I’ve been working on a tool called Creative Engine that tries to solve that problem with a visual, graph-based workflow instead of a text-only form. You type a single theme (e.g. “cyberpunk”), and the system expands it into a semantic map of related concepts:
- Subject: richer variants like “cyberpunk cyborg cat”, “interstellar wanderer”, etc.
- Vibe: moods such as “psychedelic neon”, “cinematic cold tone”, “dreamy soft glow”.
- Details & composition: fine-grained texture, perspective, framing options.
- Styles: integrated with a style library (e.g. “Ghibli-esque hand-drawn”, “Unreal Engine 5 render”).
You can then click through this “idea map” to pick the nodes you like, and an intelligent combination engine turns your selections into multiple, fully formatted prompts with different emphases. The goal is to make the process feel more like “visual brainstorming” and less like memorizing prompt recipes.
- The UI is built around an interactive canvas that animates how your concept branches into related nodes, so you can literally watch your idea “grow”.
- Under the hood it uses graph-like structures and semantic associations to keep the expansions coherent instead of random tag spam.
- Power users can export the whole inspiration map (image/JSON) and reuse prompts across other tools or pipelines.
If you often get stuck at the “what should I actually type?” stage of AI art, this might be useful as a thinking aid or as a way to systematize prompt engineering for a team.
You can try it here (free, but desktop, login required): https://z-image.me/en/prompt-engineer
I’d love feedback on:
Whether the “graph of ideas” UX actually feels better than a big textarea.
What extra dimensions (lighting, camera, storytelling, etc.) you’d want to see in the node network.
Any obvious “gotchas” if you were to integrate something like this into your own creative or production workflow.
No comments yet.
