Filter SDK/Cplusplus API

From Avisynth wiki
Revision as of 23:03, 21 August 2013 by Admin (Talk | contribs)

Jump to: navigation, search

The header, avisynth.h, declares all the classes, structures and miscellaneous constants of the C++ API that you might need when writing a plugin. All external plugins should #include it: 

#include "avisynth.h"

Note, sometimes there is a reference to a version number (for example v3 or v5). This refers to the value of AVISYNTH_INTERFACE_VERSION. The classes and miscellaneous constants are described below.

Contents

CreateScriptEnvironment 

 IScriptEnvironment* __stdcall CreateScriptEnvironment(int version = AVISYNTH_INTERFACE_VERSION);

AviSynth exports this. It enables you to use AviSynth as a library, without writing an AviSynth script or without going through AVIFile. [link]

Classes

AvisynthError

AvisynthError(const char* _msg)

Wrap your code in try/catch statements to enable exception handling. AvisynthError will tell you what's wrong. 

try 
{	
  Val = Env->Invoke("Import", Args, 0);
  Clip = Val.AsClip();
  VidInfo = Clip->GetVideoInfo();
  Frame = Clip->GetFrame( 1, Env);
}
  catch (AvisynthError err) 
{
  printf("%s\n", err.msg);
  return 1;
}

VideoFrameBuffer

VideoFrameBuffer (VFB) holds information about a memory block which is used for video data. For efficiency, instances of this class are not deleted when the refcount reaches zero; instead they are stored in a linked list to be reused. The instances are deleted when the corresponding AVS file is closed. Or more accurately, a VideoFrameBuffer once new'd generally is not released until the IScriptEnvironment is deleted, except if SetMemoryMax is exceeded by too much then not in use VideoFrameBuffer's are forcible deleted until SetMemoryMax is satisfied.

VideoFrame

VideoFrame holds a "window" into a VideoFrameBuffer. Operator new is overloaded to recycle class instances. Its members can be called by: 

PVideoFrame src = child->GetFrame(n, env);
src->GetReadPtr(..)

VideoFrame has the following members: GetPitch, GetRowSize, GetHeight, GetReadPtr, GetWritePtr and IsWritable.

All those filters (except IsWritable) will give you a property (pitch, rowsize, etc ...) of a plane (of the frame it points to). The interleaved formats (BGR(A) or YUY2) consist of one plane, and the planar formats consists of one (Y) or three (YUV) planes. The default plane is just the first plane (which is plane Y for the planar formats).

GetPitch 

 int GetPitch(int plane=0) const;

The "pitch" of a frame buffer is the offset (in bytes) from the beginning of one scan line to the beginning of the next. The source and destination buffers won't necessarily have the same pitch. The pitch can vary among frames in a clip, and it can differ from the width of the clip. [link] NOTE that the pitch can change anytime, so in most use cases you must request the pitch dynamically.

Meaning:

Image processing is expensive, so SIMD instructions are used to speed tasks up. SSE uses 128 bit = 16 byte registers, so 8 YUY2 pixels can be processed the same time. AVX uses 256 bit = 32 byte registers, so 16 YUY2 pixels can be processed the same time.

Usage:

GetPitch must be used on every plane (interleaved like YUY2 means 1 plane...) of every PVideoFrame that you want to read or write to. It is the only way to get the size of the Video Buffer (e.g. get the size of PVideoFrame): 

int buffer_size = src->GetPitch() * src->GetHeight();  //YUY2, interleaved

This will give you the pitch of the U-plane (it will be zero if the plane doesn't exist): 

PVideoFrame src = child->GetFrame(n, env);
const int src_pitchUV = src->GetPitch(PLANAR_U);

GetRowSize 

 int GetRowSize(int plane=0) const;

GetRowSize gives the length of each row in bytes (thus not in pixels). It's usually equal to the pitch or slightly less, but it may be significantly less if the frame in question has been through Crop. This will give you the rowsize of a frame for the interleaved formats, or the rowsize of the Y-plane for the planar formats (being the default plane). const int src_width = src->GetRowSize();

GetHeight 

 int GetHeight(int plane=0) const;

GetHeight gives the height of the plane in pixels.

