Final, in solo pose point of view, it's difficult to get more fancy pose...
Re-arrange files to different folders:
lying, 40
metalstocks, 22
NSFW_all_fours, 17
NSFW_Kneeling, 35
NSFW_lying, 47
NSFW_sitting, 77 (=99-22)
NSFW_split_leg, 32
NSFW_squatting, 36
NSFW_Standing, 178
NSFW_Suspended, 41
V1.1 and after
Just download newest version, it will include all previous files.
New added poses:
metalstocks, 5 poses, 4/7
lying, 41 poses, 4/7
V1.0
The NSFW poses are created by my wildcard model.
NSFW Pose Wildcard For Depth Map Creation,2024/02/02.
Content:
Landscape mode, 768*512, 109 poses
Portrait mode, 512*768, 107 poses
File format:
All poses includes depth map, openpose bone structure and .json file.
No example image included.
File directory:
NSFW_512768, portrait mode (512*768)
NSFW_768512, landscape mode (768*512)
Prompt:
Pose: standing/sitting/kneeling/squatting/lying/(shibari,bondage,chained,suspended from the ceiling)/(shibari,bondage,standing on one leg)/(shibari,bondage)/all fours
Leg: spread legs/split leg/one leg up
Arms: arms stretched out/grabbing her own breasts/arms behind back/arms over head
Camera Angle: from below/from above/breast focus/pussy focus
-->All poses were made by the prompt matrix above, and I believe the depth map should give you the idea about necessary keyword.
ControlNet setting:
OpenPose (weight 1) + Depth (weight 0.5)
Note:
My file directory is designed for batch run of controlnet, and is not convenient for preview.
Due to the file name is organized, you can put all files into a file path, then depth map could be kind of showcase for openpose bone structure and .json file.
Description
Add 46 poses, total 262 poses.
FAQ
Comments (4)
V1.1, for metalstocks pose, in case your checkpoint don't work, lora "Joschek's Metal Bondage Stocks" shown in suggested resources is suggested.
Thank you for your excellent data.
I found your wildcard for pose too. Is it necessary to use this data of depth or openpose? No?
I made openpose/depth file from the images which prompt shown in the wildcard. When you use openpose/depth file, you still need to describe (prompt) it, and don't need to use original prompt. For example, some pose looks like "lying on bed" for me, maybe it's "sitting on chair" in your opinion. If you try both of them, maybe both the 2 prompts working.
If something show up in the image, the prompt is necessary, too. For example, cuffs,rope,baseball,boxing gloves...etc.