The CreateBitmap of the interface does not support when the pixelFormat is a native pixel format provided by Windows Imaging Component (WIC).

The native JPEG encoder uses .

inherits from and therefore also inherits the CopyPixels method. When pixels need to be moved to a new memory location, CopyPixels is often the most efficient.

Because of to the internal memory representation implied by the , in-place modification and processing using the Lock is more efficient than CopyPixels, usually reducing to a simple reference access directly into the memory owned by the bitmap rather than a as a copy. This is contrasted to procedural bitmaps which implement only CopyPixels because there is no internal memory representation and one would need to be created on demand to satisfy a call to Lock.

This interface provides a common way of accessing and linking together bitmaps, decoders, format converters, and scalers. Components that implement this interface can be connected together in a graph to pull imaging data through.

This interface defines only the notion of readability or being able to produce pixels. Modifying or writing to a bitmap is considered to be a specialization specific to bitmaps which have storage and is defined in the descendant interface .

The pixel format returned by this method is not necessarily the pixel format the image is stored as. The codec may perform a format conversion from the storage pixel format to an output pixel format.

Some formats, such as GIF and ICO, do not have full DPI support. For GIF, this method calculates the DPI values from the aspect ratio, using a base DPI of (96.0, 96.0). The ICO format does not support DPI at all, and the method always returns (96.0,96.0) for ICO images.

Additionally, WIC itself does not transform images based on the DPI values in an image. It is up to the caller to transform an image based on the resolution returned.

If the is an , the function may return the image's global palette if a frame-level palette is not available. The global palette may also be retrieved using the CopyPalette method.

CopyPixels is one of the two main image processing routines (the other being Lock) triggering the actual processing. It instructs the object to produce pixels according to its algorithm - this may involve decoding a portion of a JPEG stored on disk, copying a block of memory, or even analytically computing a complex gradient. The algorithm is completely dependent on the object implementing the interface.

The caller can restrict the operation to a rectangle of interest (ROI) using the prc parameter. The ROI sub-rectangle must be fully contained in the bounds of the bitmap. Specifying a null ROI implies that the whole bitmap should be returned.

The caller controls the memory management and must provide an output buffer (pbBuffer) for the results of the copy along with the buffer's bounds (cbBufferSize). The cbStride parameter defines the count of bytes between two vertically adjacent pixels in the output buffer. The caller must ensure that there is sufficient buffer to complete the call based on the width, height and pixel format of the bitmap and the sub-rectangle provided to the copy method.

If the caller needs to perform numerous copies of an expensive such as a JPEG, it is recommended to create an in-memory first.

The callee must only write to the first (prc->Width*bitsperpixel+7)/8 bytes of each line of the output buffer (in this case, a line is a consecutive string of cbStride bytes).

The pixel format returned by this method is not necessarily the pixel format the image is stored as. The codec may perform a format conversion from the storage pixel format to an output pixel format.

Locks are exclusive for writing but can be shared for reading. You cannot call CopyPixels while the is locked for writing. Doing so will return an error, since locks are exclusive.

This method has no effect on the actual pixels or samples stored in the bitmap. Instead the interpretation of the sampling rate is modified. This means that a 96 DPI image which is 96 pixels wide is one inch. If the physical resolution is modified to 48 DPI, then the bitmap is considered to be 2 inches wide but has the same number of pixels. If the resolution is less than REAL_EPSILON (1.192092896e-07F) the error code WINCODEC_ERR_INVALIDPARAMETER is returned.

Signing is unused by WIC. Therefore, all components .

This function can be used to determine whether a component has no binary component or has been added to the disabled components list in the registry.

If cchAuthor is 0 and wzAuthor is null, the required buffer size is returned in pccchActual.

All built-in components return "1.0.0.0", except for pixel formats, which do not have a version.

If cchAuthor is 0 and wzAuthor is null, the required buffer size is returned in pccchActual.

All built-in components return "1.0.0.0", except for pixel formats, which do not have a spec version.

