NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
pub.dev, #117 most downloaded on pub.dev
Dart Image Library provides server and web apps the ability to load, manipulate, and save images with various image file formats.
Latest stable
4.13.1
Last release 8 days ago
14 Nov 2026
Release timing varies
gaps range from 8 days to 9 months
Nearly every release is documented
notes for 59 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
13 years old
130 releases · first in 2014
Drag across the ruler to compare, or tap a release to open it.
Refuse WebP and APNG frames outside the canvas, and WebP frames holding an animation.
maxPixels overflowing for huge sizes; bound WebP Huffman groups, BMP palettes and 4 component JPEGs.maxPixels.Apply maxPixels to WebP frames and TIFF JPEG tiles; bound inflate
Apply maxPixels to WebP frames and TIFF JPEG tiles; bound inflate
maxPixels to WebP animation frames and TIFF JPEG tiles; bound memory when inflating crafted zlib data.IccProfile.decompressed throws ImageException for invalid data.srgbToLinear and linearToSrgb.linearBlend to resizing, blurring, rotating, pixelate, vignette, distortions and drawing, to blend colors as linear light.linearBlend in drawPixel and compositeImage uses the sRGB curve instead of a 2.2 gamma.gaussianBlur and separableConvolution for float32 images.clear ignoring its color for float32 images.CopyRectifyCmd ignoring interpolation.contrast with ContrastMode.scurve never applying the curve.luminanceThreshold with outputColor squaring the blue channel.hexagonPixelate centerX and centerY are pixel offsets of the grid; fix a crash on 1 pixel wide images.copyImageChannels with a mask on non-float images.pixelate average mode reading pixels it already changed; edge blocks average only their own pixels.BinaryQuantizer dithering turning almost every color white.reinhardTonemap uses the standard Reinhard operator, keeping luminance in 0-1.Image.convert to 1 channel keeps fractional luminance for float images.Image.convert(withPalette: true) keeps alpha for RGBA images.fillRect blending of partly transparent colors and masks.drawRect with a radius ignoring the mask on its straight edges.drawString centering of multi-line and wrapped text.BitmapFont.fromFnt and readFont throwing on .fnt files with a page line.copyResizeCropSquare with a radius dropping animation frames.contrast mode, pixelate amount, quantize ditherStrength, trim fuzzy and padding, drawPolygon antialias and thickness, drawPixel filter and alpha, fillRect alphaBlend, decodeJpg scale and maxPixels.ditherImage command defaults to zigzag scan order, like the function.executeThread and the other thread methods rethrow anything thrown, not only errors and exceptions.IcoDecoder.decodeImageLargest ignoring 256 pixel icons.encodeNamedImage and encodeImageFile write CUR files for .cur names.IcoEncoder and CurEncoder throw ImageException for images larger than 256.PsdDecoder.startDecode returns null for data that isn't a PSD file.CurEncoder, Rational and WebPFrame.Your coding agent can read these notes before it upgrades. Set up the MCP server →
update archive dep to 4.5.0
update archive dep to 4.5.0
Security fixes for decoding untrusted images:
decodeImage.maxPixels to every decoder, defaulting to Decoder.defaultMaxPixels; it also limits animations.JpegDecoder.defaultMaxPixels for Decoder.defaultMaxPixels.isValidFile throwing on short data, which made decodeImage throw.ImageException for bad data, not RangeError or other errors.ImageException for empty images, and Image rejects negative sizes.executeCommandImageAsync and executeCommandBytesAsync never completing when a command throws.fillFlood overflowing the stack on large shapes, filling only part of a shape, and clearing alpha.drawString with wrap redrawing each line for every word.Fix dependency constraint to archive 4.4.0.
