Anti-aliasing: translated as "anti-image folding distortion". Because in 3D images,Resolution Constraints,The edges of an object always show more or less the serticate of a triangle, while anti-serticate means to soften the edge of the image so that the edge of the image looks smoother and closer to the object.. It is a way to soften the image quality. The latest full-screen anti-sawtooth (fullsceneanti-aliasing) can effectively eliminate the dislocation of the polygon combination (especially in the combination of smaller polygon) and reduce the distortion of the image. During processing, the pixels near the image must be sampled 2-4 times to achieve different levels of anti-sawtooth effect. Simply put, the image edge and the pixel colors on both sides are mixed, then, replace the points at the original position with the newly generated points with mixed characteristics to soften the shape of the object and eliminate the Sawtooth effect.
1. Super sampling anti-aliasing (ssaa, supersampling anti-aliasing) is the earliest full-screen anti-aliasing method, which is simple and direct.
It is difficult for the human eye to tell more than 300 dpi (DPI is short for "dot Per Inch. As the name suggests, it refers to the number of points in the length of each inch .) The preceding color points. That is to say, if the resolution of a display with a diagonal line of 15 inch is 3600 × 2700, there is no need for full anti-aliasing, because the human eyes cannot tell such fine points. Super sampling and anti-aliasing means to multiply the current resolution (for example, if the current resolution is 1024 × 768, after ssaa is enabled twice, the screen is enlarged to 2048 × 1536 ), then scale the image to the current display. This is actually to increase the resolution when the display size remains unchanged, so that a single pixel becomes extremely small, which can greatly reduce the image's sawtooth. However, due to the enlargement of the entire display screen, it consumes a lot of display resources. The general process is as follows: first, the image is created to a separate buffer zone, and the image resolution in the buffer zone is higher than the screen resolution. Assume It is 2*1 (or 2x ), the horizontal size of the buffer area is twice higher than the screen resolution. If it is 2x2 (or 4x) anti-aliasing, the horizontal and vertical size of the buffer area image is twice larger than that of the displayed image. After the pixel calculation is doubled, two or four adjacent pixels are selected. This process is called sampling. After these samples are mixed, the final pixels generated have features of neighboring pixels, so the transitional colors between pixels become more similar, and the color transition of the entire image tends to be smoother. Then, the final pixel is output to the frame buffer, stored as an image, and then sent to the monitor to display a frame. Each frame is anti-sawtooth, and all images in the game become anti-sawtooth.
2. msaa is "Multi-sample anti-sawtooth", which can make the screen smoother.The principle of multisample anti-aliasing (multisampling anti-aliasing) is to multiply the current resolution and then scale the image to the current display. This is actually to increase the resolution when the display size remains unchanged, so that a single pixel becomes extremely small, which can greatly reduce the image's sawtooth. However, due to the enlargement of the entire display screen, it consumes a lot of display resources.
However, msaa looks for the pixels at the edge of an object and scales them. Because only the outer pixels of an object are scaled and the internal pixels that do not produce any serrations are ignored, the graphics card does not need as much computing as it does when processing ssaa (Super sampling and anti-aliasing, therefore, msaa is more effective than ssaa.
3. Overwrite sampling (csaa)Coveragesampling anti-aliasing (csaa) is a combination of the NVIDIA g80 series. The principle is to overwrite the subpixel coordinates to be sampled in the edge polygon, and place the original pixel coordinates in the hardware and driver's pre-calculated coordinates. This is like msaa, which has a unified sampling standard. It can run Edge Sampling with the highest efficiency, greatly improving traffic, and reducing resource usage.
4. Fast and approximate anti-tooth (fxaa) fast and approximate anti-tooth (fast approximate anti-aliasing)
It is a high-performance approximation of the traditional msaa (Multi-sample anti-aliasing) effect.. It is a one-way pixel coloring tool that runs in the post-processing stage of the rendering pipeline of the Target game like MLAA, but does not use directcompute as the latter, but simply a post-processing coloring tool, does not rely on any GPU computing API. Because of this, fxaa technology has no special requirements for graphics cards. It is fully compatible with different nvidia and AMD graphics cards (MLAA only supports a card), DX9, dx10, and dx11. 5. Time-based anti-sawtooth (txaa) allows the gaming experience of movie quality to reach a realistic level. Txaa: higher image quality than msaa, fxaa, and csaa. The film manufacturer that creates CG movies will spend a lot of computing resources on anti-aliasing, this ensures that the audience will not be distracted by unrealistic jagged lines. If you want to bring the game closer to this level of fidelity, developers need a new anti-sawtooth technology that not only reduces the amount of jagged lines, but also reduces the amount of jagged flashes without compromising performance. To help developers achieve this enhancement in fidelity, NVIDIA has designed a higher image quality anti-aliasing model called txaa, designed for direct integration into the game engine. Similar to the technology used in CG movies, txaa combines the powerful features of msaa with complex resolution filters to present smoother image effects, far surpassing all similar technologies. In addition, txaa can also perform jitter sampling for the entire scenario between frames to reduce the flickering situation, which is also known as time-based sawtooth technically. Currently, txaa has two modes: txaa 2X and txaa 4x. Txaa 2x provides visual fidelity comparable to 8x msaa, while the required performance is similar to 2xmsaa; txaa 4x delivers image fidelity better than 8 xmsaa, and the required performance is only equivalent to 4x msaa.
GPU anti-aliasing