The field of image and motion manipulation is witnessing significant advancements, driven by the development of innovative diffusion models and techniques. Researchers are focusing on improving the control and flexibility of these models, enabling more precise and intuitive editing of images and motions. A key direction is the integration of text-driven approaches, allowing for more user-friendly and accessible image manipulation. Additionally, the use of contrast information and procedural shape grammars is being explored to enhance the accuracy and realism of image generation and editing. Noteworthy papers in this area include:

• Parametric Shadow Control for Portrait Generation, which introduces a method for intuitive shadow control in portrait generation.
• MixerMDM, a learnable model composition technique for combining pre-trained text-conditioned human motion diffusion models, allowing for fine-grained control over motion dynamics.
• Pro-DG, a framework for procedurally controllable photo-realistic facade generation, combining a procedural shape grammar with diffusion-based image synthesis.