If cchAuthor is 0 and wzAuthor is null, the required buffer size is returned in pccchActual.

If cchFriendlyName is 0 and wzFriendlyName is null, the required buffer size is returned in pccchActual.

Signing is unused by WIC. Therefore, all components .

This function can be used to determine whether a component has no binary component or has been added to the disabled components list in the registry.

The usage pattern for this method is a two call process. The first call retrieves the array size needed to retrieve all the supported pixel formats by calling it with cFormats set to 0 and pguidPixelFormats set to null. This call sets pcActual to the array size needed. Once the needed array size is determined, a second GetPixelFormats call with pguidPixelFormats set to an array of the appropriate size will retrieve the pixel formats.

The usage pattern for this method is a two call process. The first call retrieves the buffer size needed to retrieve the full color management version number by calling it with cchColorManagementVersion set to 0 and wzColorManagementVersion set to null. This call sets pcchActual to the buffer size needed. Once the needed buffer size is determined, a second GetColorManagementVersion call with cchColorManagementVersion set to the buffer size and wzColorManagementVersion set to a buffer of the appropriate size will retrieve the pixel formats.

The usage pattern for this method is a two call process. The first call retrieves the buffer size needed to retrieve the full color management version number by calling it with cchDeviceManufacturer set to 0 and wzDeviceManufacturer set to null. This call sets pcchActual to the buffer size needed. Once the needed buffer size is determined, a second GetDeviceManufacturer call with cchDeviceManufacturer set to the buffer size and wzDeviceManufacturer set to a buffer of the appropriate size will retrieve the pixel formats.

The usage pattern for this method is a two call process. The first call retrieves the buffer size needed to retrieve the full color management version number by calling it with cchDeviceModels set to 0 and wzDeviceModels set to null. This call sets pcchActual to the buffer size needed. Once the needed buffer size is determined, a second GetDeviceModels call with cchDeviceModels set to the buffer size and wzDeviceModels set to a buffer of the appropriate size will retrieve the pixel formats.

The usage pattern for this method is a two call process. The first call retrieves the buffer size needed to retrieve the full color management version number by calling it with cchMimeTypes set to 0 and wzMimeTypes set to null. This call sets pcchActual to the buffer size needed. Once the needed buffer size is determined, a second GetMimeTypes call with cchMimeTypes set to the buffer size and wzMimeTypes set to a buffer of the appropriate size will retrieve the pixel formats.

The default extension for an image encoder is the first item in the list of returned extensions.

The usage pattern for this method is a two call process. The first call retrieves the buffer size needed to retrieve the full color management version number by calling it with cchFileExtensions set to 0 and wzFileExtensions set to null. This call sets pcchActual to the buffer size needed. Once the needed buffer size is determined, a second GetFileExtensions call with cchFileExtensions set to the buffer size and wzFileExtensions set to a buffer of the appropriate size will retrieve the pixel formats.

Applications can only register a single callback. Subsequent registration calls will replace the previously registered callback. To unregister a callback, pass in null or register a new callback function.

Progress is reported in an increasing order between 0.0 and 1.0. If dwProgressFlags includes , the callback is guaranteed to be called with progress 0.0. If dwProgressFlags includes , the callback is guaranteed to be called with progress 1.0.

increases the frequency in which the callback is called. If an operation is expected to take more than 30 seconds, should be added to dwProgressFlags.

There are a number of concrete implemenations of this interface representing each of the standard decoders provided by the platform including bitmap (BMP), Portable Network Graphics (PNG), icon (ICO), Joint Photographic Experts Group (JPEG), Graphics Interchange Format (GIF), Tagged Image File Format (TIFF), and Microsoft?Windows Digital Photo (WDP). The following table includes the class identifier (CLSID) for each native decoder.

?

This interface may be sub-classed to provide support for third party codecs as part of the extensibility model. See the AITCodec Sample CODEC.

