I should have realized that the unique, crocheted dress they had her wear would cause problems. I will use Qwen this weekend to edit the dress and remove her skull hairpin.
“No imagery of any Person of Interest was used in creating this LoRA.”
Please be patient. The Civitai system has flagged a few images, including the video, for review. /Sigh
Warning: I suffer from insomnia and often pass the time creating AI art. As I wait for my medication to kick in, my concentration lapses and my prompts become worse. My apologies for that. On the upside, it does create some interesting images. :)
If you enjoy my work, please click the thumbs up. Thank you! /Bow
>>> UPLOADING/SHARING MY MODELS OUTSIDE CIVITAI IS STRICTLY PROHIBITED* <<<
I considered dropping the "Synthetic partner" from the title, but people have many misconceptions about those who own them.
For instance, in many cases, it's to help with the loneliness epidemic many men face today. Coming home and seeing someone waiting for you, even a synthetic one, can make a world of difference in a person's mental state.
The genesis of the synthetic partner series: My cousin requested that I create a lora of his synthetic partner, Ariel.
Generally, we Japanese people are very accepting. As a result, it is not considered taboo to see someone in public taking photos with their synthetic partner.
I hope you enjoy her. /bow.
Description
FAQ
Comments (4)
I really like your characters Loras. But, please, can you start making for Wan 2.2?
You don't need to train for wan 2.2, with H and L noise. Just use AI toolkit, and select wan 2.1 video 14b, it works perfectly with wan 2.2 and training only takes 3h.
I tested one training for wan 2.2 using this technique, and it works like a charm, my 2.1 work perfectly with wan 2.2.
Question: Many of the videos I have watched regarding training a wan 2.1 lora state that 3,000 steps are too many. What do you recommend?
@KelevraQuakenstein I used 3000 steps, with 42 images, I could use less image like 10-15. But it was for one of my own personal daz studio futanari girl character of one of my 3DX futanari comics, so need image of her, naked with penis in many angle in addition to the normal image of her clothed closeup, medium and full body shoot, side, front, back and profil shot.
It took me 3h for the training, and i tested my Lora with Wan 2.2 it worked like a charm.
It was my very first try on training a Lora, and it works insanely good.
Here is an example video of her:
https://sendvid.com/9fwd5ln7
@KelevraQuakenstein I used default setting in AI toolkit. Just selected wan 2.1 14B