I've been using GPT Image 2 all week — here's what it's actually good for
The latest with GPT Image 2 and implementing into Seedance 2.0 workflows
Welcome back to Visually AI!
Today’s reading time is 4 minutes.
In today’s edition:
GPT Image 2 Launch
Storyboard to Animation: a real workflow using GPT Image 2 + Seedance 2.0
Cyperpunk RPG Workflow
Image & Video Prompts
GPT Image 2 Launch
OpenAI released GPT Image 2 on April 21. For most, it has been a big step up; and I would agree.
Here are the main changes:
Text rendering — finally works. Multilingual, inside the image, near-accurate. If you’ve ever needed a sign, label, poster, or UI element rendered in the image itself, this is the first time it’s reliably usable.
Instruction following — it actually does what you say. Complex layouts, specific compositions, precise edits. Less fighting the model to get what you described.
4K resolution — high-fidelity output for the first time via the API. Useful if you’re building content for large-format or print.
Reasoning before rendering — the model plans before it generates. The result is more consistent outputs on complex prompts.
Side note: DALL-E 2 and DALL-E 3 are both being retired on May 12, 2026. If you’re using those via the API, you’ll need to migrate.
For visual creators, the text rendering alone has been a massive boost. Mockups, branded visuals, title cards, and graphic design-adjacent work are now firmly in reach.
Storyboard to Animation
This is something I’ve been testing out over the past few days, and the results are worth sharing. You may have seen this on social media.
The core idea: use GPT Image 2 to generate your storyboard frame first including locked compositions, consistent style, defined scene logic, audio. Then add the image to Seedance 2.0 to bring them to life.
The workflow:
Build your storyboard in GPT Image 2. Lay out your full sequence as a single image. Panel order, composition, lighting. Think like a director: establishing shot, mid shot, close-up.
Feed the storyboard into Seedance 2.0 as your single reference image.
Write your motion prompt to animate storyboard as a continuous piece.
The result:
Here is a second result:
Psst, a new Image Model may be coming soon!
Cyberpunk RPG Workflow
Here is another workflow you can use based off the previous one. My goal here was to build a cyberpunk RPG with the loadout as my real AI stack.
The concept: what if a content creator’s workflow looked like a video game character select screen? Tool cards with stats. Aesthetic presets. Finished with a world reveal when the session starts.
Workflow:
Nano Banana Pro → stylized 3D character render from a real reference photo (x3)
GPT Image 2 → four game UI screens (start menu, tool loadout, preset selector, in-world reveal)
Seedance 2.0 → animated clips with UI interactions and fade transitions
DaVinci Resolve → sequenced and edited into a single piece
Full result:
🖼️ Image Prompts
Prompt: Generate a vibrant anti-inflammatory smoothie campaign poster featuring four tall glasses arranged in a diagonal line. From left to right: golden turmeric mango lassi, deep green spinach spirulina blend, bright pink beet ginger shot in a small glass, and creamy purple açaí blueberry smoothie. Each glass has visible condensation droplets, fresh ingredient garnishes on the rim, and a small floating label beneath it showing key benefits in clean sans-serif type: “Anti-inflammatory,” “Gut health,” “Energy boost,” “Antioxidant rich.” Background is a soft gradient from warm sunrise gold to cool mint. Top header reads “DRINK YOUR MEDICINE” in bold condensed uppercase. Bottom strip shows a clean icon row: no added sugar, plant-based, cold-pressed, gluten-free. Bright natural backlighting creating translucent glow through the liquids, commercial beverage photography, high-end wellness brand aesthetic, landscape 16:9 format.
Prompt: A high-end fashion editorial shot on 120mm medium format film. A hyper-realistic, monumentally oversized bright orange nylon puffer jacket, the size of a two-story building, is draped casually over a brutalist concrete bus stop in a bleak, overcast Icelandic landscape. The jacket’s material features intricate micro-wrinkles, matte reflections, and heavy, realistic sagging under its own impossible weight. A single human figure in a sharp, tailored black suit stands beneath the bus stop shelter, providing stark scale contrast. The lighting is diffused, soft, and cinematic, casting deep, realistic shadows in the folds of the nylon. Muted color palette with the orange serving as the sole vibrant focal point. Film grain, subtle halation, hyper-detailed textures.
🎥 Video Prompts (Tool Comparison)
Prompt: Woman at tactical command vehicle console grips joystick, eating ration bar while screens flicker with tactical data. City lights streak past windows at speed. Locked-off wide interior shot, slight camera vibration from road. Warm amber panel glow against cool exterior blur.
🎨 Explore my Portfolio
I’ve had some amazing opportunities to work on a variety of projects recently, and I don’t share everything in this newsletter or online. And you know I love testing tools to share.
Take a look at my portfolio to get a quick glimpse of my work:
🚀 My Recent Top AI Tools Picks
Thank you for reading.
I hope you have a creative week!
Heather Cooper