Codecs written as TIFF container formats that are not register will decode as a TIFF image. Client applications should check for a zero frame count to determine if the codec is valid.

Custom decoder implementations should save the current position of the specified , read whatever information is necessary in order to determine which capabilities it can provide for the supplied stream, and restore the stream position.

CopyPalette returns a global palette (a palette that applies to all the frames in the image) if there is one; otherwise, it returns WINCODEC_ERR_PALETTEUNAVAILABLE. If an image doesn't have a global palette, it may still have a frame-level palette, which can be retrieved using IWICBitmapFrameDecode::CopyPalette.

Not all formats support previews. Only the native Microsoft?Windows Digital Photo (WDP) codec support previews.

None of the native formats support global thumbnails. Joint Photographic Experts Group (JPEG), Tagged Image File Format (TIFF), and Microsoft?Windows Digital Photo (WDP) support frame level thumbnails that can be accessed through a frame's GetThumbnail method.

Not all formats support previews. Only the native Microsoft?Windows Digital Photo (WDP) codec support previews.

None of the native formats support global thumbnails. Joint Photographic Experts Group (JPEG), Tagged Image File Format (TIFF), and Microsoft?Windows Digital Photo (WDP) support frame level thumbnails that can be accessed through a frame's GetThumbnail method.

To retrieve all pattern signatures, this method should first be called with pPatterns set to null to retrieve the actual buffer size needed through pcbPatternsActual. Once the needed buffer size is known, allocate a buffer of the needed size and call GetPatterns again with the allocated buffer.

There are a number of concrete implemenations of this interface representing each of the standard encoders provided by the platform including bitmap (BMP), Portable Network Graphics (PNG), Joint Photographic Experts Group (JPEG), Graphics Interchange Format (GIF), Tagged Image File Format (TIFF), and Microsoft?Windows Digital Photo (WDP). The following table includes the class identifier (CLSID) for each native encoder.

?

Additionally this interface may be sub-classed to provide support for third party codecs as part of the extensibility model. See the AITCodec Sample CODEC.

The parameter ppIEncoderOptions can be used to receive an that can then be used to specify encoder options. This is done by passing a reference to a null reference in ppIEncoderOptions. You should then set your desired encoder options on the returned, and pass this to .

Note??Do not pass in a reference to an initialized . The reference will be overwritten, and the original will not be freed.

Otherwise, you can pass null in ppIEncoderOptions if you do not intend to specify encoder options.

See Encoding Overview for an example of how to set encoder options.

To finalize an image, both the frame Commit and the encoder Commit must be called. However, only call the encoder Commit method after all frames have been committed.

IWICBitmapFipRotator requests data on a per-pixel basis, while WIC codecs provide data on a per-scanline basis. This causes the fliprotator object to exhibit n2 behavior if there is no buffering. This occures because each pixel in the transformed image requires an entire scanline to be decoded in the file. It is recommended that you buffer the image using , or flip/rotate the image using Direct2D.

Not all formats support thumbnails. Joint Photographic Experts Group (JPEG), Tagged Image File Format (TIFF), and Microsoft?Windows Digital Photo (WDP) support thumbnails.

If the codec does not support thumbnails, return WINCODEC_ERROR_CODECNOTHUMBNAIL rather than E_NOTIMPL.

Not all formats support thumbnails. Joint Photographic Experts Group (JPEG), Tagged Image File Format (TIFF), and Microsoft?Windows Digital Photo (WDP) support thumbnails.

If the codec does not support thumbnails, return WINCODEC_ERROR_CODECNOTHUMBNAIL rather than E_NOTIMPL.

This method does not fail if called on a frame whose pixel format is set to a non-indexed pixel format. The target pixel format is a non-indexed format, the palette will be ignored.

SetThumbnail should be called before calling WritePixels or WriteSource. The thumbnail will not be added to the encoded file if SetThumbnail after a call to WritePixels or WriteSource.

Successive WritePixels calls are assumed to be sequential scanline access in the output image.

