Type | |
Stats | 348 639 |
Reviews | (53) |
Published | Jun 20, 2024 |
Base Model | |
Training | Steps: 3,500 Epochs: 100 |
Usage Tips | Clip Skip: 1 Strength: 1 |
Hash | AutoV2 82F019F90D |
Gritty, damaged, out of focus old photos. No keywords needed.
No matter if you're looking for an image from a contemporary, historical, fantasy, or sci-fi scene, they're all taken with primitive photographic equipment and poorly maintained. Images generated with this LoRA will show worn edges, desaturated color, surface print scratches, and poor focus.
Not traditionally beautiful imagery, but poetic and evocative. Images intimating deeper stories.
As with all my LoRAs, training and sample images are generated with just the base SDXL 1.0 - and the aesthetic demonstrated here is by far strongest with the base. So many fine tune models work hard to steer away from "flaws" such as this LoRA embraces, that its effects can often be strongly mitigated. You may find the most you get is a shallower depth of field and mild vignette.