adaptive multisampling vs multisampling

[whiteknights]

i want to know whats is the difference between Adaptive multisampling & multisampling ..which gives better performance and which have the better graphics.

 

[Pinhedd]

i want to know whats is the difference between Adaptive multisampling & multisampling ..which gives better performance and which have the better graphics.

Super-Sample Anti-Aliasing, or Full-Scene Anti-Aliasing runs the entire render pipeline at some scalar multiple of the screen dimensions. For example, 2xSSAA applied to a viewport of 1920x1080 would render the entire scene at 3840x2160. An interpolation algorithm is then used to downsample the scene using a low-pass filter. This reduces aliasing over the entire scene. The output from the interpolation algorithm is used to fill the final viewport.
The interpolation algorithm itself is fairly quick, but rendering the entire scene at such a high resolution places a great demand on the shader programs. 2xSSAA increases the number of pixel shader runs by 4, and 4xSSAA increases it by 16.

Multi-Sample Anti-Aliasing is a particular optimization of SSAA. Rather than render the entire scene at a higher resolution and downsample it, MSAA detects locations in which two or more geometric primitives contribute to the same pixel (geometric overlap). The anti-aliasing algorithm is then only run when necessary, usually along edges, and acts as a component of the pixel shader. MSAA runs the pixel shader program only once per pixel, this is much easier on the GPU than SSAA.

Adaptive MSAA is an enhancement on MSAA that applies MSAA to textured areas that include transparencies. Imagine that there are two triangles that partially overlap. Lets call these triangles A and B. Part of triangle A sits in front of triangle B, so B is partially obscured by A. A and B are textured, but the texture applied to A includes a transparency. MSAA would smooth out the edges where A and B connect on the projection, but MSAA would not smooth out the parts of B that are visible through the transparent texture applied to A. Adaptive MSAA smooths this out as well.

 

[whiteknights]

i want to know whats is the difference between Adaptive multisampling & multisampling ..which gives better performance and which have the better graphics.

Super-Sample Anti-Aliasing, or Full-Scene Anti-Aliasing runs the entire render pipeline at some scalar multiple of the screen dimensions. For example, 2xSSAA applied to a viewport of 1920x1080 would render the entire scene at 3840x2160. An interpolation algorithm is then used to downsample the scene using a low-pass filter. This reduces aliasing over the entire scene. The output from the interpolation algorithm is used to fill the final viewport.
The interpolation algorithm itself is fairly quick, but rendering the entire scene at such a high resolution places a great demand on the shader programs. 2xSSAA increases the number of pixel shader runs by 4, and 4xSSAA increases it by 16.

Multi-Sample Anti-Aliasing is a particular optimization of SSAA. Rather than render the entire scene at a higher resolution and downsample it, MSAA detects locations in which two or more geometric primitives contribute to the same pixel (geometric overlap). The anti-aliasing algorithm is then only run when necessary, usually along edges, and acts as a component of the pixel shader. MSAA runs the pixel shader program only once per pixel, this is much easier on the GPU than SSAA.

Adaptive MSAA is an enhancement on MSAA that applies MSAA to textured areas that include transparencies. Imagine that there are two triangles that partially overlap. Lets call these triangles A and B. Part of triangle A sits in front of triangle B, so B is partially obscured by A. A and B are textured, but the texture applied to A includes a transparency. MSAA would smooth out the edges where A and B connect on the projection, but MSAA would not smooth out the parts of B that are visible through the transparent texture applied to A. Adaptive MSAA smooths this out as well.

thank you very much for this useful informations .. so this is means that Adaptive Multisampling is better than Multisampling graphically , and more gpu demanding?

 

[Pinhedd]

i want to know whats is the difference between Adaptive multisampling & multisampling ..which gives better performance and which have the better graphics.

Super-Sample Anti-Aliasing, or Full-Scene Anti-Aliasing runs the entire render pipeline at some scalar multiple of the screen dimensions. For example, 2xSSAA applied to a viewport of 1920x1080 would render the entire scene at 3840x2160. An interpolation algorithm is then used to downsample the scene using a low-pass filter. This reduces aliasing over the entire scene. The output from the interpolation algorithm is used to fill the final viewport.
The interpolation algorithm itself is fairly quick, but rendering the entire scene at such a high resolution places a great demand on the shader programs. 2xSSAA increases the number of pixel shader runs by 4, and 4xSSAA increases it by 16.

Multi-Sample Anti-Aliasing is a particular optimization of SSAA. Rather than render the entire scene at a higher resolution and downsample it, MSAA detects locations in which two or more geometric primitives contribute to the same pixel (geometric overlap). The anti-aliasing algorithm is then only run when necessary, usually along edges, and acts as a component of the pixel shader. MSAA runs the pixel shader program only once per pixel, this is much easier on the GPU than SSAA.

Adaptive MSAA is an enhancement on MSAA that applies MSAA to textured areas that include transparencies. Imagine that there are two triangles that partially overlap. Lets call these triangles A and B. Part of triangle A sits in front of triangle B, so B is partially obscured by A. A and B are textured, but the texture applied to A includes a transparency. MSAA would smooth out the edges where A and B connect on the projection, but MSAA would not smooth out the parts of B that are visible through the transparent texture applied to A. Adaptive MSAA smooths this out as well.

thank you very much for this useful informations .. so this is means that Adaptive Multisampling is better than Multisampling graphically , and more gpu demanding?

