- Mandatory trigger words: wind lift, wind lifted skirt,
To use it together with character LoRAs I always recommend a range between 0.2 - 0.3
(Although I honestly have no idea how versatile this model is)
Creating images with this model and without some LoRA character model to accompany it does not give me good results... Maybe it's just me that this happens to me...
- You can use or ignore the following prompts, BUT I recommend that if you want to prevent your results from deviating from what you are looking for, you add SOME prompts you don't want to the negative prompts: showing her ass, from behind, leaning forward, bent over, all fours, cameltoe, thigh gap, cowboy shot, from front, from below, looking back, looking at viewer, ass, panties, pussy, anus, close up,
- The "close up" tag can change the result quite a bit. Keep this in mind when generating your images.
- Maybe the model is trying to dress the subject in a "school uniform", so add that to the negative instructions if that's not what you're looking for.
Help for "all fours"
- Describing the background/location can help improve image generation: on floor, on grass, on bed, wood floor, indoors, outdoors, etc, etc...
- Adding tags of different positions to the negative prompts also helps: standing, lying, sitting, etc.
Help for "leaning forward/bend over"
- Likewise, add descriptions of the environment to help generate the image (for example, if you intend to obtain a view from below, use "blue sky, clouds, etc" for a low angle) or add labels such as "balcony or table" to help the "bend over" tag.
Help for "from front"
-The "from front" doesn't work very well xD
But after a few tries you may get good results.