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.