~Beach Ball~ While aiming to get exactly the pose I want in my AI images, I have expanded my 3D-to-AI workflow to now include a depth map! My OpenPose plugin is used in Poser to extract and export the 3D OpenPose image, and I run a 2nd pass to render a scene depth map. I then feed ControlNet with both maps (OpenPose + Depth), which allows me to place extra objects in the scene, in this case, the beach ball, with the girl kneeling over it exactly as I wanted. My prompt said there was a beach ball in the scene, and the AI looked at the depth map in ControlNet, and understood the size and placement extracted from it. This also helps disambiguating OP poses where limbs overlap with the body, like the hand over the chest, and foot over the shin. This also helps the AI to understand what side the character is facing, which can often be ambiguous when dealing with OP stick figure representations from OpenPose. One side-effect is that the depth map may impose the hairless naked 3D model I used to create the pose, but this can be easily helped by reducing the depth map weight to values as low as 10-15% in ControlNet, allowing the AI to fill in the blanks with whatever I prompt. This OP+Depth workflow allows me to add objects from my 3D Poser scene to the AI image that match the character pose, size, and placement, while removing the pesky pose and orientation ambiguities that often emerge when dealing with OP stick figures. 3D pose and camera set in Poser with my OpenPose plugin, rendered in Stable Diffusion, postwork in PaintShop.