The Jellyfish community may be small, but it’s still very much alive. We continue to enjoy every piece of video footage that has survived over the years. Unfortunately, most of it is only available in poor quality, so I want to use the latest AI diffusion models to restore and upscale as much of it as possible.
As many of you know, I’ve been working for nearly 15 years to preserve and discover new Jellyfish material, which I’ve made available for everyone to download from the Jellyfish Archive (jellyfish-archive.de).
Unlike traditional upscaling, modern diffusion models can reconstruct realistic details, but they require enormous GPU memory and computing power. Rendering locally would take many hours for just a single minute of video, so the process currently has to rely on cloud-based GPUs, which comes with significant costs.
It’s also important to understand that not every source will produce spectacular results. The final quality always depends on how much detail is present in the original footage.
If you’d like to help preserve and improve this rare Jellyfish footage, I’d truly appreciate your support. Every contribution helps cover the cloud computing costs.
Please note that payment processing fees of approximately 2.9% will be deducted from your donation.
Below are just a few examples of what these models can do.
Updates on this project will be posted here, in the Jellyfish Facebook group, and on the Jellyfish Archive website.
