A lora model with a collection of cartoon character expressions

In the higher weight will appear more reductive expression, in the lower weight will appear better generalization performance. Use about 700 images for training, training resolution 512*512

Use base mode T-anime v4 (pruned)

By Tonade

Very useful model ~

Look here!!

Primary cue word

chibi

Character cue word

cattype_1

cattype_2

cattype_3

cattype_4

nottype_1

foxtype_1

foxtype_2

foxtype_3

girltype_1

girltype_2

girltype_3

gruatype

nachonekotype

shizukutype

nekoparatype

Recommended weight

0.3-0.8

If you think it's good, like it

Figure prompt word picture display

cattype_1

cattype_2

cattype_3

cattype_4

foxtype_1

foxtype_2

foxtype_3

girltype_1

girltype_2

girltype_3

gruatype (I typed the wrong cue word when training the model)

i

nachonekotype

shizukutype

nekoparatype(v2 model update)

There are more training sets, but they are so few that I haven't labeled them, but you can still restore them if you use a lot of prompt words. If you think this model is good, remember to return the picture "like" comments ~