Noticing a complete lack of high-quality LoRA models dedicated specifically to the male-on-male 69 position, I took it upon myself to train a custom one using a diverse dataset of 40 unique art styles. To achieve optimal generation results, I highly recommend using horizontal landscape resolutions—such as 1216x832—or a standard square 1216x1216 ratio. For the best prompting workflow, make sure to place side view near the beginning of your prompt before the primary action. Additionally, you will get significantly cleaner anatomical results if you explicitly define the positioning of each character, clearly specifying who is on top versus who is located on the bottom.
