QualityScaler 3.5: Windows setup and video upscaling

Uploaded on June 5, 2024, this Vietnamese-language tutorial covers QualityScaler 3.5 for local AI video upscaling. The presenter describes it as an open source application for Windows 10 and Windows 11, citing at least 8 GB of RAM and roughly 4 GB of graphics memory. These are the presenter’s stated requirements for the dated version, rather than minimum specifications verified in this example.

The walkthrough starts with separate input and output folders. An 11-second clip with 334 frames goes from 576 × 1024 to a stated output resolution of 1152 × 2048. The presenter explains input scaling, full or half precision, GPU selection and CPU allocation, with MP4 or AVI output. He leaves multithreading disabled for this example and discusses the interpolation setting. The source does not demonstrate increased frame rate. He reports clearer detail and stronger color in the result, while warning that repeated upscaling can introduce errors and take longer. His comments about processing speed describe this demonstration rather than a general benchmark.

The installation section shows how to clone the repository, enter its folder, create and activate a Python virtual environment, and install requirements.txt. The presenter then downloads a model archive described as 193 MB, extracts its contents into the AI-onnx folder and launches QualityScaler.py. Viewers can follow the historical setup sequence and learn how the demonstrated controls affect output; the video does not establish whether those download links or commands still apply to later releases.