GetReadPtr 

 const BYTE* GetReadPtr(int plane=0) const;

GetReadPtr gives you a read pointer to a plane. This will give a read pointer to the default plane: 

PVideoFrame src = child->GetFrame(n, env);
const unsigned char* srcp = src->GetReadPtr()

GetWritePtr 

 BYTE* GetWritePtr(int plane=0) const;

GetWritePtr gives you a write pointer to a plane.

Any buffer you get from NewVideoFrame is guaranteed to be writable (as long as you only assign it to one PVideoFrame). Our filter's dst came from NewVideoFrame, so we can safely call dst->GetWritePtr(). However, frames you get from other clips via GetFrame may not be writable, in which case GetWritePtr() will return a null pointer. 

PVideoFrame dst = env->NewVideoFrame(vi);				
unsigned char* dstp = dst->GetWritePtr();

If you want to write a frame which is not new (the source frame for example), you will have to call MakeWritable first: 

PVideoFrame src = child->GetFrame(n, env);
env->MakeWritable(&src);
unsigned char* srcp = src->GetWritePtr(PLANAR_Y);

See IsWritable for more details.

IsWritable 

 bool IsWritable() const;

All frame buffers are readable, but not all are writable. This method can be used to find out if a buffer is writable or not, and there's a MakeWritable callback (described below) to ensure that it is.

The rule about writability is this: A buffer is writable if and only if there is exactly one PVideoFrame pointing to it. In other words, you can only write to a buffer if no one else might be reading it. This rule guarantees that as long as you hold on to a PVideoFrame and don't write to it yourself, that frame will remain unchanged. The only drawback is that you can't have two PVideoFrames pointing to a writable buffer. 

PVideoFrame src = child->GetFrame(n, env);
if (src->IsWritetable()) {...}

AlignPlanar 

 AlignPlanar(PClip _clip);

AlignPlanar does nothing, if the pitch of a frame is at least mod16 (16 bytes, being the default frame alignment for luma and chroma). Otherwise it realigns the image, by blitting it to a larger buffer.

Filters can enforce a lower pitch, but they must always apply the AlignPlanar filter after itself, if they intend to return a frame with a lower pitch. VFW delivers a 4 byte alignment for example, so the AlignPlanar filters needs to be applied on all frames when using AviSource.

FillBorder 

 FillBorder(PClip _clip);

This function fills up the right side of the picture on planar images with duplicates of the rightmost pixel if the planes are not aligned. That is, if src->GetRowSize(PLANAR_Y) != src->GetRowSize(PLANAR_Y_ALIGNED).

ConvertAudio 

 ConvertAudio(PClip _clip, int prefered_format);

ConvertAudio converts the sample type of the audio to one of the following sample types: SAMPLE_INT8 (8 bits), SAMPLE_INT16 (16 bits), SAMPLE_INT24 (24 bits), SAMPLE_INT32 (32 bits) or SAMPLE_FLOAT (float).

The following example converts the sample type of the clip child to SAMPLE_INT16 (16 bit) if the input isn’t 16 bit. 

ConvertAudio(child, SAMPLE_INT16);

IScriptEnvironment

AviSynth exports an IScriptEnvironment interface. It enables you to use AviSynth as a library, without writing an AVS script or without going through AVIFile. Its members can be called by: 

IScriptEnvironment* env
env->Invoke(..)

IScriptEnvironment has the following members: ThrowError, GetCPUFlags, SaveString, Sprintf, VSprintf, Invoke, BitBlt, AtExit, AddFunction, MakeWritable, ... . They are described in the following subsections.

ThrowError 

 __declspec(noreturn) virtual void __stdcall ThrowError(const char* fmt, ...) = 0;

ThrowError throws an exception (of type AvisynthError). Usually, your error message will end up being displayed on the user's screen in lieu of the video clip they were expecting: 

if (!vi.IsRGB()) {
  env->ThrowError("RGBAdjust requires RGB input");
}

GetCPUFlags 

 virtual long GetCPUFlags();

GetCPUFlags returns the instruction set of your CPU. To find out if you're running for example on a CPU that supports MMX, test: 

env->GetCPUFlags() & CPUF_MMX

There's a complete list of flags in avisynth.h.

