Loona / Hellhounds style lora for realistic images of Hellhounds of course include Loona too ;-)
for the image Gallery i used my model Sleipnir for the gallery also to use the workflow for it https://civarchive.com/models/241631/workflow-for-sleipnir-sdxl-turbo-hybrid-v2?modelVersionId=272623
pick the Lora workflow
i recommend to use a strength between 0.5-0.8 for this model but it's up to you
for bifröst project i recommend using 0.8
Note; this Lora is very skimpy if recommend for sfw pictures putting sfw in negative and describe clothing also reducing the lora weight can help
V1: First Version
V2: changed the training method and did use my Bifröst Project model for training instead of the base model i recommend also for using that with the Lora
more details
more versatile
more effects like example wet fur
Disclaimer; You are responsible for your own creations
PS: will report every sick photo that is posted here that is against the law!
This disclaimer is not here without reasons. please have in mind this is like a virtual showroom and posting extreme stuff here is like 💩 on my desk in real live 😉
Description
First Version created in first place to test new hardware for training upcoming models
(more a tech demo)
FAQ
Comments (7)
Sometimes I have a problem: black and red stains on the object. How could I avoid it?
depends on the used model and sampler i guess
had not the problem with the model sleipnir so far and the new added workflow i build it for testing new hardware and training methods i will implement next year in upcoming models .
another way is get the Lora down to 0.5 but it depends on the model used
@freek22 That's right. Tried it on 10 other models, only works on some of them. Changing scheduler, sampler - no result. Reducing the weight of LoRA reduces the probability, but worsens the quality of the wool. I will try again)
Try upping the steps, that sometimes works for me on other LoRAs
i can also recommend using the bifröst project model on tensor.art at 0.6 strength
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.