If SetSize is not called prior to calling WriteSource, the size given in prc is used if not null. Otherwise, the size of the given in pIBitmapSource is used.

If SetPixelFormat is not called prior to calling WriteSource, the pixel format of the given in pIBitmapSource is used.

If SetResolution is not called prior to calling WriteSource, the pixel format of pIBitmapSource is used.

If SetPalette is not called prior to calling WriteSource, the target pixel format is indexed, and the pixel format of pIBitmapSource matches the encoder frame's pixel format, then the pIBitmapSource pixel format is used.

When encoding a GIF image, if the global palette is set and the frame level palette is not set directly by the user or by a custom independent software vendor (ISV) GIF codec, WriteSource will use the global palette to encode the frame even when pIBitmapSource has a frame level palette.

Windows Vista:The source rect width must match the width set through SetSize. Repeated WriteSource calls can be made as long as the total accumulated source rect height is the same as set through SetSize.

To finalize the image, both the frame Commit and the encoder Commit must be called. However, only call the encoder Commit method after all frames have been committed.

This method does not fail if called on a frame whose pixel format is set to a non-indexed pixel format. The target pixel format is a non-indexed format, the palette will be ignored.

SetThumbnail should be called before calling WritePixels or WriteSource. The thumbnail will not be added to the encoded file if SetThumbnail after a call to WritePixels or WriteSource.

The bitmap lock is simply an abstraction for a rectangular memory window into the bitmap. For the simplest case, a system memory bitmap, this is simply a reference to the top left corner of the rectangle and a stride value.

To release the exclusive lock set by Lock method and the associated object, call IUnknown::Release on the object.

Note the stride value is specific to the , not the bitmap. For example, two consecutive locks on the same rectangle of a bitmap may return different references and stride values, depending on internal implementation.

The reference provided by this method should not be used outside of the lifetime of the lock itself.

GetDataPointer is not available in multi-threaded apartment applications.

Note the stride value is specific to the , not the bitmap. For example, two consecutive locks on the same rectangle of a bitmap may return different references and stride values, depending on internal implementation.

Images can be scaled to larger sizes; however, even with sophisticated scaling algorithms, there is only so much information in the image and artifacts tend to worsen the more you scale up.

The scaler will reapply the resampling algorithm every time CopyPixels is called. If the scaled image is to be animated, the scaled image should be created once and cached in a new bitmap, after which the may be released. In this way the scaling algorithm - which may be computationally expensive relative to drawing - is performed only once and the result displayed many times.

The scaler is optimized to use the minimum amount of memory required to scale the image correctly. The scaler may be used to produce parts of the image incrementally (banding) by calling CopyPixels with different rectangles representing the output bands of the image. Resampling typically requires overlapping rectangles from the source image and thus may need to request the same pixels from the source bitmap multiple times. Requesting scanlines out-of-order from some image decoders can have a significant performance penalty. Because of this reason, the scaler is optimized to handle consecutive horizontal bands of scanlines (rectangle width equal to the bitmap width). In this case the accumulator from the previous vertically adjacent rectangle is re-used to avoid duplicate scanline requests from the source. This implies that banded output from the scaler may have better performance if the bands are requested sequentially. Of course if the scaler is simply used to produce a single rectangle output, this concern is eliminated because the scaler will internally request scanlines in the correct order.

For codec developer implementation details for this method, see Implementing .

When multiple transform operations are requested, the result is dependent on the order in which the operations are performed. To ensure predictability and consistency across CODECs, it's important that all CODECs perform these operations in the same order. The recommended order of these operations is:

1. Scale
2. Crop
3. Flip/Rotate

Pixel format conversion can be performed at any time, since it has no effect on the other transforms.

The first parameter, prc is used to specify the region of interest for clipping the image. By convention, scaling is performed before clipping so, if the image is to be scaled as well as clipped, the region of interest should be determined after the image has been scaled.

If a dstTransform is specified, the stride is the transformed stride, and is based on the pixelFormat specified in the CopyPixels call, not the original frame's pixel format.

