Verified: 10 months ago
SafeTensor
Type | |
Stats | 327 |
Reviews | (30) |
Published | Mar 22, 2024 |
Base Model | |
Hash | AutoV2 5EA5210648 |
v1.0
This version is finetuned based on the stable diffusion xl v1.0. It can generate realistic landscape images with very simple prompts.
Recommended sampling scheduler: DPM++ 2M SDE Karras, steps: 20~30, cfg: 2~5.
v2.0
Using more data for finetuning. This version improves the quality of the generated image, especially in contrast and satruation. We encode some "magic words" when training, which will further improve the quality of generated images.
"Landscape Photography" for landscape
"City Photography" for city scene
"Space Photography" for space or universe scene
"Auto Photography" for vehicle
Recommended sampling scheduler: DPM++ 2M SDE Karras, cfg 3~7.
Have fun with it!