SSAA > Adaptive MSAA > MSAA > FXAA

FXAA is an ultra fast variation on SSAA where the low pass filter algorithm (performed in post processing) is performed without rendering the scene at a higher resolution. It's a poor man's SSAA.

 

[iron8orn]

poor man's ssaa lmao

 

[whiteknights]

i want to know whats is the difference between Adaptive multisampling & multisampling ..which gives better performance and which have the better graphics.

Super-Sample Anti-Aliasing, or Full-Scene Anti-Aliasing runs the entire render pipeline at some scalar multiple of the screen dimensions. For example, 2xSSAA applied to a viewport of 1920x1080 would render the entire scene at 3840x2160. An interpolation algorithm is then used to downsample the scene using a low-pass filter. This reduces aliasing over the entire scene. The output from the interpolation algorithm is used to fill the final viewport.
The interpolation algorithm itself is fairly quick, but rendering the entire scene at such a high resolution places a great demand on the shader programs. 2xSSAA increases the number of pixel shader runs by 4, and 4xSSAA increases it by 16.

Multi-Sample Anti-Aliasing is a particular optimization of SSAA. Rather than render the entire scene at a higher resolution and downsample it, MSAA detects locations in which two or more geometric primitives contribute to the same pixel (geometric overlap). The anti-aliasing algorithm is then only run when necessary, usually along edges, and acts as a component of the pixel shader. MSAA runs the pixel shader program only once per pixel, this is much easier on the GPU than SSAA.

Adaptive MSAA is an enhancement on MSAA that applies MSAA to textured areas that include transparencies. Imagine that there are two triangles that partially overlap. Lets call these triangles A and B. Part of triangle A sits in front of triangle B, so B is partially obscured by A. A and B are textured, but the texture applied to A includes a transparency. MSAA would smooth out the edges where A and B connect on the projection, but MSAA would not smooth out the parts of B that are visible through the transparent texture applied to A. Adaptive MSAA smooths this out as well.

thank you very much for this useful informations .. so this is means that Adaptive Multisampling is better than Multisampling graphically , and more gpu demanding?

SSAA > Adaptive MSAA > MSAA > FXAA

FXAA is an ultra fast variation on SSAA where the low pass filter algorithm (performed in post processing) is performed without rendering the scene at a higher resolution. It's a poor man's SSAA.

thank you, last thing please .. if i have HD 7870 XT gpu, which settings will be balanced for me (good graphics without a huge performance hit) i didn't notice any difference between Adaptive MultiSampling and MultiSampling , I'm not sure why .. is it because i set Anti-aliasing mode on "use application settings" so putting Anti-aliasing mode on "adaptive multisampling" won't run in the game?

 

[Pinhedd]

i want to know whats is the difference between Adaptive multisampling & multisampling ..which gives better performance and which have the better graphics.

Super-Sample Anti-Aliasing, or Full-Scene Anti-Aliasing runs the entire render pipeline at some scalar multiple of the screen dimensions. For example, 2xSSAA applied to a viewport of 1920x1080 would render the entire scene at 3840x2160. An interpolation algorithm is then used to downsample the scene using a low-pass filter. This reduces aliasing over the entire scene. The output from the interpolation algorithm is used to fill the final viewport.
The interpolation algorithm itself is fairly quick, but rendering the entire scene at such a high resolution places a great demand on the shader programs. 2xSSAA increases the number of pixel shader runs by 4, and 4xSSAA increases it by 16.

Multi-Sample Anti-Aliasing is a particular optimization of SSAA. Rather than render the entire scene at a higher resolution and downsample it, MSAA detects locations in which two or more geometric primitives contribute to the same pixel (geometric overlap). The anti-aliasing algorithm is then only run when necessary, usually along edges, and acts as a component of the pixel shader. MSAA runs the pixel shader program only once per pixel, this is much easier on the GPU than SSAA.

Adaptive MSAA is an enhancement on MSAA that applies MSAA to textured areas that include transparencies. Imagine that there are two triangles that partially overlap. Lets call these triangles A and B. Part of triangle A sits in front of triangle B, so B is partially obscured by A. A and B are textured, but the texture applied to A includes a transparency. MSAA would smooth out the edges where A and B connect on the projection, but MSAA would not smooth out the parts of B that are visible through the transparent texture applied to A. Adaptive MSAA smooths this out as well.

thank you very much for this useful informations .. so this is means that Adaptive Multisampling is better than Multisampling graphically , and more gpu demanding?

SSAA > Adaptive MSAA > MSAA > FXAA

FXAA is an ultra fast variation on SSAA where the low pass filter algorithm (performed in post processing) is performed without rendering the scene at a higher resolution. It's a poor man's SSAA.

thank you, last thing please .. if i have HD 7870 XT gpu, which settings will be balanced for me (good graphics without a huge performance hit) i didn't notice any difference between Adaptive MultiSampling and MultiSampling , I'm not sure why .. is it because i set Anti-aliasing mode on "use application settings" so putting Anti-aliasing mode on "adaptive multisampling" won't run in the game?

Yes, that's correct. Forcing a particular anti-aliasing method in the driver can cause issues, so it's best to let the application handle it.