Sdcb.SimdCodecs.Bmp 1.2.0

Prefix Reserved

dotnet add package Sdcb.SimdCodecs.Bmp --version 1.2.0

NuGet\Install-Package Sdcb.SimdCodecs.Bmp -Version 1.2.0

This command is intended to be used within the Package Manager Console in Visual Studio, as it uses the NuGet module's version of Install-Package.

<PackageReference Include="Sdcb.SimdCodecs.Bmp" Version="1.2.0" />

For projects that support PackageReference, copy this XML node into the project file to reference the package.

<PackageVersion Include="Sdcb.SimdCodecs.Bmp" Version="1.2.0" />
                    

                            Directory.Packages.props

<PackageReference Include="Sdcb.SimdCodecs.Bmp" />
                    

                            Project file

For projects that support Central Package Management (CPM), copy this XML node into the solution Directory.Packages.props file to version the package.

paket add Sdcb.SimdCodecs.Bmp --version 1.2.0

The NuGet Team does not provide support for this client. Please contact its maintainers for support.

#r "nuget: Sdcb.SimdCodecs.Bmp, 1.2.0"

#r directive can be used in F# Interactive and Polyglot Notebooks. Copy this into the interactive tool or source code of the script to reference the package.

#:package Sdcb.SimdCodecs.Bmp@1.2.0

#:package directive can be used in C# file-based apps starting in .NET 10 preview 4. Copy this into a .cs file before any lines of code to reference the package.

#addin nuget:?package=Sdcb.SimdCodecs.Bmp&version=1.2.0
                    

                            Install as a Cake Addin

#tool nuget:?package=Sdcb.SimdCodecs.Bmp&version=1.2.0
                    

                            Install as a Cake Tool

The NuGet Team does not provide support for this client. Please contact its maintainers for support.

Sdcb.SimdCodecs.Bmp

Pure-managed BMP / DIB decoding and encoding for .NET 10, derived from Microsoft's documentation of the Windows DIB and IBM's of the OS/2 bitmap. Rows are converted with AVX2 / SSSE3 / AdvSimd kernels, each with a scalar twin that gives the same bytes, and decoded straight into the caller's buffer in any pixel layout without allocating.

using Sdcb.SimdCodecs.Abstractions;
using Sdcb.SimdCodecs.Bmp;

byte[] data = File.ReadAllBytes("image.bmp");
BmpInfo info = BmpDecoder.ReadInfo(data);       // size, depth, compression, header, alpha
BmpImage image = BmpDecoder.Decode(data);       // RGB, or RGBA when the bitmap can carry alpha

var bgra = new byte[info.Width * info.Height * 4];
BmpDecoder.DecodeTo(data, bgra, ImagePixelFormat.Bgra32);

What it reads:


Headers	`BITMAPCOREHEADER` (OS/2 1.x), OS/2 2.x `BITMAPINFOHEADER2` (16 to 64 bytes), `BITMAPINFOHEADER`, the 52- and 56-byte V2 / V3, `BITMAPV4HEADER`, `BITMAPV5HEADER`; the first bitmap of an OS/2 bitmap array (`BA`)
Depths	1, 2, 4 and 8-bit palettes (3- or 4-byte entries); 16, 24 and 32-bit
Compression	`BI_RGB`; `BI_BITFIELDS` and `BI_ALPHABITFIELDS` with any masks; `BI_RLE4` / `BI_RLE8` with deltas and early ends; OS/2 RLE24 and Huffman 1D; `BI_PNG`
Layout	Bottom-up and top-down rows
Metadata	A V5 header's embedded ICC profile (`BmpImage.IccProfile`)
Packed DIBs	`BmpDecoder.ReadDibInfo` / `DecodeDib` / `DecodeDibTo`: the same without the file header, as the clipboard (`CF_DIB`, `CF_DIBV5`) holds it; `iconImage: true` reads an icon or cursor image, its AND mask as alpha

BI_JPEG and 64-bit bitmaps are refused, as is anything malformed (InvalidDataException, or false from the Try methods and the BmpImageDecoder adapter). Images above 2^28 pixels are refused, to bound what a decode allocates.

Encoding

byte[] bmp = BmpEncoder.Encode(rgba, width, height, ImagePixelFormat.Rgba32);

byte[] rle = BmpEncoder.Encode(rgba, width, height, ImagePixelFormat.Rgba32,
    new BmpEncodeOptions { Format = BmpEncodeFormat.Indexed8, Rle = true, Dither = true });

Option
`Format`	`Indexed1` / `Indexed4` / `Indexed8` (exact when the colours fit, otherwise Wu-quantized, optionally dithered), `Rgb555`, `Rgb565`, `Rgb24`, `Rgb32`, `Argb32`. `Auto` (the default): 32-bit with alpha when the input's alpha is not all 255, an 8-bit grey palette for grey input, 24-bit otherwise
`Rle`	`BI_RLE4` / `BI_RLE8` for a 4- or 8-bit palette
`Header`	`Core`, `Info`, `V4` or `V5`. `Auto`: V5 with an ICC profile, V4 for alpha (`BI_BITFIELDS` with an alpha mask), `Info` otherwise
`TopDown`	A negative height, top row first
`IccProfile`	Embedded in a V5 header

Input is any ImagePixelFormat (B, G, R input written as 24- or 32-bit is copied as it is), or 1 / 3 / 4 channels. BmpEncoder.EncodeDib writes a packed DIB; with iconImage: true, an icon or cursor image: a 40-byte header with the doubled height, then the AND mask (alpha below 128 is transparent).