SaveString 

 virtual char* SaveString(const char* s, int length = -1);

This function copies its argument to a safe "permanent" location and returns a pointer to the new location. Each ScriptEnvironment instance has a buffer set aside for storing strings, which is expanded as needed. The strings are not deleted until the ScriptEnvironment instance goes away (when the script file is closed, usually). This is usually all the permanence that is needed, since all related filter instances will already be gone by then. The returned pointer is not const-qualified, and you're welcome to write to it, as long as you don't stray beyond the bounds of the string.

Example usage (converting a string to upper case): 

AVSValue UCase(AVSValue args, void*, IScriptEnvironment* env) {
  return _strupr(env->SaveString(args[0].AsString()));
}

Sprintf and VSprintf 

 virtual char* Sprintf(const char* fmt, ...);
 virtual char* VSprintf(const char* fmt, char* val);

These store strings away in the same way as SaveString, but they treat their arguments like printf and vprintf. Currently there's a size limit of 4096 characters on strings created this way. (The implementation uses _vsnprintf, so you don't need to worry about buffer overrun.)

Invoke 

 virtual AVSValue Invoke(const char* name, const AVSValue args, const char** arg_names=0);

You can use this to call a script function. There are many script functions which can be useful from other filters; for example, the Bob filter uses SeparateFields, and several source filters use UnalignedSplice. Some functions, like Weave, are implemented entirely in terms of other functions. If you're calling a function taking exactly one argument, you can simply pass it in the args parameter; Invoke will convert it into an array for you. In order to call a function taking multiple arguments, you will need to create the array yourself; it can be done like this: 

AVSValue up_args[3] = {child, 384, 288}; 
PClip resized = env->Invoke("LanczosResize", AVSValue(up_args,3)).AsClip();

In this case LanczosResize would need to have a parameter-type string like "cii".

The arg_names parameter can be used to specify named arguments. Named arguments can also be given positionally, if you prefer.

Invoke throws IScriptEnvironment::NotFound if it can't find a matching function prototype. You should be prepared to catch this unless you know that the function exists and will accept the given arguments.

BitBlt 

 virtual void BitBlt(unsigned char* dstp, int dst_pitch, const unsigned char* srcp, int src_pitch, int row_size, int height);

This brilliantly-named function does a line-by-line copy from the source to the destination. It's useful for quite a number of things; the built-in filters DoubleWeave, FlipVertical, AddBorders, PeculiarBlend, StackVertical, StackHorizontal, and ShowFiveVersions all use it to do their dirty work.

In AddBorders it’s to copy the Y-plane from the source to the destination frame (for planar formats): 

const int initial_black = top*dst_pitch + vi.BytesFromPixels(left);
if (vi.IsPlanar()) {
  BitBlt(dstp+initial_black, dst_pitch, srcp, src_pitch, src_row_size, src_height);
  ...
}

left is the number of pixels which is added to the left, top the number which is added to the top. So the first source pixel, srcp[0], is copied to its new location dstp[x], and so on. The remaining bytes are zeroed and can be refilled later on.

AtExit 

 virtual void AtExit(ShutdownFunc function, void* user_data);

When IScriptEnvironment is deleted on script close the AtExit functions get run. When you register the function you can optionally provide some user data. When the function is finally called this data will be provided as the argument to the procedure.

