Thanks for checking out the project. I've created a new set of releases (which include "gamejam" in the filename. These point to an internet host, which through a series of SSH tunnels and Docker containers arrives at temporary server running `generate-glb` (a separate Open Source project I wrote last week to help support this Game Jam). This host is running the origin non-quantized `llama-mesh` model, which I've found has better results than Ollama.
There are a few caveats however:
- The original `llama-mesh` model is ~16 GB and the GPU on my temporary server only has 12 GB, so some of the model is offloaded to the system RAM. This makes everything run extremely slow.
- The UI of Genchanted needs some improvements to indicate when a connection has been established, progress during generation, and so on. I'm planning to add that before the voting ends, but "real life" work has caught up this week after devoting time to the Game Jam last week.
- It may literally take 2-3 minutes before you start to see the vectors and faces preview in the UI. The original effect was the semi-transparent cube would appear in the scene and slide into place as it was generated. But since generation is so slow on this host you might have to look above or behind you to spot it when it first appears
I'll follow up again once the improved UI is released, and meanwhile will try to keep an eye on the redirects in case anything in the chain falls over.