As promised, this is a much more thoroughly trained LORA for Foot-Focused images and it's pretty versatile in its use cases

I recommend weights 0.6 - 1.0 for Txt2Img and Img2Img generations, but it's pretty flexible and you can get away with using smaller smaller weights if it begins to interfere/overpower character LORAs used with it. I use a CFG scale of 7, and for inpainting uses specifically, use denoising strengths between (0.4 - 0.65) to help maintain nail polish color and avoid distorting feet. Also for inpainting, you'll want to make sure 'masked only' is set and use the tag "close-up" for better consistency.

This LORA was trained on 300 images varying from full-body to feet close-ups to make it able to do both normal generations and inpainting fixes.

Recommended Keywords:

Trigger Word: MizarFeet

mizarfeet, barefoot, feet, foot focus, toes, nail polish, toenail polish

• Specify 'soles', 'legs up', 'feet up' etc. as desired

• Specify 'close-up' when inpainting. Inpaint feet individually with 'masked only' set

• Trained on the following nail polish colors

  • Black

  • Blue/Aqua

  • White

  • Purple

  • Red/Pink

** Note: You'll have better success with nail polish color by manually adding color to nails via Adobe PS/MS Paint or some art application, then inpainting over the foot at low denoise strengths (as seen in images 1-3)

Since this LORA is really consistent in both inpainting and creating base generations, the original Mizar's - Feet Inpainting LORA isn't really necessary for use with this and will likely be archived. As always thanks for supporting, Buzz tips are appreciated so I can train other anatomy LORAs for my own works, and check out my Patreon if you feel like supporting :)