Nice that it's quick and free. However, it's very simplistic.
Tried prompting to get a top-down view of a log split into 4 quarters and it would generate images with tree rings going the wrong direction, with a huge hole in the middle. Negative prompts could not remove the hole.
Seems to be based on a small quick LLM that might be best swapped out for a larger one. Though that might not fit the model.
Is it running the LLM clientside and that limits the size?
Nice that it's quick and free. However, it's very simplistic.
Tried prompting to get a top-down view of a log split into 4 quarters and it would generate images with tree rings going the wrong direction, with a huge hole in the middle. Negative prompts could not remove the hole.
Seems to be based on a small quick LLM that might be best swapped out for a larger one. Though that might not fit the model.
Is it running the LLM clientside and that limits the size?