mo-di-diffusion on Hugging Face vs Drag Your GAN

mo-di-diffusion is a fine-tuned Stable Diffusion 1.5 checkpoint on Hugging Face that renders characters and scenes in a modern animated-film look. Creator nitrosocke trained it on screenshots from a popular animation studio using DreamBooth with prior-preservation loss and text-encoder training over 9,000 steps. Add the token modern disney style to your prompt to trigger the aesthetic.

General SD 1.5 checkpoints aim for photorealism or broad illustration styles. mo-di-diffusion narrows hard on that glossy character-animation look, which is why sample prompts for game heroes, animals, and landscapes all lean cinematic rather than photographic. The weights load through the standard Diffusers StableDiffusionPipeline, so you can run it locally, in Colab, or inside Hugging Face Spaces without a proprietary API.

Artists, hobbyists, and developers experimenting with character concepts use mo-di-diffusion when they want Disney-adjacent renders without commissioning custom model training. The model card ships Python sample code, links to Gradio demos, and notes ONNX, MPS, and FLAX export options. It is free to download under the CreativeML OpenRAIL-M license with commercial use allowed subject to the license harm restrictions.

In the realm of synthesizing visual content to meet users' needs, achieving precise control over pose, shape, expression, and layout of generated objects is essential. Traditional approaches to controlling generative adversarial networks (GANs) have relied on manual annotations during training or prior 3D models, often lacking the flexibility, precision, and versatility required for diverse applications.

In our research, we explore an innovative and relatively uncharted method for GAN control – the ability to "drag" specific image points to precisely reach user-defined target points in an interactive manner (as illustrated in Fig.1). This approach has led to the development of DragGAN, a novel framework comprising two core components:

Feature-Based Motion Supervision: This component guides handle points within the image toward their intended target positions through feature-based motion supervision.

Point Tracking: Leveraging discriminative GAN features, our new point tracking technique continuously localizes the position of handle points.

DragGAN empowers users to deform images with remarkable precision, enabling manipulation of the pose, shape, expression, and layout across diverse categories such as animals, cars, humans, landscapes, and more. These manipulations take place within the learned generative image manifold of a GAN, resulting in realistic outputs, even in complex scenarios like generating occluded content and deforming shapes while adhering to the object's rigidity.

Our comprehensive evaluations, encompassing both qualitative and quantitative comparisons, highlight DragGAN's superiority over existing methods in tasks related to image manipulation and point tracking. Additionally, we demonstrate its capabilities in manipulating real-world images through GAN inversion, showcasing its potential for various practical applications in the realm of visual content synthesis and control.