CoQui: Quantum GAN for Coordinate-Conditioned Image Generation

A new research paper introduces CoQui, a coordinate-conditioned quantum implicit generative adversarial network (GAN) for end-to-end image generation. The paper, available on arXiv (2608.11884), addresses limitations in existing amplitude-based quantum GANs (QGANs) that encode pixel locations via computational-basis indices or address qubits, causing quantum resources to scale with image resolution. CoQui reformulates quantum image generation as coordinate-conditioned implicit function learning, taking spatial coordinates and latent variables as inputs. A classical embedding network generates input-dependent circuit parameters, and a variational quantum circuit is evaluated at each coordinate. Pixel intensities are obtained directly from the expectation value of a dedicated color qubit, avoiding probability competition among pixels. The method aims to improve precision and control in quantum image generation. The paper is authored by researchers and submitted as a cross-type announcement.