DD::Image::Execute Class Reference

#include <Execute.h>

Public Types
typedef unsigned	ExecuteVersion

Public Member Functions
OpGraphLocation	generateOpGraphAt (Op *op, const OutputContext &outputContext)
OpGraphLocation	generateOpGraphForInput (Op *op, int nodeInputNumber, const OutputContext &outputContext)
Op *	generateOp (Op *op, int inputnumber, double frame, int view=0)
Op *	generateOp (Op *op, int inputnumber, const OutputContext &outputContext)
void	reset ()
void	setVersion (int newVersion)
int	getVersion () const

Friends
class	ExecuteImplementation

Detailed Description

An "execution context". It is used to ask Nuke to generate a tree of Ops that will remain locked (ie not changed by any other execution contexts, though maybe shared by them) until the Execute is destroyed.

This is used to look at multiple frames (a "clip") of an operator's input. To do this a DD::Image::Op should: Unless you want to compare each frame to the current one, make Op::split_input() return -1. This indicates that Nuke should not lock ops for that input when building the tree, so they can be reused for this. If the output does not depend on the current frame (such as an operator that averages a fixed range together) you should also make Op::input_frame() return a constant. Otherwise the hash will vary depending on the frame number, causing it to recalculate as the user changes frames. To look at all the frames, the open() or execute() should create an Execute (usually as a local variable), then inside the loop it should call reset() and then generateOp() for each frame.

Member Function Documentation

generateOpGraphAt()

OpGraphLocation Execute::generateOpGraphAt	(	Op *	op,
		const OutputContext &	outputContext
	)

Generate an Op graph for the node that owns op at outputContext. This Op graph will be "locked" and will not be altered until the Execute is destroyed or reset() is called.

References DD::Image::Op::outputContext().

generateOpGraphForInput()

OpGraphLocation Execute::generateOpGraphForInput	(	Op *	op,
		int	nodeInputNumber,
		const OutputContext &	outputContext
	)

Generate an Op graph for the input on the node that owns op at outputContext. This Op graph will be "locked" and will not be altered until the Execute is destroyed or reset() is called.

References DD::Image::Op::outputContext().

Referenced by DD::Image::GeomOpEngine::inputConnectionAt().

generateOp() [1/2]

Op * DD::Image::Execute::generateOp ( Op *  op, int  inputnumber, double  frame, int  view = 0  )

inline

Produce input n of op at frame frame. This Op will be "locked" and will not be altered until the Execute is destroyed or reset() is called.

References DD::Image::OutputContext::setView(), and DD::Image::Op::uiContext().

generateOp() [2/2]

Op * DD::Image::Execute::generateOp ( Op *  op, int  inputnumber, const OutputContext &  outputContext  )

inline

Produce input n of op using values from output context outputContext. This Op will be "locked" and will not be altered until the Execute is destroyed or reset() is called.

reset()

void DD::Image::Execute::reset ( )

inline

Same as destroy+create of a new Execute, but slightly more efficient. All previous Ops from generateOp calls are unlocked and can be reused by this or other calls. In a typical loop you would call this after processing each frame number.