Hi, I'm Steven, an indie developer. I've been working on image upscaling tools for years. Video was something I wanted to tackle next, but it turned out to be much harder than I expected.
The main challenge wasn't just running super-resolution on individual frames. Processing time, GPU memory and OOM issues, tiling, decoding, encoding, and handling larger video files all introduced their own engineering problems.
I ended up building VideoUpscaler.com with two different workflows. For smaller and lighter videos, you can upscale them locally in your browser without uploading the files. For more demanding videos, you can upload them for cloud GPU processing and produce higher-resolution output without needing to own a dedicated and expensive GPU.
I'm still actively experimenting with different models and processing pipelines, especially trying to find a good balance between processing speed and output quality.
I'd be interested in feedback from anyone working with video or ML infrastructure, or simply anyone who has some difficult old, compressed, or low-resolution footage to test.
Hi, I'm Steven, an indie developer. I've been working on image upscaling tools for years. Video was something I wanted to tackle next, but it turned out to be much harder than I expected.
The main challenge wasn't just running super-resolution on individual frames. Processing time, GPU memory and OOM issues, tiling, decoding, encoding, and handling larger video files all introduced their own engineering problems.
I ended up building VideoUpscaler.com with two different workflows. For smaller and lighter videos, you can upscale them locally in your browser without uploading the files. For more demanding videos, you can upload them for cloud GPU processing and produce higher-resolution output without needing to own a dedicated and expensive GPU.
I'm still actively experimenting with different models and processing pipelines, especially trying to find a good balance between processing speed and output quality.
I'd be interested in feedback from anyone working with video or ML infrastructure, or simply anyone who has some difficult old, compressed, or low-resolution footage to test.
[flagged]