A Color Context is an abstraction for a color profile. The profile can be loaded from a file (ie. "sRGB Color Space Profile.icm") or from a memory buffer obtained by reading. The color profile directory can be obtained by calling the GetColorDirectory API (See http://msdn.microsoft.com/library/en-us/icm/icm_58xl.asp).

A is an imaging pipeline component that knows how to pull pixels obtained from a given through a color transform. The color transform is defined by mapping colors from the source color context to the destination color context in a given output pixel format.

It is recommended that a codec report that a capability is supported even if the results at the outer range limits are not of perfect quality.

It is recommended that a codec report that this method is supported even if the results at the outer range limits are not of perfect quality.

Due to other white point setting methods (e.g. SetWhitePointKelvin), care must be taken by codec implementers to ensure proper interoperability. For instance, if the caller sets via a named white point then the codec implementer may whis to disable reading back the correspoinding Kelvin temperature. In specific cases where the codec implementer wishes to deny a given action because of previous calls, WINCODEC_ERR_WRONGSTATE should be returned.

If the named white points are not supported by the raw image or the raw file contains named white points that are not supported by this API, the codec implementer should still mark this capability as supported.

If the named white points are not supported by the raw image, a best effort should be made to adjust the image to the named white point even when it isn't a pre-defined white point of the raw file.

If the raw file containes named white points not supported by this API, the codec implementer should support the named white points in the API.

Due to other white point setting methods (e.g. SetWhitePointKelvin), care must be taken by codec implementers to ensure proper interoperability. For instance, if the caller sets via a named white point then the codec implementer may whis to disable reading back the correspoinding Kelvin temperature. In specific cases where the codec implementer wishes to deny a given action because of previous calls, WINCODEC_ERR_WRONGSTATE should be returned.

If the named white points are not supported by the raw image or the raw file contains named white points that are not supported by this API, the codec implementer should still mark this capability as supported.

If the named white points are not supported by the raw image, a best effort should be made to adjust the image to the named white point even when it isn't a pre-defined white point of the raw file.

If the raw file containes named white points not supported by this API, the codec implementer should support the named white points in .

Codec implementers should faithfully adjust the color temperature within the range supported natively by the raw image. For values outside the native support range, the codec implementer should provide a best effort representation of the image at that color temperature.

Codec implementers should return WINCODEC_ERR_VALUEOUTOFRANGE if the value is out of defined acceptable range.

Codec implementers must ensure proper interoperability with other white point setting methods such as SetWhitePointRGB. For example, if the caller sets the white point via SetNamedWhitePoint then the codec implementer may want to disable reading back the correspoinding Kelvin temperature. In specific cases where the codec implementer wants to deny a given action because of previous calls, WINCODEC_ERR_WRONGSTATE should be returned.

The codec implementer must determine what the upper range value represents and must determine how to map the value to their image processing routines.

The codec implementer must determine what the upper range value represents and must determine how to map the value to their image processing routines.

The codec implementer must determine what the upper range value represents and must determine how to map the value to their image processing routines.

The codec implementer must determine what the outer range values represent and must determine how to map the values to their image processing routines.

The codec implementer must determine what the upper range value represents and must determine how to map the value to their image processing routines.

If the named white points are not supported by the raw image or the raw file contains named white points that are not supported by this API, the codec implementer should still mark this capability as supported.

If the named white points are not supported by the raw image, a best effort should be made to adjust the image to the named white point even when it isn't a pre-defined white point of the raw file.

If the raw file containes named white points not supported by this API, the codec implementer should support the named white points in .

A decoder must be created using the value to perform in-place metadata updates. Using the option causes the decoder to release the file stream necessary to perform the metadata updates.

Not all metadata formats support fast metadata encoding. The native metadata handlers that support metadata are IFD, Exif, XMP, and GPS.

If a fast metadata encoder fails, the image will need to be fully re-encoded to add the metadata.

