anti-aliasing
Blurring:
Time-domain map fuzzing operation
A time-domain image is Fourier transformed to obtain the frequency domain, the high-frequency part corresponds to the boundary value, and the low-frequency part is taken to obtain a fuzzy image;
convolution in the time domain (for example, 1/9 * (3 * 3) matrix) takes out 8 adjacent pixels The average value of is equal to the frequency domain cross product obtained by transforming the time domain and the corresponding matrix transformation;
- MSAA: Split a pixel into multiple pixels and perform binary operations to obtain the occupancy ratio and assign the pixel color;
- FSAA: Find the boundary value of the blurred image, and replace the jagged boundary with an algorithm;
- TAA: Take the image of the previous frame, calculate it with the current image, and assign the color to the value;
- Super resolution / super sampling Super resolution: 512 512 to 1024 1024 pixels, similar to the reverse process of MSAA, assigning a single pixel to multiple pixels;
- DLSS: anti-aliasing based on deep learning;