This is a Model of Meredith From Guardian of the Galaxy (Square Enix)and my Second Lora. hope you like it!LORA notes.Serious version / TLDR:recommend using Strength 0.7 , CFG 6 or 2 depending on Checkpoint , Careful with Pos and Neg.Real Version:Lora strength: 0.7 is comfortable.Lower and you'll miss her likeness a bit but better QualityHigher and you'll get better likeness with Slight Blur and Artifacting--CFG:Results are Sharper around CFG 6Lower CFG though will give you better result for IluSamples: samples are okaySampling: Used Euler A Karras - you can Experiment--Positive Negative:For Best Outcome use your Own Neg (copy from the images if you need wink)Some Pos can conflict so careful.Update: While This Hold True for Other checkpoints, It fell apart like Nostradamus During Hasaku TestingUpdate 2: the Lora is alive, I repeat, It's Alive. nah kidding.It just so happens to work better without NEG and Idk why.--Extra Notes:- Training Data had a Lot of Orange Values. so Counter that with other lora or using NEG and POS.- Training Data was focused Solely on face for the most part.my advice would be to use LOW lora. make the pic you want (no likeness)Then use High Lora during Img2Img (now you get likeness on top of your pic choice)- Your Resolution can Also Affect this so feel free to Tweak it as you see fit--Think that's it.if you find anything else, comment down below.I HOPE YOU ENJOY IT TY TYPs: who you wanna see more of next? see you next time on Ilu Delulu

Meredith Quill

LORA notes. Serious version / TLDR: recommend using Strength 0.7 , CFG 6 or 2 depending on Checkpoint , Careful with Pos and Neg.Real Version:Lora strength: 0.7 is comfortable.Lower and you'll miss her likeness a bit but better QualityHigher and you'll get better likeness with Slight Blur and Artifacting--CFG: Results are Sharper around CFG 6Lower CFG though will give you better result for IluSamples: samples are okaySampling: Used Euler A Karras - you can Experiment--Positive Negative:For Best Outcome use your Own Neg (copy from the images if you need wink) Some Pos can conflict so careful.Update: While This Hold True for Other checkpoints, It fell apart like Nostradamus During Hasaku TestingUpdate 2: the Lora is alive, I repeat, It's Alive. nah kidding.It just so happens to work better without NEG and Idk why.--Extra Notes:- Training Data had a Lot of Orange Values. so Counter that with other lora or using NEG and POS.- Training Data was focused Solely on face for the most part.my advice would be to use LOW lora. make the pic you want (no likeness)Then use High Lora during Img2Img (now you get likeness on top of your pic choice)- Your Resolution can Also Affect this so feel free to Tweak it as you see fit--Think that's it.if you find anything else, comment down below.I HOPE YOU ENJOY IT TY TYPs: who you wanna see more of next? see you next time on Ilu Delulu

Files with SHA256: 5381d9f5a7d3284710a1f512cf6434d16d8317998f6447debfd63d5afa3cd361