Decisions

Where the documentation leaves a choice to the decoder, it is made as WIC (Windows' own decoder) makes it, with the exceptions marked; the test suite holds every one of them against WIC, ImageMagick or a hand-built stream.

Case	Decoded as
A channel narrower than 8 bits	Its bits repeated to fill the byte (5-bit `v` is `v << 3 \\| v >> 2`; a 3-bit 7 is 255), as WIC, ImageMagick and ImageSharp do for 5- and 6-bit channels. WIC writes channels of 1 to 3 bits only twice (a 3-bit 7 is 0xFC); this decoder keeps white white
A channel wider than 8 bits (10-10-10-2, …)	Its top 8 bits, as WIC and ImageMagick
An index past the palette	Opaque black, as WIC
A palette longer than the indices reach, or one the pixel offset cuts short	The entries the indices reach that lie before the pixels
Pixels an RLE stream skips (a delta, an early end of line or bitmap, data that runs out)	Opaque black, as WIC and ImageSharp (browsers make them transparent)
An RLE run or literal past the right edge	Clipped
The fourth byte of a 32-bit `BI_RGB` pixel	Alpha, unless it is 0 in every pixel, then opaque: as browsers, ImageMagick and ImageSharp. WIC always ignores it
An alpha mask (`BI_BITFIELDS` in a V3 to V5 header, `BI_ALPHABITFIELDS`)	Alpha, in any layout and depth, as ImageMagick and ImageSharp. WIC reads it only as the top byte of a B, G, R, A 32-bit pixel
A file offset to the pixels that points into the headers, or past the end	The pixels follow the palette
The last row without its padding	Accepted
A V5 profile that does not fit the data	Ignored
Compression 3 on a 1-bit bitmap, 4 on a 24-bit one	OS/2 Huffman 1D and RLE24 (Windows' `BI_BITFIELDS` and `BI_JPEG` take other depths)
OS/2 Huffman 1D	ITU-T T.4 Modified Huffman runs, each row starting white; white is index 0. Rows are not padded; EOL codes before a row are skipped. No tool at hand reads this compression, so these choices are checked only against streams built from the T.4 tables
OS/2 RLE24	RLE8's codes, each run a B, G, R triple and each literal three bytes a pixel, padded to 16 bits. Unchecked against another decoder for the same reason
An icon image	32-bit: alpha from the pixels, unless all 0, then from the AND mask; other depths: the AND mask. Without an AND mask, opaque

Product	Compatible and additional computed target framework versions.
.NET	net10.0 is compatible. net10.0-android was computed. net10.0-browser was computed. net10.0-ios was computed. net10.0-maccatalyst was computed. net10.0-macos was computed. net10.0-tvos was computed. net10.0-windows was computed.

Compatible target framework(s)

Included target framework(s) (in package)

Learn more about Target Frameworks and .NET Standard.

net10.0
- Sdcb.SimdCodecs.Abstractions (>= 1.2.0)
- Sdcb.SimdCodecs.Png (>= 1.2.0)

NuGet packages (2)

Showing the top 2 NuGet packages that depend on Sdcb.SimdCodecs.Bmp:

Package	Downloads
Sdcb.SimdCodecs Image-codec facade over the Sdcb.SimdCodecs family: sniff a byte stream's format by its magic-byte signature and dispatch to the matching decoder (PNG, JPEG, TIFF, JBIG2, JPEG 2000, WebP, GIF, BMP, ICO, CUR, QOI, PBM / PGM / PPM / PAM, PFM), returning a codec-neutral IDecodedImage (fidelity tier: native bit depth + ColorEncoding colour signalling, with a ToFloats() RGBA float32 view) or decoding zero-copy into a caller-provided 8-bit RGBA buffer (display tier, integer-sample formats). AOT-compatible.	108
Sdcb.SimdCodecs.Ico Pure-managed Windows icon (ICO) and cursor (CUR) decoder and encoder for .NET. Reads every image of the directory, each a PNG stream or a DIB with its AND mask (1/4/8/16/24/32-bit, through Sdcb.SimdCodecs.Bmp), with cursor hotspots; picks the largest, deepest image by default and opens all of them as frames. Writes multi-size icons and cursors, each image PNG or 1/4/8/24/32-bit BMP. AOT-compatible.	0

Package

Downloads

Sdcb.SimdCodecs

Image-codec facade over the Sdcb.SimdCodecs family: sniff a byte stream's format by its magic-byte signature and dispatch to the matching decoder (PNG, JPEG, TIFF, JBIG2, JPEG 2000, WebP, GIF, BMP, ICO, CUR, QOI, PBM / PGM / PPM / PAM, PFM), returning a codec-neutral IDecodedImage (fidelity tier: native bit depth + ColorEncoding colour signalling, with a ToFloats() RGBA float32 view) or decoding zero-copy into a caller-provided 8-bit RGBA buffer (display tier, integer-sample formats). AOT-compatible.

108

Sdcb.SimdCodecs.Ico

Pure-managed Windows icon (ICO) and cursor (CUR) decoder and encoder for .NET. Reads every image of the directory, each a PNG stream or a DIB with its AND mask (1/4/8/16/24/32-bit, through Sdcb.SimdCodecs.Bmp), with cursor hotspots; picks the largest, deepest image by default and opens all of them as frames. Writes multi-size icons and cursors, each image PNG or 1/4/8/24/32-bit BMP. AOT-compatible.

GitHub repositories

This package is not used by any popular GitHub repositories.