The example below (thanks to tsp) loads a library and automatically unloads it (by using AtExit) after the script is closed. It can be useful when your plugin depends on a library and you want to load the library in your script (the plugin fft3dfilter.dll depends on the library fftw3.dll for example): 

void __cdecl UnloadDll(void* hinst, IScriptEnvironment* env) {
  if (hinst)
    FreeLibrary(static_cast<HMODULE>(hinst)); 
}

AVSValue __cdecl LoadDll(AVSValue args, void* user_data, IScriptEnvironment* env){
  HMODULE hinst = 0;
  hinst = LoadLibrary(args[0].AsString()); // loads a library
  env->AtExit(UnloadDll, hinst); // calls UnloadDll to unload the library upon script exit
  return hinst!=NULL;
}

AddFunction 

 virtual void __stdcall AddFunction(const char* name, const char* params, ApplyFunc apply, void* user_data) = 0;

The main purpose of the AvisynthPluginInit2 function is to call env->AddFunction. 

env->AddFunction("Sepia", "c[color]i[mode]s", Create_Sepia, 0);

AddFunction is called to let Avisynth know of the existence of our filter. It just registers a function with Avisynth's internal function table. This function takes four arguments: the name of the new script function; the parameter-type string; the C++ function implementing the script function; and the user_data cookie.

The added function is of type AVSValue and can therefore return any AVSValue. Here are a few options how to return from the "added" function: 

AVSValue __cdecl  returnSomething(AVSValue args, void* user_data, IScriptEnvironment* env){
  
  char *strlit = "AnyOldName";
  int len = strlen(strlit);
  char *s = new char[len+1];
  if (s==NULL)
    env->ThrowError("Cannot allocate string mem");
  strcpy(s, strlit);                              // duplicate
  char *e = s+len;                                // point at null
  // make safe copy of string (memory is freed on Avisynth closure)
  AVSValue ret = env->SaveString(s,e-s);          // e-s is text len only (excl null) {SaveString uses memcpy)
  // AVSValue ret = env->SaveString(s);           // alternative, Avisynth uses strlen to ascertain length
  delete []s;                                     // delete our temp s buffer
  return ret;                                     // return saved string as AVSValue
  // return strlit;                               // alternative to MOST of above code char* converted to AVSValue.
  // return "AnyOldName";                         // alternative to ALL of above code char* converted to AVSValue.
  // String literals are read only and at constant address and so need not be saved.
}

MakeWritable 

 virtual bool __stdcall MakeWritable(PVideoFrame* pvf) = 0;

MakeWritable only copies the active part of the frame to a completely new frame with a default pitch. You need this to recieve a valid write pointer to an existing frame. 

PVideoFrame src = child->GetFrame(n, env);
env->MakeWritable(&src);

FunctionExists 

 virtual bool __stdcall FunctionExists(const char* name) = 0;

FunctionExists returns true if the specified filter exists, otherwise returns false: 

if (env->FunctionExists("Import")) {
  env->ThrowError("Yes, the IMPORT function exist.");
} else {
  env->ThrowError("No, the IMPORT function don't exist.");
}

GetVar 

 virtual AVSValue __stdcall GetVar(const char* name) = 0;

GetVar can be used to access AviSynth variables.

Internal and external (plugin) functions are, for example, exported as AviSynth variables:

  • $InternalFunctions$ Should contain a string consisting of function names of all internal functions.
  • $InternalFunctions!Functionname!Param$ Should contain all parameters for each internal function.
  • $PluginFunctions$ Should contain a string of all plugins in your autoloading plugin folder.
  • $Plugin!Functionname!Param$ Should contain all parameters.

Use env->GetVar() to access them. This example returns a string consisting of all parameters of ConvertToYV12: 

const char* plugin_dir;
plugin_dir = env->GetVar("$Plugin!ConverttoYV12!Param$").AsString();

This example returns the plugin folder which is used to autoload your plugins (and returns an error if it’s not set): 

try {
  const char* plugin_dir;
  plugin_dir = env->GetVar("$PluginDir$").AsString();
  env->ThrowError(plugin_dir);
} catch(...) {
  env->ThrowError("Plugin directory not set.");
}

If you are making a conditional filter you can use it to get the current framenumber: 

// Get current frame number
AVSValue cf = env->GetVar("current_frame");
if (!cf.IsInt())
  env->ThrowError("MinMaxAudio: This filter can only be used within ConditionalFilter");
int n = cf.AsInt();
PVideoFrame src = child->GetFrame(n, env);

SetVar 

 virtual bool __stdcall SetVar(const char* name, const AVSValue& val) = 0;

Return values: 

true if the variable was created and filled with the given value
false in case the variable was already there and we just updated its value

SetVar can be used to set/create AviSynth variables. The created variables are only visible in the local scope, e.g. script functions have a new scope.


