Image version 0.67.0

Enhancements

• Adds Image.vignette/2.

• Adds Image.Palette.extract/2 — perceptual palette extraction from an image via K-means in Oklab plus the Color.Palette.Cluster merge / phantom-guard / centroid-aware-rep pipeline (requires :scholar and :nx).

• Adds Image.gamma/2 — per-band gamma curve adjustment (wraps Vix.Vips.Operation.gamma/2).

• Adds Image.sepia/2 — single-pass sepia tone via a 3×3 colour-recombination matrix; accepts a 0.0..1.0 strength that blends the matrix with the identity (matching imgix's sepia=N percentage).

• Adds Image.posterize/2 — quantises each band to 2..256 evenly-spaced levels for a flat-shaded, comic-strip rendering.

• Adds Image.opacity/2 — multiplies the alpha band by a 0.0..1.0 factor; adds an opaque alpha band first when the input has none.

• Adds Image.set_orientation/2 — overrides the EXIF orientation tag without rotating the underlying pixels (paired with Image.open/2's default autorotate: false for full caller control over orientation; matches imgix's or=N).

• Adds :lossy and :chroma_subsampling options to Image.write/3. :lossy (boolean) toggles the lossless wire format on WebP / AVIF and palette-quantisation on PNG. :chroma_subsampling selects :auto / :on (4:2:0) / :off (4:4:4) on JPEG and AVIF.

• Adds Image.tint/2 — colour-tinted monochrome via a single 3×3 luminance + tint colour-recombination matrix. Used for the imgix monochrome=#hex / ImageKit e-monochrome family of CDN options.

• Adds Image.fade/2 — alpha-gradient fade-out on one or more edges via SVG <linearGradient> masks combined with a per-pixel min. Supports :top, :bottom, :left, :right, or :all, with :length as either pixels or a fraction of the relevant dimension. Used for Cloudinary e_fade.

• Adds Image.drop_shadow/2 — soft drop shadow under the image's alpha-shaped silhouette. Composites a Gaussian-blurred, opacity-scaled, tinted copy of the alpha band beneath the original. Used for ImageKit e-shadow and Cloudinary e_shadow.

• Adds Image.minimize_metadata/2 with a :keep option — caller-controlled list of EXIF fields to preserve when minimising metadata. The 1-arity variant continues to default to [:copyright, :artist]; the 2-arity variant lets callers pass keep: [:copyright] (preserve only copyright), keep: [] (strip everything), or any other subset.

• Adds Image.enhance/2 — content-aware automatic enhancement composed of luminance equalisation + mild saturation boost + mild sharpen. Approximates the CDN-style "improve" / "auto-enhance" calls used by Cloudinary, imgix, and ImageKit. Tunable via :saturation and :sharpen_sigma options.

• Adds Image.to_colorspace/3 — ICC-profile-driven colourspace conversion. Accepts the libvips built-in profile atoms (:srgb, :cmyk, :p3) or a path to an .icc file via Image.ICCProfile.known?/1 validation. Options: :input_profile, :intent (:relative / :perceptual / :saturation / :absolute), and :depth (8 / 16). Wraps Vix.Vips.Operation.icc_transform/3.

Removed

• Image.QRcode is removed. QR encoding and decoding move to the sibling image_qrcode package, which is built on Nayuki's QR-Code-generator + quirc and does not depend on :evision. Migration: replace Image.QRcode.encode/2 / Image.QRcode.decode/1 with Image.QRCode.encode/2 / Image.QRCode.decode/1 (note the capital "C") and add {:image_qrcode, "~> 0.1"} to your deps. The Image.to_evision/2 and Image.from_evision/1 interop helpers are unchanged.

Bug Fixes

• Image.add_alpha/2's :opaque and :transparent atoms now produce alpha = 255 and alpha = 0 respectively, matching the standard libvips / RGBA convention. The previous values were inverted relative to their names; integer values pass through unchanged.

Image version 0.66.0

Enhancements

• Relax Nx support to allow ~> 0.10 (not ~> 0.11). This allows Bumblebee to be configured in image_vision since Bumblebee only supports Nx ~> 0.9.0 or ~> 0.10.0.