If the commit fails and returns WINCODEC_ERR_STREAMNOTAVAILABLE, ensure that the image decoder was loaded using the option. A fast metadata encoder is not supported when the decoder is created using the option.

If the commit fails for any reason, you will need to re-encode the image to ensure the new metadata is added to the image.

If you do not have a predefined palette, you must first create one. Use InitializeFromBitmap to create the palette object, then pass it in along with your other parameters.

dither, pIPalette, alphaThresholdPercent, and paletteTranslate are used to mitigate color loss when converting to a reduced bit-depth format. For conversions that do not need these settings, the following parameters values should be used: dither set to , pIPalette set to null, alphaThresholdPercent set to 0.0f, and paletteTranslate set to .

The basic algorithm involved when using an ordered dither requires a fixed palette, found in the enumeration, in a specific order. Often, the actual palette provided for the output may have a different ordering or some slight variation in the actual colors. This is the case when using the Microsoft?Windows palette which has slight differences among versions of Windows. To provide for this, a palette and a palette translation are given to the format converter. The pIPalette is the actual destination palette to be used and the paletteTranslate is a fixed palette. Once the conversion is complete, the colors are mapped from the fixed palette to the actual colors in pIPalette using a nearest color matching algorithm.

If colors in pIPalette do not closely match those in paletteTranslate, the mapping may produce undesireable results.

can be useful in format conversions from 8-bit formats to 5- or 6-bit formats as there is no way to accurately convert color data.

selects the error diffusion algorithm and may be used with any palette. If an arbitrary palette is provided, WICBitmapPaletteCustom should be passed in as the paletteTranslate. Error diffusion often provides superior results compared to the ordered dithering algorithms especially when combined with the optimized palette generation functionality on the .

When converting a bitmap which has an alpha channel, such as a Portable Network Graphics (PNG), to 8bpp, the alpha channel is normally ignored. Any pixels which were transparent in the original bitmap show up as black in the final output because both transparent and black have pixel values of zero in the respective formats.

Some 8bpp content can contains an alpha color; for instance, the Graphics Interchange Format (GIF) format allows for a single palette entry to be used as a transparent color. For this type of content, alphaThresholdPercent specifies what percentage of transparency should map to the transparent color. Because the alpha value is directly proportional to the opacity (not transparency) of a pixel, the alphaThresholdPercent indicates what level of opacity is mapped to the fully transparent color. For instance, 9.8% implies that any pixel with an alpha value of less than 25 will be mapped to the transparent color. A value of 100% maps all pixels which are not fully opaque to the transparent color. Note that the palette should provide a transparent color. If it does not, the 'transparent' color will be the one closest to zero - often black.

The format converter does not necessarily guarantee symmetricality with respect to conversion; that is, a converter may be able to convert FROM a particular format without actually being able to convert TO a particular format. In order to test symmetricality, use CanConvert.

To determine the number of pixel formats a coverter can handle, set cFormats to 0 and pPixelFormatGUIDs to null. The converter will fill pcActual with the number of formats supported by that converter.

When a decoder is created using this method, the file handle must remain alive during the lifetime of the decoder.

Other values may be available for both guidContainerFormat and pguidVendor depending on the installed WIC-enabled encoders. The values listed are those that are natively supported by the operating system.

Component types must be enumerated seperately. Combinations of component types and are unsupported.

The native image formats provided by Windows Imaging Component (WIC) do not support metadata at the decoder level. WIC codecs only support metadata on image frames. To create a fast metadata encoder from an image frame, see the image factory's CreateFastMetadataEncoderFromFrameDecode method.

A metadata query reader uses metadata query expressions to access embedded metadata. For more information on the metadata query language, see the Metadata Query Language Overview.

If the query reader is relative to the top of the metadata hierarchy it will return an empty string.

If the query reader is relative to a nested metadata block this method will return the path to the current query reader.

GetMetadataByName uses metadata query expressions to access embedded metadata. For more information on the metadata query language, see the Metadata Query Language Overview.

