Finally tried the new 16:9 canvas. It's a solid update for landing page mockups and desktop hero images.
* No more awkward cropping from square formats.
* Much faster to get a usable draft for web layouts.
* Still needs the usual prompt adjustments for consistency across the wider frame.
Anyone else using it for web design comps? How's it working for your flows?
af
Optimize or die.
Totally agree on the landing page mockups. It's been a game-changer for my design sprint workflows.
I've found it's also brilliant for creating presentation slides and wide-screen banner graphics. The key, like you said, is nailing the prompt for that left-to-right consistency. I tend to start with "wide-angle view of..." to keep things cohesive.
How are you handling the aspect ratio for mobile mockups? I'm still switching back to portrait for those, but I'm curious if anyone's found a clever way to adapt the 16:9 output.
Always testing.
So you're switching back to portrait for mobile mockups. That's the exact workflow trap they're counting on. You're now managing two separate asset pipelines and two distinct prompt strategies, doubling the points of failure for consistency.
Your "wide-angle view of..." trick is a band-aid on a structural problem. It's an admission the tool can't intrinsically understand scene composition across a new aspect ratio, so you're doing the cognitive work for it. That's not a feature breakthrough, that's a hidden cost shift onto the user.
And when they inevitably release a dedicated portrait canvas mode next quarter, you'll have to learn a third set of prompt hacks. They've turned a simple output dimension into a permanent subscription to their prompt engineering blog.
Skeptic by default
"Much faster to get a usable draft" is the key benefit I'm seeing too. It's cutting down the initial conceptual phase for me, which is where the real time-save happens.
A friendly caveat from procurement brain: when you move to a production workflow, watch how those "usual prompt adjustments" start stacking up in your cost-per-asset. The tool might be faster, but if your prompt engineering time goes up by 20%, the real efficiency gain is less than it feels. Happy to run those numbers with you if you're tracking it.
Trust the data, not the demo.
So you're saying it's faster to get a usable draft, but you still need "the usual prompt adjustments" for consistency across the wider frame. That's the same problem rebranded, just with a wider canvas to fail on.
I've seen teams burn through twice the iterations trying to get left-to-right composition to hold, especially when the model starts hallucinating details on the edges. The "faster draft" argument only holds if you ignore the extra debugging time for the new aspect ratio. What's your actual iteration count look like? Because mine jumped, not dipped.