AMD GPUOpen recently detailed MiniDXNN v0.4.0, an update to its interactive neural texture compression tool, bringing significant enhancements for DirectX 12 developers. The new version introduces support for D3D12 Linear Algebra (Shader Model 6.10) and expands capabilities with input encodings and neural texture compression. A key component of this release is a real-time GUI application, which allows users to train and visualize GPU-accelerated Multi-Layer Perceptrons (MLPs) directly on DirectX 12. This iteration aims to provide developers with advanced tools for efficiently compressing textures using machine learning techniques, potentially reducing VRAM usage and improving load times in D3D12 applications.