imageformats

.NET modules for decoding obscure graphics formats:

• .PPM, .PGM, .PBM (Netpbm images)
• .TGA (Targa)
• .PCX (ZSoft Paintbrush)
• .IFF (Amiga ILBM images)
• .SGI, .RGB, .BW (SGI images)
• .RAS (Sun raster images)
• .MAC (MacPaint)
• .CUT (Dr. Halo)
• .XPM (X Window PixMap)
• .DEEP (TVPaint IFF DEEP images)
• .FITS, .FIT, .FTS, .FZ (astronomical images, including image cubes, multi-extension and tile-compressed files)
• .DCM (DICOM medical images, including JPEG, JPEG-LS, JPEG 2000 and RLE compressed ones)
• .JP2, .J2K, .J2C, .JPF (JPEG 2000)
• .ART (AOL Johnson-Grace images)
• .PSP, .PSPIMAGE (Paint Shop Pro images)
• .AWD (Microsoft Fax documents)
• .PSD, .PSB, .PDD (Adobe Photoshop and PhotoDeluxe images)
• .FLI, .FLC (Autodesk Animator animations)
• .FPX (FlashPix images)
• .PPP (Serif PagePlus publications; reads the preview image stored in the file)
• .TIF, .TIFF (TIFF images, including BigTIFF, multi-page, tiled, CCITT, LZW, Deflate and JPEG compressed ones)
• .SER (Lucam Recorder videos, as used for planetary imaging, including raw Bayer ones)
• .CIN, .CINE (Phantom high-speed camera videos, including raw color and packed 10-bit ones)
• .DDS (DirectDraw Surface textures, including BC1 to BC7 compressed ones, cube maps, texture arrays and volume textures)
• .LFP, .LFR (Lytro light field camera pictures, both raw ones and refocus stacks)
• .MNG, .JNG (Multiple-image Network Graphics animations, including ones with JNG and Delta-PNG images, and JPEG Network Graphics images)
• .CR2, .CR3, .CRW, .NEF, .NRW, .ARW, .SR2, .SRF, .PEF, .ORF, .RW2, .RWL, .RAF, .DNG, .SRW, .KDC, .MRW, .MOS, .X3F, and other camera raw images (reads the preview image embedded in the file)

Usage

Every decoder returns an ImageData object: a plain buffer of raw pixels, with no dependency on any particular imaging library. Hand the buffer to whichever library you prefer.

using DmitryBrant.ImageFormats;

// Detects the format automatically, and returns null if it isn't one we know.
ImageData image = ImageData.Load("picture.tga");

The pixels in image.Data are stored top-down, one row after another, four bytes per pixel in blue, green, red, alpha order. (Read as a 32-bit little-endian integer, that's the usual packed ARGB layout.) This is the same layout as GDI+ Format32bppArgb, WIC/Direct2D B8G8R8A8, Avalonia's Bgra8888, and ImageSharp's Bgra32, so it can go straight into any of them:

// Avalonia
var bitmap = new WriteableBitmap(new PixelSize(image.Width, image.Height),
    new Vector(96, 96), PixelFormat.Bgra8888, AlphaFormat.Unpremul);
using (var buffer = bitmap.Lock())
    for (int y = 0; y < image.Height; y++)
        Marshal.Copy(image.Data, y * image.Stride, buffer.Address + (y * buffer.RowBytes), image.Stride);

// ImageSharp
var bitmap = Image.LoadPixelData<Bgra32>(image.Data, image.Width, image.Height);

// System.Drawing
var bitmap = new Bitmap(image.Width, image.Height, PixelFormat.Format32bppArgb);
var bits = bitmap.LockBits(new Rectangle(Point.Empty, bitmap.Size), ImageLockMode.WriteOnly, bitmap.PixelFormat);
for (int y = 0; y < image.Height; y++)
    Marshal.Copy(image.Data, y * image.Stride, bits.Scan0 + (y * bits.Stride), image.Stride);
bitmap.UnlockBits(bits);

DICOM, FITS, FLI/FLC, TIFF, SER, Cine, DDS, Lytro and MNG files can contain more than one frame (for example, an ultrasound cine loop, the planes of an astronomical image cube, an animation, the pages of a fax, a planetary or high-speed video, the faces of a cube map, or the focal planes of a refocusable picture). ImageData.Load returns the first frame (or, for a refocusable Lytro picture, the one focused at its default depth); to get at the others, use DicomReader, FitsReader, FliReader, TiffReader, SerReader, CineReader, DdsReader, LfpReader or MngReader directly:

int count = DicomReader.GetFrameCount("series.dcm");
ImageData frame = DicomReader.Load("series.dcm", count - 1);

// Or decode them all, one at a time, parsing the file only once:
foreach (ImageData frame in DicomReader.LoadFrames("series.dcm"))
    ...

Photoshop files normally contain a flattened copy of the picture, which is what ImageData.Load returns. When that copy is missing or only a placeholder (as in PhotoDeluxe files, or Photoshop files saved without "Maximize Compatibility"), the layers are composited instead. To make that choice yourself, use PsdReader directly:

ImageData image = PsdReader.Load("picture.psd", PsdImageSource.Layers);

For camera raw files, ImageData.Load returns the largest of the JPEG previews that the camera embedded in the file (often full size), rotated the way the picture was taken. A raw file that's also a TIFF file, but has no JPEG preview, is read as a TIFF file instead, which usually gives a small uncompressed thumbnail. RawReader.Load reads the preview of any raw file directly.

Occasionally a source format stores its pixel data as an embedded image in some other well-known encoding that the library can't decode itself (for example, a DICOM file that wraps an 8-bit JPEG using arithmetic coding). Those bytes are passed through untouched: image.IsEncoded is true, image.Data is null, and the encoded bytes are in image.EncodedData (with image.EncodedFormat naming the encoding) for you to decode yourself.

The ImageViewer project in this repository is a small Avalonia app that does all of the above; see its Interop.cs for a complete example.

Copyright 2013+ Dmitry Brant

https://dmitrybrant.com

License: MIT