请教显卡里面的2x和2xQ有什么区别？

jarry168

请教显卡里面的2x和2xQ有什么区别？
显卡里面的3D平滑处理，有选项2x，那就是2倍抗锯齿，2xQ是什么？Q代表什么？什么样的效果呢？

能请明白的大大告知一二吗？比如和4x有什么区别？ 有什么比较的文章吗？

万分感谢

2xQ就是5点梅花采样。
GF3Ti的招牌AA，后来因为它的固有原因，NV放弃了对它的优化，没落了。

2XQ开了以后画面特模糊,不知道怎么回事

原帖由 hejianfei520 于 2006-10-23 22:35 发表
2XQ开了以后画面特模糊,不知道怎么回事

没解，这就是5点梅花的固有缺陷。
NV的20.30驱动里还有专门针对5点梅花的锐化选项，到4X.XX后消失了。名字也换成了2XQ这个不能体现它特点的名字了。估计NV放弃它了。

我现在用的91.48好像选择2xQ没有模糊的效果了

那么请问 2xQ和4x 差别有多大？

From nHancer Online Documentation
Anti-AliasingVisual comparison of the various Anti-Aliasing modes
A common problem when a scene has to be rendered with a finite resolution (which is always the case with the current technology) are the "jaggies" that are especially visible alongside the edges of rendered objects. This effect is called aliasing.
Increasing the resolution can reduce the visibility of the aliasing effect, but as long as the resolution isn't at least twice the resolution of the human eye (and we're still far far away from that resolution), it will always be visible.
Another effect of aliasing is the shimmering, that happens if the character moves in the scene or if the scene itself changes (e.g. waving trees). This shimmering is caused by the movement of the "jaggies". This is especially bad, if you have fine details, like tree branches, which are too small to be correctly drawn at the given resolution. These details tend to "pop in" and "pop out" with every smallest movement, which usually causes a heavy flickering of these structures. This effect is equally apparent with high resolutions (the detailed structure has just to be farther away from the viewer), so increasing the resolution is not really an alternative.
To elliminate the effects of aliasing, the graphic card employs a technique called Anti-Aliasing. nVidia offers various variations of this technique, each with their own advantages and disadvantages. All these variations have in common, that the "jaggies" are made less visible by causing a smoother transition of colors along the edges. The human eye tends to blend these transitions, so the viewer can't see the individual pixels anymore. Only if you zoom in into such a rendered image (which is done with the examples in the Visual Presentation), you will see how this technique works. Unfortunately the most important advantage cannot be seen on the presentation, since it is only apparent if the scene is moving.

MultisamplingMultisampling is currently the most common way of doing Anti-Aliasing these days. On some graphic cards, it's the only mode available to the user.
With multisampling, each pixel at the edge of a polygon is sampled multiple times. For each sample-pass, a slight offset is applied to all screen coordinates. This offset is smaller than the actual size of the pixels. By averaging all these samples, the result is a smoother transition of the colors at the edges.
The number of samples taken for each pixel indicates the level of multisampling. So the 2x multisampling mode means that two samples are taken, while 4x means that four samples are taken for each pixel at the edge of a polygon.
An important factor of this method is how these coordinate offsets are arranged. A so called ordered grid smoothes all edges that aren't close to exactly horizontal or vertical angles. But on the other hand, most edges tend to be close to horizontal or vertical angles, so this grid is usually replaced by a rotated grid where this disadvantage is removed. Unfortunately, only the latest GeForce cards (starting with the 6xxx line of products) is capable of using a rotated grid in the 4x mode, all previous models are using an ordered grid. For the 2x mode, however, all models are using a rotated grid.
Advantages

• Very efficient. All multisampling modes causes a relatively small performance hit but have a huge effect on the image quality.
• Available in all resolutions.

Disadvantages

• Only the edges of polygons are smoothed. The inner parts of textures are not changed at all. This is very apparent with alpha-test textures (meaning that they have "look-trough" holes). Those are often used for grates, fences, tree branches and similar structures, because these details aren't usually modelled using polygons to reduce the complexity of the scene.
• Doesn't reduce texture shimmering, which is caused by textures with a very detailed structure. Usually the texture creators avoid such details, but in some games, this texture shimmering is still very apparent.