Image version 0.65.0

Bug Fixes

• Use the updated colorspace names in Color 0.4.0.

Image version 0.64.0

The primary intent of this release is to stablise the code in readiness for a 1.0 release. Deprecated code has been removed, a standardised color model introduced (via the new library color) and objective classification and detection has been removed to a new image_detection library.

Breaking Changes

• Five long-deprecated functions have been removed:

  • Image.interpretation/1 — use Image.colorspace/1.
  • Image.type/1 (was format/1) — use Image.band_format/1.
  • Image.convert_to_mask/1 — use Image.convert_alpha_to_mask/1.
  • Image.convert_to_mask!/1 — use Image.convert_alpha_to_mask!/1.
  • Image.map_pages/2 — use Image.map_join_pages/2.

• Image.Color has been removed. Color handling now lives in two new modules and one new dependency:

• Image.Classification and Image.Generation have moved to a new sibling package, :image_detection.

• :bumblebee is no longer a dependency of :image. It is configured in the new library image_detection.

• Image.Video is now backed by Xav (a thin Elixir wrapper around FFmpeg) instead of :evision / OpenCV. The public API surface is largely unchanged but the underlying type, options, and a few semantic details have moved:

  • The video struct is now %Image.Video{} (with fields :reader, :source, :fps, :duration_seconds, :frame_count, :width, :height) rather than %Evision.VideoCapture{}. Pattern-match on the new struct module if your code does so.

  • Image.Video.open/2's :backend option has been removed. FFmpeg picks the demuxer automatically and there is no concept of pluggable backends in Xav.

  • Image.Video.known_backends/0, available_backends/0, known_backend?/1, known_backend_values/0, and available_backend?/1 have been removed for the same reason. Image.Options.Video (the module that owned the backend table) has been deleted.

  • Camera input is now opened via a platform-specific device path. :default_camera resolves to /dev/video0 on Linux, "0" (AVFoundation device 0) on macOS, and "video=0" on Windows. An integer camera index is mapped to the corresponding /dev/videoN (or platform equivalent). For non-default cameras you can also pass an explicit FFmpeg device string.

  • Frame-based seeking (Image.Video.seek/2 with frame: n, and Image.Video.image_from_video/2 with frame: n) is now implemented as a time-based seek to n / fps followed by zero or more next_frame calls. For most files this lands on the requested frame; for very inter-frame-compressed files FFmpeg may snap to the nearest preceding keyframe.

  • Image.Video.close/1 is now a no-op that returns {:ok, %Image.Video{reader: nil}}. Xav garbage-collects the underlying FFmpeg context, so explicit close is no longer necessary. The function is retained for source compatibility; subsequent operations on the closed struct return {:error, %Image.Error{reason: :video_closed}}.

  • Image and audio frames are decoded by FFmpeg + libswscale rather than by OpenCV's videoio backend. Pixel-exact comparisons against fixtures generated by the previous version will not match; the test fixture test/support/validate/video/video_sample_frame_0.png has been regenerated.

• :xav is now an optional dependency. It requires FFmpeg ≥ 6.0 to be installed on the system. Add it to your mix.exs if you use Image.Video:

  {:xav, "~> 0.10", optional: true}
  

• :evision is no longer needed for Image.Video. It is still required for Image.QRcode and for the Image.to_evision/2 / Image.from_evision/1 interop helpers, which are unchanged. The README's optional-dependency table reflects the new split.

• Image.Error is now a structured public exception. It carries :reason (atom or {atom, value} tuple), :operation, :path, :value, and a derived :message. Every fallible function in the library now returns {:ok, value} or {:error, %Image.Error{}} — bare-string error tuples have been eliminated. Bang variants raise the same struct. The new Image.Error.wrap/2 helper attaches structured context to a raw libvips or File.* error. Pattern-match on :reason instead of scraping :message:

  case Image.open(path) do
    {:ok, image} -> ...
    {:error, %Image.Error{reason: :enoent}} -> not_found_handler()
    {:error, %Image.Error{} = err} -> raise err
  end
  

Enhancements

• Image.dominant_color/2 now accepts a :method option of either :histogram (the existing default) or :imagequant. The :imagequant method routes through libimagequant (via vips_gifsave_buffer) and returns a palette of {r, g, b} tuples ordered by perceptual importance. New :effort and :dither options tune the quantiser. See guides/performance.md for a comparison of the two methods.

