This model was trained at 512px x 640px. Consequently it rarely 'doubles up' when using the popular 512px x 768px format.<hr />A Kohya-trained LoRA, using 130 highly-curated reference images, of American actress Raquel Welch, who rose to her greatest prominence in the 1960s, going on to become the premier sex symbol of the 1970s, culminating in a pictorial for Playboy in 1979.This replaces the previous checkpoint that I did for this actress, which is far less versatile and accurate.The CFG range is quite forgiving - you can slide it a fair way along before bad artifacts ruin the effect. Works well at a wide range of steps as well.I found that using only the best and richest color material for this personality produced a pure cheesecake generator with limited versatility, so I went looking for more spontaneous images, which were harder to find than I would have thought.Anyway, since there is very little color reportage of Raquel Welch in this heyday period, a great deal of the training data was colorized at <a target="_blank" rel="ugc" href="http://Palette.fm">Palette.fm</a> and by various other means. Colorization is never the ideal solution, but with vintage stars, there's often no alternative. It doesn't appear to affect the model too badly.I generally use these models in complex workflows, inpainting faces after initial T2Is, and using ControlNet extensively. So if you're hoping for one-click prompt magic, my models aren't data-curated with this in mind, but rather as tools for traditional workflows that use Photoshop and other older methods.This model is very likely to produce NSFW and nude renderings unless counter-prompted.

Raquel Welch (1975-80 era - LoRA)

Trained on Kohya_ss for two hours, on an RTX 3090 with 24GB of VRAM. Six epochs, 26,400 steps, using 130 reference images and 2,200 regularization images (including repeats, there were only 1980 available). Extensively manually labeled, with extra WD14 labels created in Kohya, and then manually curated in BooruDatasetTagManager.

Files with SHA256: d570854b0566a29fe6da2561be1de2c3e63b827c76b32fc0fcdfe5b55889c122