trigger = smg

My LoRA for American actress Sarah Michelle Gellar, which is focused on photorealism and trained on 66 images from roughly the same years that “Buffy The Vampire Slayer” originally aired (a period which also catches "Scream 2", "I Know What You Did Last Summer", etc.)

All training images are at least 1024 x 1024 pixels and were trained for 5,000 steps using DrawThings AI. (I used 55 images for the first 3,000 steps, then I decided her eyes weren't quite right, so I added 11 more images focusing on Sarah’s eyes for the final 2,000 steps).

The LoRA is was trained on the SDXL 1.0 base model, and my four sample images were generated with the LoRA on the following base models: SDXL 1.0 (looking back), sdxlYamersRealistic_v5 (closeup portrait), cinemaxAlphaSDXLCinema_alpha1 (bleak landscape) and thinkdiffusionxl_v10 (slayer in the cemetery). I usually apply this LoRA by itself at 1.0, but you may want to bump up or down for effect or based on other resources.

Please use this LoRA of a real person legally and responsibly, and enjoy!