author: asdrabael
baseModel: Hunyuan Video
hashes:
AutoV1: 201E416E
AutoV2: 177CE4EA30
AutoV3: 3C773595A217
BLAKE3: 7AF38A2652F0CCB30C802F9B65145ABD88792FE20C90A67D8E3A57283AC26211
CRC32: 368B829C
SHA256: 177CE4EA302224F32D85ABB50DCCAB8806A14A18E68006E636D61C5B3C5702EB
metadata:
format: SafeTensor
modelPage: https://civitai.com/models/1112858?modelVersionId=1285032
preview:
- >-
https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/10e5ad58-c9e2-4408-bbe0-bc6bc9b8091d/width=450/51720036.jpeg
website: Civitai
Trigger Words
doggystyle
About this version
Same as Version 2 but changed to the correct format to work for everyone.
Doggystyle from the side
Go to v2 for the prompts used in the other videos.
This is my updated version of this model. I used longer clips, so the motion works much better. A good example prompt is:
a side-view angle of a woman with long light-brown hair on her hands and knee's on a bed being fucked doggystyle. She has one hand on the bed, and the other on her pillow with her back sharply arched as her breasts jiggle and sway beneath her as the man thrusts into her. Her hair is covering her face, and her entire body is rocking rapidly. The man is only visible from the waist up and he is gripping her waist.
I also recommend you make the image slightly wider, like 512x384 or 768x512 and so on. If you do it square, it changes the pose to where you see from the front more. If you make it taller than wide, it changes the angle a little bit again so I guess it makes it like 3 lora's in one where you just change the dimensions to get different outputs. I also recommend that you keep the weight between 0.9-1.4. If you go lower it gets weird.