Photo Booster XL makes your image more detailed, and photo-realistic.
SDXL 1.0 model used for training. Tensor Art trainer certified.
Please check the following settings and my sample postings for example prompts and LoRA combinations:
For checkpoint models, any realism models such as Extra Realistic(+) XL can be used for the best realistic results . For low step generations, there are also many realism models such as Atomix XL as well as other Turbo and LCM models.
If you use the LoRA together with my other Booster LoRAs such as Real Booster XL, Art Booster XL and Anime Booster XL, you can experience richer and more fun results.
Use 0.6~1.0 setting depending on your applications. Please refer to my sample images for various settings and prompts if you want some initial ideas.
VAE: can be automatic, None or vae-ft-mse-840000-ema-pruned.ckpt
Sampling method: Euler, Euler a (You can try others too.)
sampling steps 25+, cfg 7~10 (With Turbo or LCM checkpoints, it can be much lower of course)
Adetailer denoising strength: 0.3~0.5
Upscaling: do it as post-processing if needed. (4X-UltraSharp, 4X-AnimeSharp, etc.)
Thanks. Enjoy~
Any advice is appreciated!
** Please follow Tensor Art guideline about the explicit usage of celebrity images that could result in liability.