This example sets the autoloading plugin folder to “C:\\” 

if(env->SetVar("$PluginDir$", AVSValue("C:\\"))) {
  //variable was created
} else {
  //variable was already existing and updated
}

This example sets variables in GetFrame which can be accessed later on in a script within the conditional environment: 

// saves the blue value of a pixel
int BlueValue;
BlueValue = srcp[x];
env->SetVar("BlueValue", AVSValue(BlueValue));

SetGlobalVar 

 virtual bool __stdcall SetGlobalVar(const char* name, const AVSValue& val) = 0;

Usage:

SetGlobalVar can be used to create or set AviSynth variables that are visible within global scope. It is possible that a single filter may want to use SetVar in order to exchange signals to possible other instances of itself.

There are at least 4 different components that make use of Set(Global)Var functions:

  • the core itself
  • the user within the avs script
  • filters/plugins
  • a custom application that invoked the environment

All of above may have their own requirements for the SetVar function. Some may want to be visible globally, others may not.


PushContext 

 virtual void __stdcall PushContext(int level=0) = 0;

?

PopContext 

 virtual void __stdcall PopContext() = 0;

?

PopContextGlobal 

 virtual void __stdcall PopContextGlobal() = 0;

?

NewVideoFrame 

 virtual PVideoFrame __stdcall NewVideoFrame(const VideoInfo& vi, int align=FRAME_ALIGN) = 0; // align should be 4 or 8 (default 16)

The NewVideoFrame callback allocates space for a video frame of the supplied size. (In this case it will hold our filter's output.) The frame buffer is uninitialized raw memory (except that in the debug build it gets filled with the repeating byte pattern 0A 11 0C A7 ED, which is easy to recognize because it looks like "ALLOCATED"). "vi" is a protected member of GenericVideoFilter. It is a structure of type VideoInfo, which contains information about the clip (like frame size, frame rate, pixel format, audio sample rate, etc.). NewVideoFrame uses the information in this structure to return a frame buffer of the appropriate size.

The following example creates a new VideoInfo structure and creates a new video frame from it: 

 VideoInfo vi;
 PVideoFrame frame;
 memset(&vi, 0, sizeof(VideoInfo));
 vi.width = 640;
 vi.height = 480;
 vi.fps_numerator = 30000;
 vi.fps_denominator = 1001;
 vi.num_frames = 107892;   // 1 hour
 vi.pixel_type = VideoInfo::CS_BGR32;
 vi.sample_type = SAMPLE_FLOAT;
 vi.nchannels = 2;
 vi.audio_samples_per_second = 48000;
 vi.num_audio_samples = vi.AudioSamplesFromFrames(vi.num_frames);
 frame = env->NewVideoFrame(vi);

CheckVersion 

 virtual void __stdcall CheckVersion(int version = AVISYNTH_INTERFACE_VERSION) = 0;

CheckVersion checks the interface version (avisynth.h). It throws an error if ‘version’ is bigger than the used interface version. The following interface versions are in use:

AVISYNTH_INTERFACE_VERSION = 1 (v1.0-v2.0.8), 2 (v2.5.0-v2.5.5), 3 (v2.5.6-v2.5.8), or 5 (v2.6.0) [version 4 doesn’t exist].

This example will throw an error if v2.5x or an older AviSynth version is being used: 

env->CheckVersion(5)

This can be used in a plugin, for example, if it needs at least a certain interface version for it to work.

Subframe 

 virtual PVideoFrame __stdcall Subframe(PVideoFrame src, int rel_offset, int new_pitch, int new_row_size, int new_height) = 0;

Subframe (for interleaved formats) extracts a part of a video frame. For planar formats use SubframePlanar. For examples see SubframePlanar.