Fix encodeJpg modifying the pixels of RGBA source images.
encodeJpg modifying the pixels of RGBA source images.gaussianBlur and separableConvolution only blurring the first frame.gaussianBlur and separableConvolution crashing when the radius is
larger than the image.trim copying the first frame into every frame.GifEncoder.ditherStrength being ignored after the first frame.flip (both directions) and bakeOrientation (orientation 3) not
flipping the middle row of odd-height images.Image.convert(withPalette: true) modifying the source image.ditherImage crashing on palette images.OutputBuffer growth taking quadratic time; 8-bit BMP encoding is ~5x
faster.decodeImage.ditherImage are ~2x faster, and GIF decoding faster.gaussianBlur and separableConvolution are ~7x faster for 8-bit images,
without a full-size temporary image.flip, flipVertical, flipHorizontal, copyRotate (90, 180
and 270 degrees) and grayscale.copyResize is 4-10x faster, and resize up to ~13x.copyCrop, trim and direct compositeImage.gamma and adjustColor are up to ~50x faster.convolution (and filters using it), sobel and edgeGlow are much faster.copyResize applies the EXIF orientation while resizing, without a rotated
copy: ~4x faster and ~8x less memory for portrait photos.bakeOrientation uses less memory on rotated photos.JpegComponent.blocks is replaced by a flat coefficients list.scale to decodeJpg and JpegDecoder, to quickly decode a JPEG at
1/2, 1/4 or 1/8 size (~4x faster, with ~2% of the memory, at 1/8).PngCicpData.colourPrimaries is renamed to colorPrimaries.IfdValueUndefined.view, which doesn't copy its data.Fix a denial of service where a tiny JPEG declaring huge dimensions made the decoder allocate gigabytes of memory and spend minutes decoding. JPEG dec
ImageException for images with more than
JpegDecoder.defaultMaxPixels pixels (2^28 by default). The limit can be
set per call with decodeJpg(bytes, maxPixels: n) or
JpegDecoder(maxPixels: n), and a value <= 0 disables it.JpegDecoder.startDecode no longer allocates the image's pixel data.Fix alpha blending onto translucent pixels, which darkened translucent animated WebP frames (#804).
PixelFloat16.a setter writing the green value.Add lossy WebP encoding with encodeWebP(lossless: false) and the quality, method and alphaQuality options.
encodeWebP(lossless: false) and the quality,
method and alphaQuality options.cwebp.exact and singleFrame options to encodeWebP.WebPEncoder.exact now defaults to true, preserving the color under fully
transparent pixels at the cost of larger files.Add the DitherKernel.burkes and DitherKernel.jarvisJudiceNinke error-diffusion kernels.
DitherKernel.burkes and DitherKernel.jarvisJudiceNinke
error-diffusion kernels.ditherImageBayer and the
DitherKernel.bayer2x2, DitherKernel.bayer4x4, DitherKernel.bayer8x8
kernels. Unlike the error-diffusion kernels, the pattern is position-based
and uses a fixed threshold matrix, so it is deterministic and fast.DitherScanOrder (raster, serpentine, zigzag, hilbert) and the
ditherImage(scanOrder:) / quantize(ditherScanOrder:) /
encodeGif(ditherScanOrder:) / GifEncoder(ditherScanOrder:) parameters.
zigzag visits pixels along the anti-diagonals, which breaks up the
horizontal worm artifacts of raster scanning.findDecoderForData.injectJpgExif dropping any segment (such as the JFIF APP0 header) that
precedes the EXIF APP1 block, and losing the embedded EXIF thumbnail.OctreeQuantizer mapping colors to palette index 0 when their octree
path hit a folded-away branch, which turned bright pixels near-black once an
image had more distinct colors than the requested palette size. Palettes
built from partially-folded internal nodes now include those nodes' entries,
so the palette also no longer comes up short of numberOfColors.* Remove meta dependency.
