[ShaShekhar]

instructpix2pix is fine-tuned on sd-v1.5 which is a inpainting model (aware of contexts and semantics) that why it don't require mask.