ad
ad

an experiment with A.I. & 3D rotation

Film & Animation


Introduction

In this article, we delve into an innovative experiment using artificial intelligence to create 3D rotating animations from single images. The experiment revolves around a software called Stable Video 3D, which operates on the principles of Stable Diffusion but enhances the capabilities to animate a static image in a 360-degree rotation.

The Initial Setup

The process begins with generating static images using Stable Diffusion. For example, one of the prompts used was "ice Statue Of An Angel," which creates a visually appealing image. After generating the image, the next step involves setting up the Comfy UI workflow to render a 360-degree rotating animation. While the initial results may not always meet expectations, they lay the groundwork for further enhancements.

Challenges Faced

Throughout the experiment, various challenges arise, particularly concerning the seed used for generation. Sometimes, a bad seed can result in suboptimal animations, making it necessary to generate multiple images to find a satisfactory output. The importance of adjusting parameters, such as the batch size and seed settings, is highlighted during this exploration.

Iterative Improvements

As the experimentation continues, adjustments are made to the prompts and parameters in Stable Diffusion to refine the results. By changing descriptive terms (e.g., from "ice" to "stone") and modifying the details of images, the renderer aims to produce more realistic representations. The concept of "open pose" is also introduced as a potential solution to improve the figure's pose within the image.

The Role of Post-Processing

Once a satisfactory animation is obtained, the article discusses how to utilize other software, such as Photoshop and After Effects, to enhance the images further. By keying out backgrounds or adjusting certain features, the overall appearance of the animated assets can be improved. Furthermore, multiple attempts to generate variations of a nun figurine lead to discovering different aesthetics and interesting outcomes.

Entering the Glitch Realm

After settling on a design, the process extends into glitch art. Using the application Resom, the generated images are manipulated to achieve unique effects, adding different layers to explore the texture and visual appeal further. This combination of AI-generated content and traditional editing results in exciting new assets suitable for creative projects.

Conclusions Drawn

The experiment demonstrates the potential of integrating artificial intelligence with creative processes in unique ways. The possibilities for creating rotating 3D animations from static images open new doors for artists and designers seeking innovative visual styles. As the technology continues to evolve, so too will the methods we use to exploit it creatively.

Keyword

  • AI
  • 3D rotation
  • Stable Video 3D
  • Stable Diffusion
  • Animation
  • Comfy UI
  • Photoshop
  • After Effects
  • Glitch Art

FAQ

What is Stable Video 3D?
Stable Video 3D is an advanced software that utilizes the principles of Stable Diffusion to animate static images into 360-degree rotating animations.

How does this experiment work?
The experiment involves generating static images using prompts in Stable Diffusion and then processing them through various AI tools and software to create animated assets.

What challenges are faced during the experiment?
Challenges include finding the right seed settings for image generation, ensuring realistic representations, and achieving the desired animation effects.

What post-processing tools are utilized?
Photoshop and After Effects are used for applying finishing touches to images, such as removing backgrounds, enhancing details, and generating glitch effects.

How can glitch art be incorporated?
Glitch art can be created by manipulating the output images using resolutions and filters, contributing to the unique visual style of the project.