Trained without a text encoder to make shadowrun dwarves on a handcrafted collection on images this Lora is more versatile than the original intent. Essentially it works to do the following things: reduce height, bulk up the body and face, gives pointy ears. With some effort you might be able to do other short fantasy races (gnomes, halflings, etc), but my success in that area has not been reliable so far.

Prompting:

1) Shadowrun dwarves: Add "dwarf, short, stocky, broad shoulders, pointy ears" to positive prompt, "furry, pointy ears, elf ears, dwarf ears, cat ears, horns, tusks" to negative. Experiment on weights depending on the model.

2) Fantasy dwarves: Add "dwarf, short, stocky, broad shoulders" to positive prompt, "pointy ears, horns, tusks" to negative. Experiment on weights depending on the model.

Do note that full body shots can be tricky to pull off. I recommend starting with my succesful prompts from examples.

General negative prompting advice: (nude, shirtless, nudity, nipples:1.5) to avoid unintended nudity. For female characters add (beard, man:1.2) too. Sometimes closes eyes on non-realistic models, tag (eyes wide open:1.7) or so to fix

Style and Models: Meant for use with Realistic Vision 1.3 and Deliberate, but seems to work with the most checkpoints I tried to various extents. All examples are straight txt2img. Here are recommended weights for checkpoints I tested:

Works well with Realistic Vision (0.7-0.9), Deliberate V2 (0.7-0.9), Dreamshaper V4 (0.5-0.8), , expmixLine_v20 (0.7-1.1), RevAnimated_v1.1 (0.7-1.1), Chilloutmix (0.8-1).

Technically works, but requires quite a bit of effort with Anything V3 and V4 (0.7-1.3) and AOM3 0.7-1.1 .

Does not work at all with RPG v4 and Dalcefo models. Works when merged into RPG v4 via Supermerger.

Settings: Works best at low to mid CFG (2.5-5), I personally prefer to pick lower end of the scale. Seems to mix with other Loras just fine.