You're right, I mixed some things up. I checked again only the MP4-Files are over 6Mbit/s; all other files are max. 2Mbit/s. My bad. TGO used 6-10 Mbit/s; which was not really good for the delivered quality either.
Nevertheless, NLT uses VP8 (which is bad enough) with a poor compression algorithm, then there's the low resolution and the icing on the cake is the low bitrate.
AI video enhancement concepts
At its core, AI video enhancement and upscaling involve using machine learning algorithms to improve video quality by increasing resolution, reducing noise, and enhancing details. Unlike traditional methods that rely on interpolation techniques, AI approaches employ neural networks trained on vast datasets to predict and generate high-resolution images from low-resolution inputs. The prediction-algorithms are getting better and better every day. In the near future the compute-power of an Nvidia AI-Server can increase by a factor of 100.000 - 1.000.000. Which you can use to make faster or better images.
Limitations of AI-Upscaling
- Artifacts and inconsistencies
AI video enhancer can sometimes introduce visual artifacts, such as unnatural textures or halo effects around edges. These artifacts arise because the AI is making educated guesses, which might not always align with the original content's intent.
- Processing time and computational Resources
High-quality AI video enhancement is computationally intensive. Processing even a few minutes of footage can take hours on standard hardware, necessitating high-end GPUs or cloud-based solutions, which can be costly especially for private enhusiasts.
- Overfitting issues
It usually happens when AI models are trained on specific datasets might not perform well on content that differs significantly from their training data.