Been sitting on a side project idea for weeks and kept getting stuck at the same wall: I know what I want the final graphic, product shot or storyboard frame to look like in my head, but I don't want to spend two hours in an editor just to test whether the idea even works visually. Anyone else run into this?
What's worked for me is treating the first pass as disposable. Instead of opening a full design tool and committing to layers, fonts and exact placement, I write the scene out in plain language first - subject, setting, mood, maybe one reference object I want kept consistent across a few variations. Then I generate rough versions fast, look at five or six side by side, and only then decide which direction is worth the real time investment. Saves me from falling in love with an idea that looks bad the second it is actually rendered.
For that rough-draft stage something like Nano Banana 2 Lite fits - a third-party AI image site (not something official from Google or DeepMind, just built on that kind of tech) aimed at throwing prompts and reference objects at it to get quick concept variations for mockups, ads or storyboard panels. Useful for exploring direction, not for final output.
Limitation worth knowing: none of these tools nail small text, exact brand colours or fine product details on the first try, so don't skip your actual design pass - this is for deciding WHAT to build, not building it. Treat it like a sketchpad, not a print shop.
Curious whether anyone else has a workflow for killing bad ideas before they eat a whole weekend.
