Seated Sniper Cradle

Trained on Illustrious-XL.

This lora replicates a sniper pose where the rifle is resting on the marksman's arm supported by one leg in a seated position (seated sniper cradle). This Lora focuses on the seated position but can also achieve kneeling positions.

Example Prompts:

Sitting with one leg up:

holding sniper, sniper on arm, crossed arms, arm on knee, arm on top, thumb grip, aiming, (leg on floor, one leg up:1.3), wrist grab, front view, cross-legged

Cradle:

holding sniper, sniper on arm, crossed arms, arm on knee, arm on top, sitting, cradle, wrist grab

Kneeling:

holding sniper, sniper on arm, kneeling, crossed arms, arm on knee, aiming, arm on top

Prompt Engineering:

Trigger:
- 5n1p3r_cr4dl3
Base Prompt (Should be used):
- holding sniper, sniper on arm, crossed arms
- arm on knee (not strictly needed)
Sitting Pose:
- sitting, cross-legged, leg on floor, one leg up
- sitting, cradle, cross-legged (optional)
Kneeling:
- kneeling, leg on floor, one leg up
Hand Positioning:
- arm on top, wrist grab (helps to place the holding arm on top of the trigger arm)
- thumb grip (helps to place thumb on top of the rifle)
- one hand covered (places hands on top of each other)

Usage Tips:

Strength: 0.8 - 1
Seems to work slightly better for Landscape or Square resolutions
Only works for positions where the marksman is pointing the rifle to the right side of the picture (You may need to give it some tries).
Sometimes the arm might not be positioned as intended. Try to use prompts like:
- crossed arms, arm on top, thumb grip, wrist grip
Use background and environment prompts, also describe the character for more variety
This model is limited on the variety of snipers and it likes to draw scopes, can be fixed by reducing the strength and inpainting.

Resolutions I use for good results:

896x1152 (Portrait)
1024x1024
1152x896 (Landscape)
1536x1024 (Landscape)

Usually I also apply a 2xHighResFix (0.3-0.5 denoising strength)

Future Versions:

I may will improve this lora in the future with a larger weapon variety or by supporting aiming to the left hand side of the picture. A larger dataset might also help to improve the stability.

If you are interested in any of the datasets I used to create my loras (I mainly use ai generated datasets) feel free to contact me.

I also love to see the images you create using my loras so I would appreciate any post.