Version 2.0 update:
after I connected the grid pattern sometimes visible in bright or dark parts of the images with the fp4 training with OneTrainer, I switched to fp8 training, which got rid of the pattern. But I have to redo all my LoRAs to fix them all. Nothing changed, except the tag is now <lora:LeaningToSide-step01705:1>.

After the series of lying pose LoRAs and the experience won, I did a rework of the leaning back LoRA and split that up. The control is simply better if you train each pose separately, because it is very different how much training each one needs. This is: "leaning to the side and supporting herself with her arm" from a sitting position. Include the <lora:LeaningToSide-100:1> tag for consistency.

There are several leg positions trained, but the dataset is not big enough to trigger them with prompting alone: a more stretched out position and the z-sit position where the legs are bent and tilted to the side. See the examples for that. It is still at the testing stage, but with a extra leg position LoRA in low strength you may be able to trigger the legs.

Something I started here but had to be placed too in separate LoRAs to work is a finer control over face angles, body angles, camera angles and image area. But starting with this one, the captions are consistent as a preparation, even if they do not (yet) work standalone. The scheme is:

Face angles:
Full face view | Three-quarter face view | Two-third face view | Profile face view

Body angles:
Front angle shot | Three-quarter angle shot | Side angle shot | Three-quarter rear angle shot

Image area:
Full-body portrait | Three-quarter-body portrait | Half-body portrait | Quarter-body portrait

It would be great if I get this to work!

Search AI Models & Addons

very few results