A model for rendering images in which a giant woman is about to sit on a shrunken person. Typical prompt is as follows:

immBC, 1girl, <t1ny_man/t1ny_woman>, <pose>, <other details about characters and setting>

I used LoRA strength of 1 on everything without issue, though if you load up other LoRAs, you may need to adjust.

common poses: about_to_sit, kneeling, standing, squatting, falling, lying on_side, all_fours

Sometimes you can get multiple giantesses by replacing 1girl with 2girls and multiple_girls, but it's unreliable. Usually you just end up enforcing the shrunken person as female. Sometimes t1ny_crowd can produce a crowd of shrunken people, but with a low success rate.

You can also get views from_above and from_side, especially with a tag like (from_above:1.5) and aerial_view, but it does strongly prefer views from behind and below. I've had better luck forcing frontal views using anatomical figures in front (e.g., "breasts") than by prompting "from_front" or something like that.

Standard giantess LoRA disclaimer - if the shrunken figure is very small on the canvas, it'll probably be low-resolution and distorted. Upscaling and inpainting are your friend here. I inpainted two cases as an example (indicated). In those cases, I left the prompt from the original generation As usual, see the example images as a jumping off point.

Absolutely not trained on realistic images, but included one with some bare-bones inpainting out of curiosity.