I do not use EA and don't charge for use so please tip if you enjoy this lora, even a small amount can add up. Thank you!
--
Diamond, a personal lora I've finally decided to let out in the world. She is 100% artificial, made from a small dataset of AI images (of fake people) I've been using to get better faces out of Flux, originally with face swap at first then later on with this lora.
Why the name Diamond you ask? Because she brings a sparkle to my eye.
She's trained to be 28-32 years old or "mature" if you like. Though she will certainly do young adult and can sorta do a elderly adult, though her resemblances I trained into it start to fade off a bit.
She usually had a defined button nose, brown eyes and either dirty blonde or light brunette hair and full lips. She had a good amount of skin texture, blemishes, so on so on.
Description
Late but here she is! Krea 2 version. Trigger, tho not needed, is D14M0ND. If you want her as I intended her to be, prompt her as a 30 year old/early-30s and sometimes depending on the model you can also use middle aged, but that might push her into much older area.
The two most common trained phrases in the dataset to achieve that was "woman in her early-30s" and "photography of a caucasian woman, she's in her very early 30s"
weight 1.0 is suggested but Ive also had her works as low as 0.4, just depends on the model and if you have other loras going on.
ALL OF MY EXAMPLES HAVE BEEN UP-SCALED USING SEEDvr2 -- this often hides minor artifacts. These artifacts ARE NOT caused by the lora in almost all cases but caused by the used checkpoint model, mainly seen with int8 because people still don't make them correctly. I also used Realistic Snapshot pretty frequently because it just makes the images more interesting. Totally not needed but you'll see it often in the meta.
NO FEE BUT I WOULD LOVE SOME TIPS TO MAKE UP FOR IT!




