working on a new process + style

made with stable diffusion : textual inversion + img2img + text prompt