Well, it's not really a one-click magic. The answer will depend on what do you already know about SD.
I'm using
You must be registered to see the links
for Stable Diffusion,
You must be registered to see the links
checkpoint as a model and
You must be registered to see the links
extension for Automatic 1111.
I used img2img combined with ControlNet ("depth" preprocessor and module) to capture basic poses from the render. When I got more or less acceptable image I used inpainting with different prompts for specific parts of the image to refine details like fingers, faces and so on. Male genitals are a separate pain because there are no really good (or at least acceptable) models trained to draw dicks. You can try to use
You must be registered to see the links
or just draw them yourself in photoshop and hope for the best.
Other parameters like sampling method (euler a), steps (28), cfg scale (5), seed (random)) are not that important because of using Controlnet to control poses and inpainting for details.
Positive prompt for a general composition (again, it probably won't give you similar results without Controlnet):
(40yo curve woman:1.2), hand hugging, black parted bangs long hair, wet hair
(boy grab boob:1.2)
nude, correct anatomy, correct proportions, eyes open
(looking away to the side:1.6), (embarrassed laughing:1.2)
(detailed hands, perfect fingers:1.3)
intricately detailed, fine details, hyperdetailed, intricate, absurd details
Negative prompt:
lowers, jpeg artifacts, low quality, (emotionless:1.2), twisted anatomy, deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy, extra limb, extra arm, extra hand, extra finger, missing limb, floating limbs, disconnected limbs, mutation, mutated, ugly, disgusting, blurry,
