its weak point — switch tools instead
The real numbers
Tools cover 95% of every image job you'll ever need
Variations from one photo via Nano Banana edits
Photoshop skill required to ship pro-quality visuals
Bonus drill: Same prompt, three tools
Feel the difference between the three tools in your bones.
- 01Write ONE prompt for a hero image for your idea (use the S·S·M·D formula).
- 02Run it in DALL·E via ChatGPT, Nano Banana, and Imagine Art.
- 03Screenshot all three side by side.
- 04In one sentence: which tool won, and why?
Which of the 4 power moves (persona, step-by-step, examples, self-critique) will you use most — and on what task?
Pick ONE move to master this week.
Name the specific prompt you'll rewrite with it.
What would 'good' look like after the upgrade?
Chain-of-thought prompting works because…
You are now the art department.
Tomorrow: give your stills motion.