Catherine Élise Blanchett is an Australian-American actor and film producer. Lilith is a character from the Borderlands series.
SDXL Recommendations:
Checkpoints: SocaRealism XL
Pony Recommendations:
Checkpoints: NostraMix, NostraRealisticMix.
try use "wearing gloves" in negative. Hands with gloves badly trained,ruins hands comletely.
Description
FAQ
Comments (7)
Cate Blanchett Galadriel would be awesome
Also want a Galadriel version
Going make a request for Jamie Lee Curtis's Tannis.
Since so far you're the only one who is making anything from this film.
Other request:
Annalee Call - Alien Resurrection
Lydia Deetz - Beetlejuice Beetlejuice (2024)
Ripley 8 (Outfit) - Alien Resurrection for the Ellen Ripley model since it's same actress with a different outfit
Maybe add "freckles" in the kim pines lora's trigger words for her light freckles in the film if you haven't yet.
Apparently the lora for kim pines was only made with 13 images, if you need more images here:
https://fancaps.net/movies/MovieImages.php?name=Scott_Pilgrim_vs_the_World&movieid=607
https://scottpilgrim.fandom.com/wiki/Kim_Pine/Gallery#Scott_Pilgrim_vs._the_World
"Since so far you're the only one who is making anything from this film."
It's fun to interpret this in another way lol
I mean it wasn't the most well receive films of this year.
@yume99
Not interested in Jamie Lee Curtis's from borderlands.
I think i will make lora with Winona Ryder in future.
I made Kim Pine lora from best images what i can find. I still experimenting with refinementing pictures, but result is unpredictable, most cases blury pictures causes bad results, and training checkpoint unpredictably “adapts” the images, I have to remove parts of the dataset that are bad for learning. Maybe i return to this later..
Too many interesting things to do and so little time to do it...
Too many interesting things to do and so little time to do it...
That's alright
I still experimenting with refinementing pictures
I'm just surprise you mange to make the lora in 13 images and it still looks good with kim since other users say they need 50+ images for training and some made loras with 1-3 images. The lora is Impressive, i just asking for her light freckles to be tagged in the trigger words. "guy90" has a "Background Changer" script if that can help with training. If it makes it any easier, loras can be pony only
https://civitai.com/articles/5697/tools-scripts-i-made-and-use-in-my-workflow
Winona Ryder in future
Nice, still hoping for her goth mom self in Beetlejuice Beetlejuice.