A lora for hiding your sword from the police
throatfucking
trained on 97 images
Image info
Model used abyssorange2hard
Model used based64
Notes(Update)
v2.0 is a LORA NOT A LYCROIS
works much better and much more consistent
much more flexible( prompt for the girl to be kneeling/squatting/sitting etc)
Redid the entire dataset (available for download)
Training settings?
The most important part of training is getting a good dataset!
Training settings can be found by installing the additional networks extension in auto1111
Extensions -> available -> load from -> additional networks
Restart webui
place lora into stable-diffusion-webui\extensions\sd-webui-additional-networks\models\lora
To view the training settings
Go to additional network tab -> choose model -> training paramters
Description
FAQ
Comments (21)
Must replace sword with something else, it draws a sword
From my testing, it seems you dont need the trigger word at al. "throat fuck" seems to be sufficient to use with the lyco set to 1.
Plus, don't forget to group tags with quotes, parenthesis, or underscores instead of spaces.
(sword swallowing position:1) "( )" alone usually means more emphasis unless you specify the value
sword_swallowing_position
"sword swallowing position" WITH quotes
At least for Automatic 1111, the above will all read as the same and (from my experience at least) are interchangeable but perhaps that will vary from model to model depending on the tokenizer. Can also put "sword" in your negative prompt but sometimes the model just doesn't have the anatomical awareness or positioning to sufficiently generate a result.
why not just train with NAI as the base? training on AOM is going to cause compatibility issues
I think you misunderstood. I always train on nai, the models used section refers to the model I used to generate the images
oh sorry, i got scared that it did because it looks so good ;)
the trigger word is incorrectly written:
sword swallowing positionHello there, could you tell me please what is concept type and where should i put this file?
just download the lycrois extension and place it into stable-diffusion-webui\models\LyCORIS
@rinrinrinrinr thank you!
got it working with the lora version, i think the site i was using this lycoris on just didn't had lycoris support. great lora.
thats an issue which alot of people had *insert not on my machine meme* I have no idea what caused it so I remade it into a lora instead of a lycrois. I hope it works well now!
it works well your just a shit who cant follow simple instructions
Can confirm~ does in fact generate shit! In fact it's "Da shit!" 😏 (check gallery, just made some)
V1 was Lycoris. V2 is a Lora. Watch your prompting, Embeds, and Negative prompts. If your negative prompt is three paragraphs it's practically useless after the first two lines. Refine. Use only what you need. Don't spam out of horny frustration. 😉
Also, PLEASE remember to do ' GIT PULL' if you use stable diffusion like A1111. I used a <1.5 version and it didnt work well. Now it works in 1.5.1 and beyond.
@taiconan >If your negative prompt is three paragraphs it's practically useless after the first two lines.
this is BS btw. Thats definitely not how it works. Longer is not necessarily better but shorter is also not necessarily better.
@veryfirstquestion800 That's fair, it's just often folks are already using embeds (which have much higher token presence than the usual tags) you do eventually reach a threshold where it's doing more harm than good and so I'll often advocate for less (rather than more) negative prompting. If what I'm generating is an object, why use "bad_anatomy" Amirite? 😋
@pinkmanjessie757278 shut up dumbass, again i literally copy pasted the fucking prompts including the same model etc. i didn't even blame the lycoris either. i've been gening my stuff just fine all this time with no issues and know what i am doing so please kindly fuck off.
@mnemosynekotorin got it to work with the lora version, very consistent, thanks a lot
@taiconan yes it depends. sometimes more is less, sometimes more is more.Just make sure you do your trial and error to learn, its part of the fun tbh ^_^
@pinkmanjessie757278 what is wrong with you man. chill...