Pretty good results
Been testing out a bunch of Turbo LoRA from DrBaph, silverOxide, FeiHou and now this.
This LoRA at about 0.7-0.8 is a little (15%-20%) better at liquid clarity, motion, reference adherence, facial expressiveness compared to the above. At 1.0 it causes the visuals to look greasy and over baked.
For those that haven't tried it yet, don't over look this. It works even with other hybrid checkpoints like smhfacct's Minimax-H3-fl2va-ref2va-hybrid models.
Been testing out a bunch of Turbo LoRA from DrBaph, silverOxide, FeiHou and now this.
This LoRA at about 0.7-0.8 is a little (15%-20%) better at liquid clarity, motion, reference adherence, facial expressiveness compared to the above. At 1.0 it causes the visuals to look greasy and over baked.
For those that haven't tried it yet, don't over look this. It works even with other hybrid checkpoints like smhfacct's Minimax-H3-fl2va-ref2va-hybrid models.
what samplers combo did you try?
Been testing out a bunch of Turbo LoRA from DrBaph, silverOxide, FeiHou and now this.
This LoRA at about 0.7-0.8 is a little (15%-20%) better at liquid clarity, motion, reference adherence, facial expressiveness compared to the above. At 1.0 it causes the visuals to look greasy and over baked.
For those that haven't tried it yet, don't over look this. It works even with other hybrid checkpoints like smhfacct's Minimax-H3-fl2va-ref2va-hybrid models.
what samplers combo did you try?
res_multistep \ simple
Been testing out a bunch of Turbo LoRA from DrBaph, silverOxide, FeiHou and now this.
This LoRA at about 0.7-0.8 is a little (15%-20%) better at liquid clarity, motion, reference adherence, facial expressiveness compared to the above. At 1.0 it causes the visuals to look greasy and over baked.
For those that haven't tried it yet, don't over look this. It works even with other hybrid checkpoints like smhfacct's Minimax-H3-fl2va-ref2va-hybrid models.
what samplers combo did you try?
res_multistep \ simple
Have you mixed any of the lora's together? I saw someone was mixing them and getting good results. From what they say.
wow, no lies found, it works pretty well, it rivals silver's and that EMA600
wow, no lies found, it works pretty well, it rivals silver's and that EMA600
I actually even think its better than those. At least for me. Its literally the best lora I found to work
i'm still testing, i'm checking if mixing it with cwb from silver (the best i tried) on low helps, but still alone in 7-8 does so good
i'm still testing, i'm checking if mixing it with cwb from silver (the best i tried) on low helps, but still alone in 7-8 does so good
Agreed. I tried mixing it with the others and honestly this lora by itself is the beez kneez!!! Thank you tons for bringing awareness to it.
Been testing out a bunch of Turbo LoRA from DrBaph, silverOxide, FeiHou and now this.
This LoRA at about 0.7-0.8 is a little (15%-20%) better at liquid clarity, motion, reference adherence, facial expressiveness compared to the above. At 1.0 it causes the visuals to look greasy and over baked.
For those that haven't tried it yet, don't over look this. It works even with other hybrid checkpoints like smhfacct's Minimax-H3-fl2va-ref2va-hybrid models.
Howdy... so I'm thinking that I may have interpreted your statement previously differently. Are you talking about the lora's strength set to .7-.8 or the resolution? because at 1 mp resolution size I get that greasy over baked, but at .7 resolution with lora strength at 1.0 I see amazing results. Is that what you were referring to?
Been testing out a bunch of Turbo LoRA from DrBaph, silverOxide, FeiHou and now this.
This LoRA at about 0.7-0.8 is a little (15%-20%) better at liquid clarity, motion, reference adherence, facial expressiveness compared to the above. At 1.0 it causes the visuals to look greasy and over baked.
For those that haven't tried it yet, don't over look this. It works even with other hybrid checkpoints like smhfacct's Minimax-H3-fl2va-ref2va-hybrid models.
Howdy... so I'm thinking that I may have interpreted your statement previously differently. Are you talking about the lora's strength set to .7-.8 or the resolution? because at 1 mp resolution size I get that greasy over baked, but at .7 resolution with lora strength at 1.0 I see amazing results. Is that what you were referring to?
hopefully they say strenght, some turbo loras overcook results after 0.8 strenght but if you get good results it's fine
wow, no lies found, it works pretty well, it rivals silver's and that EMA600
It is partly built off of that, just added the 4 step reference deltas fully which is why the file is so big, and some more of the raw larry turbo one as well. I didn't make or train it, it's a merge.
wow, no lies found, it works pretty well, it rivals silver's and that EMA600
Which one？，give me a link,thx!
wow, no lies found, it works pretty well, it rivals silver's and that EMA600
Which one？，give me a link,thx!
Bro.... where are you commenting at right now?
May I ask if you are using the 10Eros_Max_h3_hybrid_beta5 model along with this LoRA?