T & H

I haven't updated it for a long time. Send it to lower the previous article.

 

* In D9, setstreamsource/drawindexedprimitive often forgets the meanings of these parameters for a long time.
Offsetinbytes: Specifies the starting position of the VB data, that is, the 0th vertices in IB, starting from this address.
Basevertexindex: add this value to each index in Ib. For example, if 2 is set, 5 in IB is changed to 7... It is not very common, but sometimes it has unexpected effects.
Minvertexindex: tells d3d that the vertex used starts from the first vertex in VB and does not affect rendering after the settings. It only helps d3d improve performance in some cases...
Numvertices: Tell d3d how many vertices are used in this DP and there is no impact on rendering after the settings. Note that this has nothing to do with minvertexindex and is not counted from minvertexindex, it only helps d3d improve performance in some cases...
Startindex: this parameter is of the offsetinbytes type and sets the start point for reading IB data.

 

* The cry terrain does not use a chart to calculate the alpha of each layer, but is represented by the color level 256 in the vertex. Each value represents a layer. Therefore, it cannot represent a translucent layer. It can only be dropped from 1 to 0 at the edge of 1 lattice.

 

* D3dformat makefourcc ('n', 'U', 'l', 'L') does not need to set the target to 1x1 (in CRY). No matter how large the target is, it will not consume the video memory, and the number of pixels written to the draw is also 0, and the speed does not change.

 

* Simple bone Animation:

The number of changes in the transformation from the skeleton to the skeleton of the current frame * When the mesh is bound = the vertex x when the mesh is bound.

= X y h.

= X y.

Where,

Vertex when the mesh is bound: it is exported by Max.

The inverse matrix of the skeleton transformation during binding: Obtained from getboneinittm of Max's skin modifier (or getinitnodetm of physique modifier.

The transformation of the skeleton of this frame to the parent node is obtained from the nodetm of the Max skeleton and the nodetm of the parent bone. Note that for the row matrix, the matrix sequence is parenttochild = childtoworld * inverse (parenttoworld ). The column matrix is the opposite. The root node stores the nodetm directly. The interpolation between frames is mainly used for interpolation of the Data. When there is no major change between the two frames, there is no big difference between the matrix and the matrix plus vector.

Transformation of the parent skeleton of this frame: It is accumulated from the root node.

 

* Simple method for volume painting by light in deferred shading.
When the eyes are outside volume, draw volume directly with CCW; when the eyes are about to enter volume or in volume, draw with CW, and slightly enlarge volume to include camera. In addition to changing the cullmode, the rendering process does not need to be changed at all.

 

* In CSM, if shadowmap uses Atlas to place it in a graph, for example, when four slice is placed in one graph, a very small (64x64) can be used to calculate shadow receive) to select the slice UV. Faster than direct computing.