I don't like canon Akane. She is probably the worst of the Ranma waifus except for Azusa Shiratori. Then again while i was collecting the Ranma-chan dataset I found some delicious images by Mage which inspired me to complete this. Busty Thicc Akane is a lot more tolerable than canon Akane. LOCON Version also available.
V2:
As I said before V1 was an experimental Lora, now with a busty thick Akane at my disposal my opinion of her rose a few notches, not too much but enough to revisit this LORA and fix it to modern standards(It honestly just pissed me off to have to add short hair to my v1 prompt, Customizing triggers simply make prompting stupidly easier.).
works fine with: <lora:TendouAkaneV2:0.7>, tendouakane, <outfit>
Available outfits are:
furinkanuniform: Her normal furinkan uniform
AkaneYellowMicrodress: Micro dress from Nihao my concubine.
ovadress: dress of the second part of Nihao my concubine.
AkaneYellowSwimsuit: Yellow swimsuit from Nihao my concubine.
Honestly, Nihao my concubine. seems to have the best outfits for the ranma cast. All other of Akane's outfits i found were boring(non ero nun level boring) sundresses, well and the fetish version of her yellow swimsuit with her wearing a sailor fuku top(Dafuq character animators? that went too far even in the fanservice nonsense department.). But well the dataset was getting too big as is to add even more crap.
Anyway as always I will add the tags summary with the dataset if you want to experiment more.
V1:
Also as I was less invested in this LORA, I was able to experiment more with tags and dataset processing. It came out surprisingly clean... Miracles exist, this was able to not crap at weight 1. Also seems the waifudifussion taggers gives more concrete results than deepbooru.
Might be a good idea to add short_hair and blue_hair for consistency.
It seems stable on all the range but I liked the results more at .6 or .7
Got Good results with: TendouAkane, big_breasts, short_hair<lora:TendouAkane:.6>,
I added triggers for 3 common outfits:
AkaneYellowdress
furinkanuniform
ovadress for this one you may want to add flower_on_head as it is not always displayed
I will add the tags summary with the dataset if you want to experiment more.