Your First Pro Image: GPT Image & Nano Banana

What you will walk away with

By the end of this lesson you will have generated and conversationally edited a social-post image in two free tools — GPT Image inside ChatGPT and Nano Banana inside Gemini — talking to them in plain language ("make the background warmer", "give her a red jacket") instead of learning any software. You'll know which tool to reach for and own one reusable prompt that turns any concept into a post image with three swappable variables.

Watch the class

Primary video: ChatGPT Image Generation for Beginners

• URL: https://www.youtube.com/watch?v=KkLj5gzZff0
• Duration: TBD
• Status: PENDING EMBED VERIFICATION (oEmbed check, embed permission, youtube-nocookie, no regional restrictions — per video-first policy checklist)

Assigned fallback: Nano Banana Pro — Beginner Guide 2026

• URL: https://www.youtube.com/watch?v=ZCw325FiS78
• Note: channel unconfirmed — verify before publishing.

Watch it once end to end, then use the manual below to execute at your own pace — you can do everything here without re-watching.

The concept: why "conversational editing" changes everything

You don't "design" these images — you describe them in a sentence, then refine by chatting. Older tools made you write one long perfect prompt and start from zero if it was wrong. The 2026 tools here work like a conversation: you generate once, then say "keep her face, change the jacket to red", and the tool edits that image instead of inventing a new one. You correct in plain language, the way you'd direct a designer — which is why beginners now get professional results. Both tools have real free tiers (Verified: Jul 2026).

That decides which tool fits which job:

• GPT Image (in ChatGPT) is most natural if you already write captions there, and handles readable on-image text well.
• Nano Banana (Google's model in Gemini) is strongest at keeping a subject consistent and editing a photo you upload — exactly what a content series needs (deep-dive in 2.2).

Neither is "better" — keep both open and pick per task. The durable skill is the prompt structure and the follow-up steering, not any one tool's name or button positions; if a label has moved, search the same words in the app.

Step by step: generate and edit one post image, in both tools

You need: a free ChatGPT account (https://chatgpt.com), a free Google/Gemini account (https://gemini.google.com), and one concept to make an image of (use a real post idea from your pillars in lesson 0.2).

Part A — GPT Image in ChatGPT

1. Open ChatGPT and start a new chat. In the message box, click the "+" button and choose "Create image" — or simply type your request; ChatGPT routes image prompts to GPT Image automatically.
2. Paste Prompt 1 below (filled in) and send it. Wait for the image to render.
3. Edit by talking. In the same chat, reply with one change at a time: "Make it a 9
   vertical." · "Warm up the lighting." · "Move the headline text to the top third." Each reply edits the same image.
4. To fix one region, click the image to open it, use the "Select" brush to paint over the area, and describe only that change ("replace this sign with a blank one"). This is inpainting — it leaves the rest untouched.
5. When happy, open the image and click the download icon (down-arrow). Save at the largest size offered.

Part B — Nano Banana in Gemini

1. Open Gemini (https://gemini.google.com). In the prompt box, type your image request directly, or click the "+" / image icon to upload a photo first if you want to edit an existing image.
2. Paste the same filled-in Prompt 1 and send. Gemini generates with Nano Banana.
3. Edit conversationally: "Keep the subject identical, change the background to a studio gradient." · "Make three variations of the color palette." Nano Banana holds the subject steady while you change everything around it.
4. Download: click the image, then the download icon.

Part C — Compare and choose

Put the two results side by side. Which nailed the subject, handled the text better, feels on-brand? Note which tool won for this kind of image — that's your default next time.

Prompts (copy and use)

Prompt 1 — Social Post Image Generator

Purpose: turn any concept into a ready-to-post image by swapping three variables — style, subject, and format. When to use: Step 2 in both Part A and Part B; reuse it for every post image. Works in ChatGPT (GPT Image) and Gemini (Nano Banana).

Create a social media post image.

SUBJECT: [SUBJECT — what's in the frame, described plainly]
STYLE: [STYLE — e.g. clean flat illustration / warm film photo / bold 3D render]
FORMAT: [FORMAT — aspect ratio + where text goes, e.g. "9:16 vertical, leave the
top third empty for a headline"]

Requirements:
- Strong single focal point; uncluttered background.
- Leave clear negative space where text will go; do not add lorem/placeholder text
  unless I ask for specific words.
- Colors and mood consistent with the STYLE.
- High contrast so it reads on a small phone screen.

After you generate it, ask me what one thing I'd like to change next.

Example (filled in): [SUBJECT] = "a ceramic coffee cup with steam, on a linen cloth" · [STYLE] = "warm film photo, soft morning light" · [FORMAT] = "9

Variations: ask for "3 style options of the same subject" to compare looks; add "square 1

Warnings: on-image text is where AI still slips — read every letter. Never upload a photo of someone who hasn't agreed, or paste confidential/client material into a tool you don't control. A realistic depiction of a real person or event may owe an AI disclosure when you post — covered in Module M8.

Common errors and how to fix them

Exercise

Take one concept from your content pillars and make three versions of it across the two tools:

1. Run Prompt 1 in ChatGPT (GPT Image). Refine it in two follow-up turns.
2. Run the same filled-in prompt in Gemini (Nano Banana). Refine it in two turns.
3. Make a third version by pushing the style hard in whichever tool you preferred ("same subject, but bold 3D render").
4. Download all three and put them side by side.

Expected result: three saved images of one concept — at least one from each tool — each with a clear focal point, deliberate space for text, and the correct aspect ratio for where you'd post it. Plus one sentence of judgment: which tool you'd default to for this kind of image, and why. Keep the winner — it becomes raw material for the character work in 2.2 and, later, an image-to-video clip in Module M3.

Final checklist

• Generated the concept in both GPT Image and Nano Banana
• Edited conversationally ("keep the subject, change X") — not by restarting
• Used Prompt 1 with all three variables filled: [SUBJECT] [STYLE] [FORMAT]
• Aspect ratio matches where you'd post it (9
  / 1
  )
• On-image text (if any) read letter-by-letter and correct
• Single clear focal point; background uncluttered
• Three versions saved at the largest size offered
• Noted your default tool for this kind of image

Toolkit for this lesson

Required

Also useful

Prices and free-tier limits change fast — treat them as dated benchmarks, not permanent facts.

Next step

Lesson 2.2 — "Consistent Character: Your Visual Identity with AI": one great image is easy; making the same character or aesthetic appear across a whole series is the skill that turns scattered posts into a recognizable brand. You'll use reference workflows in Nano Banana and Midjourney to lock it in.