Basic key :
capelet
frills shirt > use emphasis " summer clothes "
striped dress ( opsional )
ribbon trim
I don't know why now, the results generated from training understand more about the concept of clothing but when it's in automatic diffusion it's very different.
Description
FAQ
Comments (3)
Dauymmm, this looks clean. Would love to know what settings you use, will try out the LoRA later
Emm..
I just divided the folders, first the character lora folder in which there are 3 log, image, and model folders, then in the image I created another folder because there are 18 sets of my photos so I named the folder 100_name character
the number 100 is because the photos that I will practice are 18 if under 15 photos I usually do the calculation of the number of photos divided by 1500 I don't know why I just follow the tutorial, for parameters I leave the basics from kohyaa like batch 6 etc. it's just that for the previous DDIM sample I changed it to Eulr a the rest just added details to the caption and prompt.
When generating and looking for results, I just do it as usual, there are no other settings, I just add some negative TI
and Vae that I like clear Vae
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.