SubframePlanar 

 virtual PVideoFrame __stdcall SubframePlanar(PVideoFrame src, int rel_offset, int new_pitch, int new_row_size, int new_height, int rel_offsetU, int rel_offsetV, int new_pitchUV) = 0;

SubframePlanar (for planar formats) extracts a part of a video frame. The example below returns the first field of a frame: 

vi.height >>= 1; // sets new height in the constructor

PVideoFrame frame = child->GetFrame(n, env);
if (vi.IsPlanar()) { // SubframePlanar works on planar formats only
  const int frame_pitch = frame->GetPitch(PLANAR_Y);
  const int frame_width = frame->GetRowSize(PLANAR_Y);
  const int frame_height = frame->GetHeight(PLANAR_Y);
  const int frame_pitchUV = frame->GetPitch(PLANAR_U);
  return env->SubframePlanar(frame, 0, 2*frame_pitch, frame_width, frame_height>>1, 0, 0, 2*frame_pitchUV);
}

Note that it copies the first row of pixels and moves on to the third row (by moving the offset by ‘2*frame_pitch’). After frame_height/2 it stops reading.

The following example keeps the left 100 pixels of a clip (it leaves the height unaltered) and throws away the rest: 

vi.width = 100; // sets new width in the constructor

PVideoFrame frame = child->GetFrame(n, env);
if (vi.IsPlanar()) { // SubframePlanar works on planar formats only
  const int frame_pitch = frame->GetPitch(PLANAR_Y);
  const int frame_height = frame->GetHeight(PLANAR_Y);
  const int frame_pitchUV = frame->GetPitch(PLANAR_U);
  return env->SubframePlanar(frame, 0, frame_pitch, 100, frame_height, 0, 0, frame_pitchUV);
}

Note that it copies 100 pixels and moves on to the next row (by moving the offset by ‘frame_pitch’).

You need to check somewhere that the source frames is more than 100 pixels wide, otherwise throw an error.

SetMemoryMax 

 virtual int __stdcall SetMemoryMax(int mem) = 0;

There is a builtin cache automatically inserted in between all filters. You can use SetmemoryMax to increase the size.

SetMemoryMax only sets the size of the frame buffer cache. It is independent of any other memory allocation. Memory usage due to the frame cache should ramp up pretty quickly to the limited value and stay there. Setting a lower SetMemoryMax value will make more memory available for other purposes and provide less cache buffer frames. It is pointless having more buffers available than are needed by the scripts temporal requirements. If each and every frame generated at each and every stage of a script is only ever used once then the cache is entirely useless. By definition a cache is only useful if a generated element is needed a second or subsequent time.

SetWorkingDir 

 virtual int __stdcall SetWorkingDir(const char * newdir) = 0;

Sets the default directory for AviSynth.

DeleteScriptEnvironment, v5 

 virtual void __stdcall DeleteScriptEnvironment() = 0;

Provides a method to delete the ScriptEnvironment which is created with CreateScriptEnvironment.

ApplyMessage, v5 

 virtual void _stdcall ApplyMessage(PVideoFrame* frame, const VideoInfo& vi, const char* message, int size, int textcolor, int halocolor, int bgcolor) = 0;

ApplyMessage writes text on a frame. For example: 

char BUF[256];
PVideoFrame src = child->GetFrame(n, env);
env->MakeWritable(&src);
sprintf(BUF, "Filter: Frame %d is processed.", n);
env->ApplyMessage(&src, vi, BUF, vi.width/4, 0xf0f080, 0, 0);

PVideoFrame

PVideoFrame is a smart pointer to VideoFrame.

In this example it gives a pointer to frame ‘n’ from child: 

PVideoFrame src = child->GetFrame(n, env);

"child" is a protected member of GenericVideoFilter, of type PClip. It contains the clip that was passed to the constructor. For our filter to produce frame n we need the corresponding frame of the input. If you need a different frame from the input, all you have to do is pass a different frame number to child->GetFrame.

In this example it gives a pointer to a new created VideoFrame from vi (which is a VideoInfo structure): 

PVideoFrame dst = env->NewVideoFrame(vi);

"vi" is another protected member of GenericVideoFilter (the only other member, actually). It is a structure of type VideoInfo, which contains information about the clip (like frame size, frame rate, pixel format, audio sample rate, etc.). NewVideoFrame uses the information in this struct to return a frame buffer of the appropriate size.

