Hey guys, this is my first time training my own lora and this was just a test which I did
so I made a lora for Ad-nim from the manhwa love quest and the results turned out pretty good the lora was trained for the "Anima-base-V1.0" model
Test it out and let me know how it is, any feedback is valuable
Description
FAQ
Comments (12)
well done, this is good
Thank you! enjoy <3
this is a really good lora. it works with a variety of poses and clothes so your dataset and training parameters were well chosen
Thanks man, appreciate it, enjoy the loRA <3
Will you do more character from love quest? and how did you set up you dataset and how long did it take? the lora is great
I have made the lora for the other 2 characters from love quest, and yes I will be doing more characters from love quest in the future, and as for the dataset, i just took 30 good images of ad-nim from the manhwa and most of them were full body, the training took me only 20 mins approx with my rtx 3060 12gb vram model.
@kuraretto2 Thank you for replying to my question. now i have more understanding of training the lora. also so did you not create different angle of her for the dataset?
@ArtificalEverything ye , but its not a must that you have to get the images from every angle, just a few full body pics are more than enough the ai can generate the back view and the side, but if you want the images to be very accurate then yes, you would need images from every angle and, probably also proper fully nude images as well, i think what matters the most is the tagging and how well you tag, i am still experimenting and learning how to make better loras so its a learning curve, finding which kinda images to use to train the lora.
@kuraretto2 Thank you. That helps me as well to know what i should focus on.
@ArtificalEverything anytime <3
Why removing the NTR previews? :(
just wanted to keep the thumbnail different, coz it looks like civit ai takes the first post as the thumbnail, you can post your creative generations here ;)