• Colour arguments to drawing, embedding, trimming, gradient, chroma key, comparison, warp-perspective, meme, replace-colour, flatten, and if_then_else operations are now correctly converted to the target image's colour space. Drawing :red on a Lab image now produces actual Lab red, not the bytes [255, 0, 0] reinterpreted as Lab.

• New Image.Pixel and Image.ICCProfile modules. Image.Pixel.to_pixel/3 is the canonical way to turn any user-friendly colour input into a libvips-ready pixel for a particular image; Image.Pixel.to_srgb/1 is the image-independent equivalent for callers (SVG renderers, gradients) that need a fixed sRGB output.

• The :color library is now a dependency of :image. It is now the canonical colour science layer for the project.

• Image.Video now supports HTTP/HTTPS/RTSP/RTMP URLs as video sources for free, since FFmpeg supports them natively.

• Image.xav_configured?/0 is the new compile-time predicate that gates the Image.Video module (analogous to Image.evision_configured?/0 and Image.bumblebee_configured?/0).

• The Image.Video.frame_to_image/1 helper exposes the raw Xav.Frame → Vix.Vips.Image.t() conversion (used internally by image_from_video/2 and stream!/2). Useful if you have a frame from elsewhere in the Xav ecosystem and want to bring it into Image.

Image version 0.63.0

Breaking Changes

• Image is supported on Elixir 1.16 and greater only. Previously support was claimed for Elixir 1.12 and later.

Bug Fixes

• Fix compiler warnings on Elixir 1.20.

• Fix RGB to hex conversion. Thanks to @akoutmos for the issue and PR. Closes #196.

Image version 0.62.1

Bug Fixes

• Fix decoding EXIF component configuration if the value is invalid. Fixes #194. Thanks to @ethangunderson for the report.

• Fix Image.minimize_metadata/1 when the image does not have an author or copyright field.

Image version 0.62.0

Breaking Change

• Image.histogram/1 previously normalized the results which meant that the pixel counts for each bucket could not be resolved. The function no longer normalises the histogram so the raw pixel counts per bucket are retained.

Bug Fixes

• Image.histogram/1 is fixed to not normalize histogram entries. Therefore the histogram now returns correct pixel counts.

• Fix Image.Text.add_background_padding/2 when the padding is derived from a base image.

Enhancements

• Add Image.to_list/1 to return an image as a nested list.

• Added an example for Image.histogram/1 to illustrate what data is returned and how to interpret it.

Image version 0.61.1

Bug Fixes

• Fix "function get_req_message/1" compiler warning.

Image version 0.61.0

Enhancements

• Adds Image.from_req_stream/2. This function returns a Vix.Vips.Image.t/0 from streaming a Req request using the Req.get/2 option into: :self,

Image version 0.60.0

Breaking Changes

• Image.Math.maxpos/2 is renamed to Image.Math.top_n/2 to better reflect its intent. The keyword options argument is also replaced with a simple integer argument n.

• Image.Math.minpos/2 is renamed to Image.Math.bottom_n/2 to better reflect its intent. The keyword options argument is also replaced with a simple integer argument n.

• The return value from Image.Math.top_n/2 and Image.Math.bottom_n/2 have changed. They will now return the form {max, max_x, max_y, [{x_max_1, y_max_1}, {x_max_2, y_max_2}, ...]}.

Bug Fixes

• Improve the error messages for Image.crop/5 when the crop bounding box is invalid. Closes #190.

Enhancements

• Add Image.band_and/1, Image.band_or/1 and Image.band_xor/1 to apply the appropriate boolean operation across the bands of an image.

• Add new (and different) Image.Math.maxpos/2 and Image.Math.minpos/2 which return only coordinates whose values match the image maximum or minimum.