nVidia also offers an additional mode called 2x Quincunx. This mode adds a blurring filter on the normal multisampled image. This helps to reduce the disadvantages of multisampling, but the result is a visibly blurred image, which is especially apparent with texts or other details on the screen. This mode is therefore quite rarely used.
Another mode, which is not officially available but can be forced using nHancer, is called 4x gamma. This mode applies a gamma correction to the inbetween colors at the edges. This can improve the subject smoothing, if the monitor uses the correct gamma setting. But this mode also causes an additional performance hit, and the improvement of the image quality doesn't usually outweigh this hit. Right now, this gamma correction is the only difference in the multisampling modes between the current ATI and nVidia graphic cards. All ATI anti-aliasing modes are using gamma corrected multisampling.
SupersamplingSupersampling is a very straight forward method, that was also first introduced on Voodoo 4/5 cards.
The image is just rendered with a higher resolution internally. After a whole image has been rendered, it is then scaled down to the target resolution. While doing that, each pixel is colored from the avarage of all apropriate pixels of the high-resolution image.
Supersampling modes are named after the size of the internally used resolution. I.e. the mode 2x1 means that the internal resoltion has twice the horizontal resolution and the same vertical resolution. The result is, that each final pixel is made from the average of 2 pixels. With the 4x4 supersampling mode, each final pixel is created from the average of 16 pixels.
Advantages

• Excellent image quality.
• Even the edges within alpha-blended textures (see above) are correctly smoothed.
• Also reduces texture shimmering.

Disadvantages

• Heavy performance hit.
• Limitations on the target resolution. The maximum internal resolution is 4096 in both directions. So the 4x4 mode can only be used with 1024x768 or less.
• Some supersampling modes are only available using Direct3D, but not in OpenGL games.

For most games, especially new ones, the performance hit is not acceptable, if you want to use a resolution of 1280x1024 or more. But if you're playing an older game and if you're using a high-end model of you graphic card, the supersampling modes cause a tremendous improvement on the image quality, while still producing acceptable framerates.
There are even some current games (e.g. "Chronicles of Riddick"), which produce very heavy texture shimmering (because of the extremely high detailed textures and the bump-mapping effects). It can actually be preferrable to reduce the resolution and apply the 2x2 supersampling mode in such cases.
Combined modesnVidia also offers a couple of combined anti-aliasing modes. These modes are using mutlisampling and supersampling at the same time.
Currently there are the following modes:

• 4xS:
  This mode combines 1x2 supersampling with 2x multisampling. This mode is still quite cheap on performance but can already have great effects on the image quality, if the game uses many alpha-blended textures or causes heavy texture shimmering.
• 8xS:
  This is the only combined mode which is available through the standard control panel. Just like the 4xS mode, it has a 1x2 supersampling component, but combined with 4x multisampling. This mode doesn't cause a much higher performance hit as 4xS but can produce a better result.
• 8xS old:
  In older driver versions (pre 6x.xx), the official 8xS mode used 2x2 supersampling and 2x multisampling. This mode is still available through nHancer, and to differentiate it from the current official 8xS mode it has been renamed to 8xS old. This mode offers a better quality through its higher supersampling component, especially with alpha-blended textures, while also causing a much higher performance hit than the current official 8xS mode.
  Note: It has been discussed that this mode doesn't work as intended in the latest driver versions, which is quite possible. It's said that the multisampling component isn't working correctly. But if you compare the pure 2x2 with this mode using the Visual Presentation a slight improvement of the quality is visible (note the horizontal edge in the upper part of the zoomed wall section). Still, this mode offers very little in comparison with the pure 2x2 supersampling mode.
• 16xS:
  This mode combines 2x2 supersampling with 4x multisampling. This mode offers the second best anti-aliasing, a GeForce card is capable of (best being 4x4 supersampling) at a rather high performance cost.

To compare the various anti-aliasing modes, you can use this Visual Presentation.

