All the content I make is and will be completely free on CivitAI.
But if you want to support me with a coin, then you can do it here.
Sample images was generated with Luma and AOM2
Recommended embeddings:
1. easynegative
2. bad-hands-5
3. badhandv4
Strength 0.6-0.8 is optimal.
PLEASE CHECK YOUR LORA NAME AFTER DOWNLOAD.
Male Nightborne: 1boy, henightborne, purple skin, athletic
Female Nightborne: 1girl, shenightborne, purple skin, athletic
Skin color: grey, purple, blue, dark purple, dark blue
Also you may improve it with glowing eyes, tattoo, facepaint
Description
FAQ
Comments (14)
Another great Lora! Really impressed how consistent it is out of the box and how it captures the more exotic traits of the race, namely the very unique shape of the ears and the glowing tattoos. Cheers!
Thank you for your great feedback mate! Really appreciate it!
Really love your wow series of loras ❤️
Thank you!
Any chance you'd consider going through some of the classic D&D races and doing a set of LORAS for them? You make some of the best fantasy LORAS out there man. You're BORN for this!
Hehe, thank you for your words, pal! I can't promise anything particular, but i guess i will do some dnd stuff in near future. But i really will begin with smaller things, i guess.
@sadxzero Tieflings and Aasimars seems like a piece of cake for you.
Btw how do you make your faces this good? It always in your style mine gets that Restored face ai look without using the restore
Can I ask how many images do you usually use to create your "loraas" and what model do you use? As a 3D character, it generates doubts for me. Thank you very much anyway, excellent work as always 🙌😊
Thank you! About images - it really depends. For example, when i make some specific characters lika Leon or Ada, i go with 15+ cause i don't nedd too much variability in dataset. But when it's a WoW race or something less specific, then i go with 25+ pics for females and males each. When it's a style i really try to collect as much as possible, cause it gives flexibility to lora. But there is a thing that quality over quantity. And dataset with good tagging is about 80% of success. But most time remaining part is testing lora on flexibility and prompt understanding. As for a models, i usually train on AOM2 or NovelAI. There is not much what i can tell, i still experimenting with stuff.
@sadxzero ohhhh! Thank you very much for answering, my friend 💖💖. I really appreciate the information. I thought that we needed to use many more images for the process. I'm thinking about creating Loras myself. Thank you very much. Your Loras are great. Keep up the excellent work! 🙌🙌🥰
@BlessPibe ah, no biggie, glad to help, mate) Thank you and good luck!)
Any plans to update this Lora for Pony XL?
It'd be amazing to have a ponyXL version! :)
Pony pls. ; u ;
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.