The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Yesterday OpenAI released GPT‑Image‑2, and the generative graphics race just sped up again.
The new model finally handles text inside images properly. Posters, banners, UI mockups — the letters no longer “swim,” and both Cyrillic and hieroglyphs are readable. For anyone building interfaces and promo for a global audience, that’s a meaningful step forward.
GPT‑Image‑2 handles complex scenes better, keeps dense layouts from falling apart, and can generate series of related images. It can also pull fresh content from the web before generation, so app screenshots and interfaces look like they do now, not like they did a year ago.
At IZUM.STUDY we’re currently preparing a new large course on front‑end layout. We needed realistic studio shots for promo and covers, so I decided to test GPT‑Image‑2 in a real scenario. The image from the model is attached to this post — just look at the level of detail. The plasticky AI look and glossy lighting are gone; the frame feels much closer to a real studio.
It’s clear the bar for generative visuals has gone up again. The only question now is who and how will integrate these capabilities into real workflows.
How do you plan to use GPT‑Image‑2: interfaces, promo, storyboards? Share your ideas in the comments 👇
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
You've built the email. Layout is done. Copy is approved. Then someone notices the product image background clashes with the design.
Now you're opening Photoshop. Re-exporting. Re-uploading. Waiting for the designer if it's not your strength. All for a fix that should take 30 seconds.
Stripo just added AI image editing directly inside the editor. You select an image in your layout, describe what needs to change in a short text prompt, and the image updates in place. No tool switching. No re-uploading.
A few things it handles well:
• Background removal
Product photo has a messy background. Select "remove background" from the preset options and get a clean cutout you can place on any color.
• Banner color adjustments
Campaign palette changed last minute. Update the banner's dominant colors without redesigning it from scratch.
• Text translation in images
Sending to multiple regions. Change the text language inside the image directly, no need to create separate versions in a design tool.
The key thing for freelancers: these fixes usually happen at the worst possible moment. The email is done, the deadline is close, and one visual detail is off. Being able to handle it without leaving the editor or pinging a designer saves the send.
It works best for quick, single-step changes. Not a replacement for full design work, but for last-minute production fixes it removes a real bottleneck.
Full walkthrough with examples: https://stripo.email/blog/improve-email-visuals-with-stripo-ai-quick-image-edits-inside-the-editor/
What's the last-minute image fix that's cost you the most time on a project?
Yeah... nostalgic for when quality work meant a human put thoughtful time into it :)
The Build
For the Flora technique challenge, I made a jewelry product shot generator. You give it a rough product image (bad lighting, messy background; doesn't matter) and an optional short theme description. It returns editorial-quality shots.
How It Works
You upload a product image along with a short description of the theme or style you want; that's all the setup needed. From there, the AI selects the most fitting photography concepts from a 110-idea library and builds multiple tailored prompts around your input. The result is a set of distinct visual variations covering different compositions, lighting styles, and scene setups, ready to use directly or refine further.
The Revisit
I came back to it today to see if it still holds up. It does. Tests across different pieces came out consistently strong; not pixel-perfect, and the product can shift slightly, but for the time and credits spent, the quality is hard to beat.
The Ring Worry
I was worried about rings going in. AI usually botches hands and fingers. It didn't happen here. Hand poses came out natural, no weird gestures.
READ THE FULL CASE STUDY HERE
Trending
Runway
AI video generation is exploding. What are you dreaming up in Runway?
Contra University
Learn from expert creatives how to earn more using next-gen AI tools.
creativeaiflow
Creative AI workflows are evolving. What tools do you use, and what are their strengths and weaknesses?
portfolioreview
The best portfolios tell a story, not just show a grid. Share yours for feedback.
freelancerlife
Freelancer life is wins, pivots, and everything in between. What’s yours right now?