IClip

An Avisynth filter is simply a C++ class implementing the IClip interface. IClip has four pure virtual methods: GetVideoInfo, GetFrame, GetParity, and GetAudio. The class GenericVideoFilter is a simple do-nothing filter defined in avisynth.h. It derives from IClip and implements all four methods. Most filters can inherit from GenericVideoFilter rather than directly from IClip; this saves you from having to implement methods that you don't care about, like GetAudio.

IClip has the following members: GetVersion, GetFrame, GetParity, GetAudio, SetCacheHints and GetVideoInfo. They are described in the following subsections.

GetVersion 

 virtual int __stdcall GetVersion() { return AVISYNTH_INTERFACE_VERSION; }

GetVersion returns the interface version of the loaded avisynth.dll:

AVISYNTH_INTERFACE_VERSION = 1 (v1.0-v2.0.8), 2 (v2.5.0-v2.5.5), 3 (v2.5.6-v2.5.8), or 5 (v2.6.0) [version 4 doesn’t exist].

GetFrame 

 virtual PVideoFrame __stdcall GetFrame(int n, IScriptEnvironment* env) = 0;

GetFrame returns a video frame. In this example, the even frames (0, 2, 4, ...) of ‘child’ are returned: 

PVideoFrame src = child->GetFrame(2*n, env);

You should do all the GetFrame() calls BEFORE you get any pointers and start manipulating any data.

GetParity 

 virtual bool __stdcall GetParity(int n) = 0;

GetParity returns the field parity if the clip is field-based, otherwise it returns the parity of first field of a frame. In other words, it distinguishes between top field first (TFF) and bottom field first (BFF). When it returns true, it means that this frame should be considered TFF, otherwise it should be considered BFF.

GetAudio 

 virtual void __stdcall GetAudio(void* buf, __int64 start, __int64 count, IScriptEnvironment* env) = 0;

Audio processing is handled through the GetAudio method. You must fill in the buffer with count samples beginning at the sample start. A sample may vary from one to four bytes, depending on whether the audio is 8, 16, 24 or 32-bit (float is also 32-bit). The flag vi.SampleType() will tell you this.

If you cannot do your audio processing in-place, you must allocate your own buffer for the source audio using new or malloc.

In this example, the audio of frame ‘n’ is returned (in the buffer ‘samples’): 

VideoInfo vi = child->GetVideoInfo();
SFLOAT* samples = new SFLOAT[count*vi.AudioChannels()];
PVideoFrame src = child->GetFrame(n, env);
const __int64 start = vi.AudioSamplesFromFrames(n);
const __int64 count = vi.AudioSamplesFromFrames(1);
child->GetAudio(samples, max(0,start), count, env);

SetCacheHints 

 int __stdcall SetCacheHints(int cachehints, int frame_range) = 0 ;  // We do not pass cache requests upwards, only to the next filter.

SetCacheHints should be used in filters that request multiple frames from any single PClip source per input GetFrame call. frame_range is maximal 21.

The possible values of cachehints are: 

CACHE_NOTHING=0  // Filter requested no caching.
CACHE_RANGE=1  // An explicit cache of "frame_range" frames around the current frame.
CACHE_ALL=2  // This is default operation, a simple LRU cache.
CACHE_AUDIO=3  // Audio caching.
CACHE_AUDIO_NONE=4  // Filter requested no audio caching.
CACHE_AUDIO_AUTO=5  // Audio caching (difference with CACHE_AUDIO?).

When caching video frames (cachehints=0, 1, 2), frame_range is the radius around the current frame. When caching audio samples (cachehints=3, 4, 5), the value 0 creates a default buffer of 64kb and positive values allocate frame_range bytes for the cache.

E.g. If you have a single PClip source, i.e. child and you get asked for frame 100 and you in turn then ask for frames 98, 99, 100, 101 and 102 then you need to call CACHE_RANGE with frame_range set to 3: 

child->SetCacheHints(CACHE_RANGE, 3);

Frames outside the specified radius are candidate for normal LRU caching.

