Stable diffusion - anyone a heavy user?

Im playing with it atm using controlnet and hitting a brick wall and the GPTs are also failing me. I’m trying to get it generate an image around the black spaces and for the outcome to be closer to the first photo group of images than the last one

Prompt:
(Design a calligraphy artwork that fills only the white spaces surrounding the machine overlay’s black shapes:1.2). (Ensure the black shapes are completely untouched:1.4). The piece should exude a sense of harmony and balance between the written word and the geometric precision of the overlay. Avoid modifying, filling, or obscuring the black shapes with any calligraphic elements. Keep the integrity of the black circles, lines, and squares pristine. Do not place any calligraphy within the black areas; they should remain as negative space

Ask gemini what they think?

Dont think it can do the spatial positioning, stable is still top dog ;P. gpt admitted it dint have the same advance features as stable :smiley: