Bad Apple!! reimagined as a space themed AI animation.
This project was created with Python, Stable Diffusion XL, and ControlNet Canny.
The original video was extracted into 6,573 individual frames. Each frame was processed separately at a resolution of 1024×768, then composited back into a 480×360 video at 30 FPS while preserving the original audio.
To improve temporal consistency, consecutive frames were generated using identical random seeds for both the foreground and the generated space background. This significantly reduces flickering and helps maintain visual consistency throughout the animation.
In total, more than 13,000 AI generated images were created during the production of this video.
The entire generation and rendering process took approximately 30 hours on an NVIDIA GeForce RTX 4060 Laptop GPU.
This project is an experimental proof of concept exploring AI assisted video generation.
The original two color animation is remarkably well suited for 360p thanks to its simple, high contrast visuals. A space themed reinterpretation contains significantly more fine detail, making the same resolution a much greater limitation.
While the visual quality is still far from perfect, I plan to revisit this project in the future using improved AI models, higher resolutions, and more advanced workflows.
Tools used:
• Python
• Stable Diffusion XL
• ControlNet Canny
• FFmpeg
Original reference:
https://www.youtube.com/watch?v=FtutLA63Cp8
If you're interested in the technical side of the project, feel free to leave a comment. I'd be happy to answer questions about the workflow or the implementation.
This project was created with Python, Stable Diffusion XL, and ControlNet Canny.
The original video was extracted into 6,573 individual frames. Each frame was processed separately at a resolution of 1024×768, then composited back into a 480×360 video at 30 FPS while preserving the original audio.
To improve temporal consistency, consecutive frames were generated using identical random seeds for both the foreground and the generated space background. This significantly reduces flickering and helps maintain visual consistency throughout the animation.
In total, more than 13,000 AI generated images were created during the production of this video.
The entire generation and rendering process took approximately 30 hours on an NVIDIA GeForce RTX 4060 Laptop GPU.
This project is an experimental proof of concept exploring AI assisted video generation.
The original two color animation is remarkably well suited for 360p thanks to its simple, high contrast visuals. A space themed reinterpretation contains significantly more fine detail, making the same resolution a much greater limitation.
While the visual quality is still far from perfect, I plan to revisit this project in the future using improved AI models, higher resolutions, and more advanced workflows.
Tools used:
• Python
• Stable Diffusion XL
• ControlNet Canny
• FFmpeg
Original reference:
https://www.youtube.com/watch?v=FtutLA63Cp8
If you're interested in the technical side of the project, feel free to leave a comment. I'd be happy to answer questions about the workflow or the implementation.
- Категория
- AI Картинки
Выполните вход или зарегистрируйтесь для отправки комментария.
Комментариев нет.