Note, SetCacheHints was of type void for v3 and older versions.

GetVideoInfo 

 virtual const VideoInfo& __stdcall GetVideoInfo() = 0;

GetVideoInfo returns a VideoInfo structure. [link]

PClip

PClip is a smart pointer to an IClip, and IClip is a generic abstract class.. It maintains a reference count on the IClip object and automagically deletes it when the last PClip referencing it goes away. For obvious reasons, you should always use PClip rather than IClip* to refer to clips.

Like a genuine pointer, a PClip is only four bytes long, so you can pass it around by value. Also like a pointer, a PClip can be assigned a null value (0), which is often useful as a sentinel. Unlike a pointer,

PClip is initialized to 0 by default.

You'll need to make sure your class doesn't contain any circular PClip references, or any PClips sitting in dynamically allocated memory that you forget to delete. Other than that, you don't have to worry about the reference-counting machinery.

AviSynth filters have a standardized output channel via IClip, but (unlike VirtualDub filters) no standardized input channel. Each filter is responsible for obtaining its own source material -- usually (as in this case) from another clip, but sometimes from several different clips, or from a file.

The clip functionality must be provided by some concrete subclass of IClip which implements the functions GetFrame(), etc. So you cannot create a PClip without having an appropriate IClip subclass. For most filters, the GenericVideoFilter class provides the basis for this, but 'source' filters (which is basically what you have) do not have a parent clip and so GenericVideoFilter is not appropriate.

AVSValue

AVSValue is a variant type which can hold any one of the following types: a boolean value (true/false); an integer; a floating-point number; a string; a video clip (PClip); an array of AVSValues; or nothing ("undefined").

It holds an array of AVSValues in the following way:

AVSValue(const AVSValue* a, int size) { type = 'a'; array = a; array_size = size; }

For example: 

AVSValue up_args[3] = {child, 384, 288}; 
PClip resized = env->Invoke("LanczosResize", AVSValue(up_args,3)).AsClip();

Note that 

AVSValue(up_args,3)

returns the following: {'a'; {child, 384, 288}; 3}.

Also Invoke returns an AVSValue (see its declaration) which in that case is a PClip.

Structures

The following structure is available: VideoInfo structure. It holds global information about a clip (i.e. information that does not depend on the framenumber). The GetVideoInfo method in IClip returns this structure. A description (for AVISYNTH_INTERFACE_VERSION=5) of it can be found here.

Constants

The following constants are defined in avisynth.h: 

// Audio Sample information
typedef float SFLOAT;

enum { // sample types
  SAMPLE_INT8  = 1<<0,
  SAMPLE_INT16 = 1<<1,
  SAMPLE_INT24 = 1<<2,
  SAMPLE_INT32 = 1<<3,
  SAMPLE_FLOAT = 1<<4
};

enum {
  PLANAR_Y         = 1<<0,
  PLANAR_U         = 1<<1,
  PLANAR_V         = 1<<2,
  PLANAR_ALIGNED   = 1<<3,
  PLANAR_Y_ALIGNED = PLANAR_Y | PLANAR_ALIGNED,
  PLANAR_U_ALIGNED = PLANAR_U | PLANAR_ALIGNED,
  PLANAR_V_ALIGNED = PLANAR_V | PLANAR_ALIGNED,
  PLANAR_A         = 1<<4,                       // v5
  PLANAR_R         = 1<<5,                       // v5
  PLANAR_G         = 1<<6,                       // v5
  PLANAR_B         = 1<<7,                       // v5
  PLANAR_A_ALIGNED = PLANAR_A | PLANAR_ALIGNED,  // v5
  PLANAR_R_ALIGNED = PLANAR_R | PLANAR_ALIGNED,  // v5
  PLANAR_G_ALIGNED = PLANAR_G | PLANAR_ALIGNED,  // v5
  PLANAR_B_ALIGNED = PLANAR_B | PLANAR_ALIGNED,  // v5
};
Retrieved from "http://avisynth.nl/index.php?title=Filter_SDK/Cplusplus_API&oldid=2097"
Personal tools