A personal Merge for Illustrious which merges various Lora and a couple of other illustrious models together to a style I like.
I do not authorise it to be used on any other sites other than civitai and tensor to which I have uploaded it
V2 Release!
A new merge based on Illustrious V2, VAE is included and fixed Loras being broken like they were in V1
Why killer polar bear? I was listening to Limp Bizkit at the time of upload 😂
Im sorry I dont remember the percentages or actual merge weights of the lora/models.
Baked VAE as per multiple requests
Note: So after some additional testing and a shout out to creedyfox for this ; it seems me baking the style in to my liking has killed the ease in which style lor are applied, they will still apply, but the weight needs to be 1.2 - 1.5 (whoops)
Description
FAQ
Comments (9)
No comments? Mores the pity. The model does what it says on the tin and turns in some fantastic results out of the box. The model has been trained well with capability for really dynamic poses and interesting camera angles, it's why I started using WAI's model. Polar Bear does not disappoint and provides a great base to work from even at low steps. I appreciate that it's not full gas anime either, so you can get a good diversity of results.
That's very kind of you to say, as I was merging the models and lora in, at each phase I was testing each iteration a host of times before moving on to the next phase. There was alot of trial and error with the percentages of the merges but after about 7 iterations of merges I was happy to call it complete - was only released cos some folk asked for it
I am curious about the training process and method of this model, can you explain it? THX
Well, for this one i think I started with Illustrious base, then i merged in (i think) about 25% of Mature Ritual using A1111 Checkpoint Merger, then merged in some lora using Koyha to get the style i liked - the lora merging killed its ability to easily accept a lora though, I used 3 or 4 style lora
@tyreal2012Â What type of Loras did you use for V1?
@TTTTT55Â Just seen this (yay Online Safety Act)
Honestly I cannot remember, I will see if I can find my original merge data to let you know but I'm not hopeful sorry :)
why i use this model produce image ,progress 100% is change to black image inevitable。。。。
That makes no sense? I don't understand what you are trying to say
Hello !
I don't know how you did it, but you created an excellent model ! I've tested a lot of different IL models, and this one really best represents the prompts i write ^^
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.