I wanted to share a little side project I developed recently called Crumly. It's an app aimed at generating instant 3D visuals for design ideas, especially for those who work in physical/spatial design like architects, interior designers, and set designers.
The idea behind Crumly is to make it super easy to create quick, clear visuals from any 3D model in just a few seconds. The goal is to help communicate design ideas more effectively, especially in the early stages when it's more about exploring different concepts rather than focusing on highly detailed, polished models.
I've seen (and experienced) frustration with the clunky, time-consuming process of translating design ideas into 3D models. The tools available can really bog down the creative process and require a lot of effort, not to mention the costs involved. Plus, if the project scope changes, a lot of that work ends up going to waste.
With Crumly, I'm trying to explore whether generative systems could help make that process smoother, especially for those early conceptual phases when quick, low-fidelity visuals are needed to convey a design's direction. It's meant to complement high-end modelling tools by offering a faster, more intuitive way to generate and communicate ideas. You can develop a low-effort, low-fidelity modeling in any tool of your choice (provided it exports one of the commonly used 3D formats) and experiment with visual creativity in this AI supplemented canvas. There are some basic tools for last-minute refinements as well.
The inspiration for this came from discussions with colleagues who work in virtual environments. They currently use midjourney generations next to block models of the actual sets to explain their designs to the production team. We started thinking about how to give more control in shaping the visual output of the 3D model. It's still in the early stages, but I'm excited about the potential of the approach and keen to see where it goes.
The web-app might hit limits on the model-size, so please be gentle with the number of triangles when you are experimenting. It also doesn't help to have very complicated models with this tool.
Also images generated is public by default for now. Uploaded models are isolated and safe.
I’d love to hear your thoughts, and if anyone is interested in giving Crumly a try, feel free to reach out!
No comments yet.
