-- created Pony v3 with an updated dataset, threw in an SDXL version while baking that with the same source. Haven't fully tested SDXL with SDXL prompting since I didn't change the data tagging approach specifically for it.. So ,Pony prompting may work better but it did seem to work well in the limited testing I did perform --With the buzz of his some of previous content out there in the wild, I thought I'd bake a Luigi luigi on Pony using my own workflow using high quality images in the dataset. I thought the results were fairly good for a relatively quick effort.With this version, ADetailer highly recommended for wide/full shots since this dataset run didn't include that many of that type.Deny. Defend. Diffusion.main trigger word: luigiman

Luigi Mangione (Pony & SDXL)

Additional HQ images to dataset - should be even more flexible with prompting.

Files with SHA256: f4c93c95886608f74aa2ef39b540c17bd5257372c15f847f18886d2e4297540c