分享：Mark Joey Tang
版本：11.0, 10.5 or later Linux, Mac, Windows
功能：本套插件可以把深度数据转换为体积位置数据，包含了'DeepToPosition' , 'DeepVolumeMatte' , 'DeepVolumeNoise' , 'DeepPosition_2d_convert'四个节点。
This tool is using tcl in deepexpression to convert deep data to position. It supports any deep data, single layer or volumetric data. Since deep render is all related to camera and depth, so the toolset required camera data, 'translation, rotation, focal length and h aperture'. Apparently deep data, it can be CG render or through ScanlineRender. Since the tool convert all deep samples to world position data, while using any tools to process with world position data, it can still keep all the volumetric data. And also this is the main purpose of this toolset.
*Beware: before Nuke11, DeepExpression has some strange behavior, thanks Foundry to fix most of them in Nuke11. So this tool is only work and tested in Nuke11. If you are using lower version, this tool will not work probably.
This toolset package included 4 nodes. Following is the decription of each :
Input is deep, output is deep. This node is the core of this toolset, it will convert all deep data to world position. Translate, Rotate, Rotate Order, Translate Order, Focal Length, and Horiz Aperture need to provide by user. These data need to be same as the input deep render. There has a auto bot function to help enter above data. User only have to do is select the camera node and click 'link to selected camera', then the node will set all knobs to camera with expression.
Input is deep, output is deep. This node will take the axis data (position, rotation, scale, skew) and set the matte with falloff in deep. If you are familiar position matte, yes, this is doing the same job but support in deep volume. User can use axis or geo to set the matte position, or type in the transformation manually. This node also provided auto bot function. User only have to do is select the axis or geo and click 'link to selected axis/geo', then the node will set all knobs to axis/geo with expression.
Input is deep, output is deep. This node will create noise in deep and support in volumetric as well. It supports 2 types of noise pattern : noise and random. Volume Transparency can be adjust. Of couse noise size, position offset, gain, gamma, color offset also do. It can output Color (RGB premulted) or black and white color. If you familiar position noise tool, yes, this is doing the same job but support in deep volume.
Input is deep, output is 2D. When the object is transparent or with transparent falloff, the position data will get mess with alpha. It is because the default DeepToImage will premult all the channel with alpha. Since Position is tech data and it shouldn't calculate in this way, so this node will handle the data conversion.
The download package also included the example script to show how all these nodes work. Feel free to contact me if any questions about it.
Set the position of the matte in 3D space
The matte falloff can be adjust in different type, individual value in x,y,z
Here is the example script looks like :