大哥不给翻译怎么看？？？？？？？？？？？？？？？？？？？？？

从网上nhancer文件反aliasingvisual比较各种反走样方式有一个共同的问题时,现场将提供一个有限草案(即总是这样,与当前科技)是"jaggies",尤其是边缘变得有形物体旁边.这一效应称为走样.草案增加了知名度,可以减少走样效果但只要不是第至少两次解决人类眼睛(和我们还是远远远离决议)它永远看得见.另一个是波光效果走样,如果这种情况发生,在现场的举动性格现场或者本身变化(如挥动树).这波光是运动造成的"jaggies".尤为恶劣,如果你有细节,像树枝,而太小会得出正确的给予解决.这些细节倾向于"流行歌曲"、"流行"与每个最小运动通常这些结构造成了沉重摇曳.这一决议的效力也同样明显偏高(只将详细的架构已经远的观众)这样的决议并不是真的越来越另类.为有理数效果走样,图形卡采用了技术称为反走样.这种技术提供各种变异真实感,每用自己的利弊得失.所有这些变化有共同语言,说"jaggies"是不太明显的平滑过渡色彩造成沿线的优势.人类眼睛这些跃迁趋向融合,让观众看不到个人象素了.只有在这样一个令你变焦形象(即做表率的视觉)你将看到这种技术的工程.可惜最重要的优势就不能看见套、由于它是明显的,如果场面感人.multisamplingmultisampling目前最常见的做法反走样这些日子.一些图形卡,它的唯一的方式提供给用户.同采样,每个像素是在边缘多边形采样成倍增长.每个样品绕道,稍微弥补应用于所有屏幕坐标.这抵消小于实际的象素大小.由这些样本平均,结果是平滑过渡色彩的边缘.样本数为每像素采样显示水平.所以倍速采样方式意味着两个样本后,而四倍体四个样本做意味着每个像素在边缘多边形.这种方法的一个重要因素是,如何协调这些冲抵则不在此限.所谓下令格边缘平滑所有那些不接近横向或纵向到底酣畅.但另一方面,也往往是最贴近边缘横向或纵向角度这通常是由一个旋转网格这一不利拆除.可惜仅最近霸主卡(从6xxx的产品线)能够利用旋转格在四倍体模式以前所有型号都使用下令格.为倍速模式,但所有型号都采用旋转格.优势非常有效率.所有采样方式造成冲击相对较小,但性能有很大的影响,图像质量.在全部决议.唯一缺点是多边形边缘平滑.党内部份织并不改变.这是很明显的阿尔法试织(亦即"看-低谷"洞).那些常常用箅子、围栏、树枝及类似结构因为这些细节并非通常用多边形为蓝本,以减少复杂现场.不减少纹理波光,这是造成织与结构提供了非常详细.通常的纹理等细节,避免创作者,但在一些游戏,这纹理波光还很明显.真实感也提供额外的模式称为倍速五株.在这种模式中加入过滤模糊不清正常multisampled形象.这有助于减少多重弊端,但结果却是明显的图像模糊,其中尤为明显的文本或其他细节上屏幕.因此,用这种方式相当罕见.另一方式,但并非正式资料可以用强迫nhancer,称为四倍体伽马.这种模式适用于伽马校正梯级边缘的色彩.这个题目可以改善平滑,如果正确使用监测伽玛设置.但这种模式也造成额外演出火爆,以及改善的图像质量通常不超过这个命中.现在,这是唯一的伽玛矫正采样模式之间的差异和目前的保险真实感图形卡.所有的保险反走样利用伽玛矫正采样方式.supersamplingsupersampling是一个非常直截了当的方法,那也首度推出巫术四月/五月卡.图像分辨率较高,只是提供了国内.整体形象呈现在舞台,那么具体而微的目标草案.在此期间,每个像素着色是从所有适当4:1~15:1像素的高清晰度图像.supersampling方式命名用于解决国内规模.即2x1模式意味着内部高分辨率两次相同的水平分辨率和垂直分辨率.结果是,最后,每个像素是由平均200像素.同4-4supersampling模式,每个像素的最终形成由平均16个像素.优点出色的图像质量.甚至在边缘阿尔法-混织(见上述)是正确平滑.也可以降低纹理波光.不利之处表现沉重打击.限制对象决议案.最高内部决议是4096年两个方向.四轮驱动模式,所以只能用1024x768或更少.有的只是提供使用Direct3D的supersampling方式,但不是在OpenGL游戏.大部分游戏,尤其是新近表现击中是不能接受的,如果你想用一个分辨率1280x1024或以上.但如果你玩游戏的年纪了,如果你用的高端图形卡,你模型、supersampling模式的形象造成了极大的改善品质,同时还生产可接受framerates.甚至还有本届运动会(如"星际传奇")产生很重纹理波光(因为详细的笔触与极高碰映射效果).它可以减少可取决议和运用人际关系supersampling模式在此种情况.合计数合计modesnvidia亦为反走样方式.这些方式都使用mutlisampling在同一时间和supersampling.目前有以下方式:4xs:这种模式结合1x2倍速supersampling与采样.这个模式还是相当便宜,但性能已经有很大的影响,图像质量、如果用许多游戏阿尔法-织或者造成沉重混纺质地波光.刃:这是唯一的方式就是透过结合标准控制面板.就像4xs模式,具有1x2supersampling组件,而是结合四倍体采样.这种模式导致更高的表现并不如击中4xs但能产生更好的效果.巷内旧:旧版本司机(预6x.xx),正式妾采实验室supersampling模式和倍速采样.这个模式仍可以通过nhancer,而从目前的区分已正式更名为钻模钻旧.通过这种模式提供了优质的高supersampling组件,尤其阿尔法-织混纺、同时也造成了冲击性能远远超过目前正式巷内模式.注:它已讨论这个模式行不通,因为打算在最新的驱动版本很可能.这并不是说采样组件正确工作.但如果你人际关系比较单纯使用这种方式略有改善视觉品质看得见(注横向力上部墙的缩放节).但是,很少有这种模式比较单纯人际关系supersampling模式.16xs:这种模式结合人际关系与四倍体supersampling采样.这种模式提供退而求其次反走样,一个具有真实感卡(最好是4-4supersampling)演出费用相当高.比较各反走样方式,你可以用这个录影.

汗楼上的
还不如看鸟文呢

原帖由 sleepbirdcyp 于 2006-10-23 22:49 发表
我现在用的91.48好像选择2xQ没有模糊的效果了

那么请问 2xQ和4x 差别有多大？

你自己开一下不就知道了