If multiple blocks or items exist that are expressed by the same query expression, the first metadata block or item found will be returned.

If a metadata item is a nested metadata block it will be passed back as a VT_UNKNOWN; otherwise, the "name" of the property will be passed back as a VT_LPWSTR. The enumerator does not enumerate content within nested metadata blocks.

A metadata query writer uses metadata query expressions to set or remove metadata. For more information on the metadata query language, see the Metadata Query Language Overview.

SetMetadataByName uses metadata query expressions to remove metadata. For more information on the metadata query language, see the Metadata Query Language Overview.

If the value set is a nested metadata block then use variant type VT_UNKNOWN and pvarValue pointing to the of the new metadata block. The ordering of metadata items is at the discretion of the query writer since relative locations are not specified.

RemoveMetadataByName uses metadata query expressions to remove metadata. For more information on the metadata query language, see the Metadata Query Language Overview.

If the metadata item is a metadata block, it is removed from the metadata hierarchy.

If the is not WICBitmapPaletteCustom, then the colors are automatically generated based on the table above. If the user subsequently changes a color palette entry the WICBitmapPalette is set to Custom by that action.

InitializeFromBitmap's fAddTransparentColor parameter will add a transparent color to the end of the color collection if its size if less than 256, otherwise index 255 will be replaced with the transparent color. If a pre-defined palette type is used, it will change to BitmapPaletteTypeCustom since it no longer matches the predefined palette.

The palette interface is an auxiliary imaging interface in that it does not directly concern bitmaps and pixels; rather it provides indexed color translation for indexed bitmaps. For an indexed pixel format with M bits per pixels: (The number of colors in the palette) greater than 2^M.

Traditionally the basic operation of the palette is to provide a translation from a byte (or smaller) index into a 32bpp color value. This is often accomplished by a 256 entry table of color values.

If a transparent color is required, it should be provided as part of the custom entries.

The entry count is limited to 256.

The resulting palette contains the specified number of colors which best represent the colors present in the bitmap. The algorithm operates on the opaque RGB color value of each pixel in the reference bitmap and hence ignores any alpha values. If a transparent color is required, set the fAddTransparentColor parameter to TRUE and one fewer optimized color will be computed, reducing the colorCount, and a fully transparent color entry will be added.

WICBitmapPaletteCustom is used for palettes initialized from both InitializeCustom and InitializeFromBitmap. There is no distinction is made between optimized and custom palettes.

WICBitmapPaletteCustom is used for palettes initialized from both InitializeCustom and InitializeFromBitmap. There is no distinction is made between optimized and custom palettes.

Images can only be progressively decoded if they were progressively encoded. The native encoders supplied by Windows Imaging Component (WIC) do not

E_NOTIMPL is returned if the codec does not support progressive level decoding.

Users should not use this function to iterate through the progressive levels of a progressive JPEG image. JPEG progressive levels are determined by the image and do not have a fixed level count. Using this method will force the application to wait for all progressive levels to be downloaded before it can return. Instead, applications should use the following code to iterate through the progressive levels of a progressive JPEG image.

A call does not have to request every level supported. If a caller requests level 1, without having previously requested level 0, the bits returned by the next call to CopyPixels will include both levels.

Users should not use this function to iterate through the progressive levels of a progressive JPEG image. JPEG progressive levels are determined by the image and do not have a fixed level count. Using this method will force the application to wait for all progressive levels to be downloaded before it can return. Instead, applications should use the following code to iterate through the progressive levels of a progressive JPEG image.

Decoders and metadata handlers are expected to create sub streams of whatever stream they hold when handing off control for embedded metadata to another metadata handler. If the stream is not restricted then use MAXLONGLONG as the max size and offset 0.

The interface methods do not enable you to provide a file sharing option. To create a file stream for an image, use the SHCreateStreamOnFileEx function. This stream can then be used to create an using the CreateDecoderFromStream method.