I’ve published 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. Video frame extractions is now using Xav rather than eVision.
As a result, there are quite a few breaking changes - and some new goodies. The breaking changes aren’t dramatic - but they are breaking.
Hopefully this is the last pre-release before 1.0. Feedback on this release will determine that outcome!
Breaking Changes
-
Five long-deprecated functions have been removed:
Image.interpretation/1— useImage.colorspace/1.Image.type/1(wasformat/1) — useImage.band_format/1.Image.convert_to_mask/1— useImage.convert_alpha_to_mask/1.Image.convert_to_mask!/1— useImage.convert_alpha_to_mask!/1.Image.map_pages/2— useImage.map_join_pages/2.
-
Image.Colorhas been removed. Color handling now lives in two new modules and one new dependency: -
Image.ClassificationandImage.Generationhave moved to a new sibling package,:image_detection. -
:bumblebeeis no longer a dependency of:image. It is configured in the new libraryimage_detection. -
Image.Videois 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:backendoption 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, andavailable_backend?/1have 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_cameraresolves to/dev/video0on 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/2withframe: n, andImage.Video.image_from_video/2withframe: n) is now implemented as a time-based seek ton / fpsfollowed by zero or morenext_framecalls. 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/1is 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.pnghas been regenerated.
-
-
:xavis now an optional dependency. It requires FFmpeg ≥ 6.0 to be installed on the system. Add it to yourmix.exsif you useImage.Video:{:xav, "~> 0.10", optional: true} -
:evisionis no longer needed forImage.Video. It is still required forImage.QRcodeand for theImage.to_evision/2/Image.from_evision/1interop helpers, which are unchanged. The README’s optional-dependency table reflects the new split. -
Image.Erroris 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 newImage.Error.wrap/2helper attaches structured context to a raw libvips orFile.*error. Pattern-match on:reasoninstead 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/2now accepts a:methodoption of either:histogram(the existing default) or:imagequant. The:imagequantmethod routes throughlibimagequant(viavips_gifsave_buffer) and returns a palette of{r, g, b}tuples ordered by perceptual importance. New:effortand:ditheroptions tune the quantiser. Seeguides/performance.mdfor 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_elseoperations are now correctly converted to the target image’s colour space. Drawing:redon a Lab image now produces actual Lab red, not the bytes[255, 0, 0]reinterpreted as Lab. -
New
Image.PixelandImage.ICCProfilemodules.Image.Pixel.to_pixel/3is the canonical way to turn any user-friendly colour input into a libvips-ready pixel for a particular image;Image.Pixel.to_srgb/1is the image-independent equivalent for callers (SVG renderers, gradients) that need a fixed sRGB output. -
The
:colorlibrary is now a dependency of:image. It is now the canonical colour science layer for the project. -
Image.Videonow supports HTTP/HTTPS/RTSP/RTMP URLs as video sources for free, since FFmpeg supports them natively. -
Image.xav_configured?/0is the new compile-time predicate that gates theImage.Videomodule (analogous toImage.evision_configured?/0andImage.bumblebee_configured?/0). -
The
Image.Video.frame_to_image/1helper exposes the rawXav.Frame→Vix.Vips.Image.t()conversion (used internally byimage_from_video/2andstream!/2). Useful if you have a frame from elsewhere in the Xav ecosystem and want to bring it intoImage.