Fix minMax returning incorrect values when the minimum or maximum was in the first pixel of a multi-channel image.
minMax returning incorrect values when the minimum or maximum was in
the first pixel of a multi-channel image.copyResizeCropSquare ignoring the crop offset when using non-nearest
interpolation, which off-centered the resized result.copyResizeCropSquare throwing a range error when a non-zero radius was
used with a non-square image.drawLine drawing diagonal lines offset by ~1 pixel.copyExpandCanvas.Image.convert now applies an explicitly provided alpha value even when
the format and channel count are unchanged.Image.fromBytes when the source row stride is smaller
than the image's row stride.RangeError when decoding images with corrupt EXIF data.BM signature instead of
crashing the decoder.noise with NoiseType.saltAndPepper producing colored pixels instead
of black/white ones.colorOffset now scales its offsets to the bit depth of the image, so the
effect is consistent regardless of the image's format.fuzzy and padding options to trim and findTrim.findEncoderForData, which returns an Encoder for a buffer of image
data, complementing findDecoderForData.dispose option to GifEncoder to control the frame disposal method.PsdImage.layers no longer throws a LateInitializationError when accessed
before decode has been called.Fix JPEG decoding issue with Adobe RGB color transformations.
Nothing published for this version
Fix issue decoding the last row of lossless webp images.
Major update to the WebP decoder to resolve errors decoding some files.
Fix issues with injectJpegExif corrupting jpeg files.
Fix for fillRect for images that that have alpha.
Improve fillPolygon to better handle concave polygons.
* Reduce Min SDK version to 3.0.
Make image library compatible with archive 3.x
Add ConstColorRgba8, ConstColorRgb8 for const color creation.
Upgrade to Archive 4.0 dependency
Fix exceptions loading some PSD files.
Fix decoding EXIF data from WebP.
JPEG images will finish decoding even if the file is incomplete or has errors.
Incomplete or JPEGs with errors will now finish loading.
Optimize copying bytes in Image.fromBytes
Fix Image.getBytes when ChannelOrder has a different number of channels than the image.
Fix crash in copyResize for non-nearest interpolation modes.
No changes, removing unnecessary files in 4.1.1.
Add maintainAspect and backgroundColor to copyResize to resize width and height of an image, without stretching the source (using background color to
* Update pub dependencies.
Fix resizing multi-frame palette images.
Fix GIF decoder not decoding some animation files transparency and frame duration.
Fix JPEG encoder for non uint8 format images.
Use Image.backgroundColor for copyRotate, copyCropCircle, and other functions that reveal background pixels.
Fix transform functions for palette images.
Fix EXIF parsing little endian data
Add decodeJpgExif and injectJpgExif functions to process JPG exif data without needing to decode the image.
Fix last-minute typo from previous release.
Fix offset error with BMP encoder for images with < 8 bits per pixel
Add Image.hasAlpha getter property that will be true if the Image has an alpha channel.
Fix Image.getRangeIterator skipping last line.
Image.remapChannels(ChannelOrder order) to remap the order of channels from rgb to bgr, etc.
Image.remapChannels(ChannelOrder order) to remap the order of channels from rgb to bgr, etc.Image.getBytes({ChannelOrder? order}) will return the image data, optionally reordering the channels.ChannelOrder? order named arg to specify the channel order of the input data.Improved antialiasing for drawLine
Fix for animated PNG with palette images, which needed a global palette for all frames.
Fix for encoding GIF transparency.
No changes, dart formatted code and fix documentation link.
Use strict analyzer settings, clean up warnings.
Major update of the Dart Image Library. Includes support for:
Fix character code issue with BitmapFont.
Fix transparency issue with animated GIF images.
Fixes for APNG: fix exception from some APNG files, and some frames were not composited correctly.
Update SDK dependency to >2.15.0 and XML package dependency to >6.0.0
Optimize Image.getWhiteBalance function, add asDouble argument to return double value.
Add BmpEncoder to encode BMP images, along with encodeBmp function. Currently, only 24-bit or 32-bit BMP images will be encoded.
Fix error loading some tiff images
Update archive version requirement
* Fix WebP lossless decoder.
* Change LICENSE to MIT.
One column per quarter.