This is a prerelease. It is not intended for production use.
Please report any issues you find.

Example Hosted App

https://rerun.io/viewer/commit/342b6d5

Wheels can be installed with:

pip install --pre --no-index -f https://build.rerun.io/commit/342b6d5/wheels --upgrade rerun-sdk

or

pip install --pre --no-index -f https://github.com/rerun-io/rerun/releases/download/prerelease --upgrade rerun-sdk

CMake fetch-content for C++ SDK

include(FetchContent)
FetchContent_Declare(rerun_sdk URL https://build.rerun.io/commit/342b6d5/rerun_cpp_sdk.zip)
FetchContent_MakeAvailable(rerun_sdk)

or

include(FetchContent)
FetchContent_Declare(rerun_sdk URL https://github.com/rerun-io/rerun/releases/download/prerelease/rerun_cpp_sdk.zip)
FetchContent_MakeAvailable(rerun_sdk)

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.20.3/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.20.3/rerun_cpp_sdk.zip

We introduced a Map view in 0.20, but did not enable its feature flag by default, which meant that all web viewer builds did not include it. This is now fixed, and the map view may be used in web viewer builds published by Rerun.

🔎 Details

🪳 Bug fixes

• Fix web viewer feature flags #8295

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.20.2/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.20.2/rerun_cpp_sdk.zip

🔎 Details

🪳 Bug fixes

• Fix a drag-and-drop display regression #8228

📚 Docs

• Add map_view to the default features and improve how the nasm feature is handled and documented #8243

🧑‍💻 Dev-experience

• Gracefully handle cargo-metadata failures in users' environments #8239

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.20.1/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.20.1/rerun_cpp_sdk.zip

• Fix Rust docs #8168

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.20.0/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.20.0/rerun_cpp_sdk.zip

📖 Release blogpost: https://rerun.io/blog/maps
🧳 Migration guide: http://rerun.io/docs/reference/migration/migration-0-20

✨ Overview & highlights

• 🗺️ There is now an map view!
• 🎬 Native viewer now supports H.264 video if ffmpeg is installed.
• 📽️ Videos now load a lot faster use less RAM.
• 📂 Improvements to the existing Open (Viewer) & log_file (SDK) workflows, and addition of a new Import workflow.
  • Blueprints can now easily be re-used across different applications, recordings and SDKs
  • The new Import feature allows you to drag-and-drop any data into an existing recording, directly in the viewer.
• ☰ Dataframe queries are now streamed, reducing memory usage.
• 💊 Add capsule archetype.
• 📚 Doc improvements
  • Arrow schemas are now documented for all types.
  • Better structure to the how to section and a few more pages

⚠️ Breaking changes & deprecations

• 🐍 Python 3.8 is being deprecated
• 🔌 connect & serve got deprecated in favor of connect_tcp & serve_web
• 🎨 In Python, lists of numbers without type information are now assumed to be packed integer color representations, unless the length is exactly 3 or 4
  🧳 Migration guide: http://rerun.io/docs/reference/migration/migration-0-20

🔎 Details

🎬 Video

• Support H.264 video on native via user installed ffmpeg executable #7962
• Make mp4 parsing a lot faster & tremendously lower memory overhead #7860
• Fix playback of HDR AV1 videos in native viewer #7978
• Show all samples/frames in a video in a nice table #8102
• Calculate and show video frame number #8112
• Expose basic information about group of pictures in video data in the selection panel #8043
• Fix some videos having offsetted (incorrect) timestamps #8029
• Fix video backward seeking / stepping back sometimes getting stuck (in the presence of b-frames) #8053
• Make sure videos all end up in different space views #8085
• Fix video on web sometimes not showing last few frames for some videos #8117
• Fix issues with seeking in some H.264 videos on native & web #8111
• Fix view creation heuristics for videos #7869
• Improve video doc page #8007
• Update re_mp4 to fix integer overflow bug #8096

🪵 Log API

• Add Capsules3D archetype #7574 (thanks @kpreid!)
• rr.log_file_from_path now defaults to the active app/recording ID #7864
• Allow overriding albedo color on Asset3D #7458 (thanks @EtaLoop!)
• rr.serve -> rr.serve_web, rr.connect -> rr.connect_tcp #7906

🌊 C++ API

• C++: Improve error message when finding X11 macro Unsorted #7855
• Forward CMAKE_TOOLCHAIN_FILE to arrow build for sdk cross-compilation #7866 (thanks @SunDoge!)
• Update the python package to support python 3.13, update C++ arrow to 18.0.0 #7930

🐍 Python API

• Allow passing seconds/nanoseconds to VideoFrameReference archetype #7833
• Officially deprecate support for python 3.8 #7933
• Update the python package to support python 3.13, update C++ arrow to 18.0.0 #7930
• Remove the upper bound constraint on python version #7949
• Enable dataframe streaming across Python FFI #7935
• Fix python SDK's shutdown unsafely dropping cross-FFI resources #8038
• Improve edge-cases and warn on ambiguity for Rgba32 datatype #8054
• Check rerun notebook version on first import #8030

🦀 Rust API

• Allow logging individual components directly (Impl AsComponents for all ObjectKind::Component) #7756 (thanks @oxkitsune!)
• re_query::Caches -> re_query::QueryCache #7915

🪳 Bug fixes

• [bugfix] Make sure blueprint gets sent to the notebook view being created #7811
• Fix too short picking ray in pinhole-only scenarios #7899
• Update zune-jpeg to fix crash on bad JPEGs #7952
• Consistent open/import/log_file behaviors in all common scenarios #7966
• ChunkStore: fix row-id computation when removing dangling static chunks #8020
• EntityTree: only check for entity deletions when necessary #8103
• WebSocket server now indefinitely keeps track of non-data RPC commands #8146

🌁 Viewer improvements

• A Rerun Viewer session now matches 1:1 to a Rerun TCP server #6951 (thanks @petertheprocess!)
• Implement support for in-place drag-n-drop #7880
• Implement Menu > Import and associated command #7882
• Expose additional information about decoded frames in the viewer #7932
• Update crates, including rfd for better file dialogs #7953
• Line strips are no longer a disconnected series of quads #8065
• Show data density graph in collapsed time panel #8137
• Show the root entity "/" in the streams panel #8142

🚀 Performance improvements

• Don't keep around additional CPU copy of loaded mesh files #7824
• Make mp4 parsing a lot faster & tremendously lower memory overhead #7860
• Fix slow receive when using native WebSocket #7875
• Implement support for fully asynchronous QueryHandles #7964

🧑‍🏫 Examples

• Fix Rust DNA sample writing to a temporary file #7827
• Add ml_depth_pro example #7832 (thanks @oxkitsune!)
• Add map view to nuscenes python example #8034 (thanks @tfoldi!)
• Add an example to display OpenStreetMap-sourced data on the map view #8044
• Improve NuScenes example with more geo data & blueprint #8130

📚 Docs

• Clarify viewport documentation and reference the type list for view classes #7826
• Finish dataframe reference page #7865
• Docs: static data #7856
• Docs: concepts > recordings #7896
• Docs: "How-to: reuse blueprints across languages" #7886
• Docs: application model part 1: native workflows #7905
• Document arrow datatypes #7986

🖼 UI improvements

• Map View and GeoPoints archetype #6561 (thanks @tfoldi!)
• Replace the "Options" submenu with a settings screen #8001
• Improve error message style slightly #8092
• Much nicer looking error and warning messages #8127

🧑‍💻 Dev-experience

• Show list of enabled features with rerun --version [#7885](https://github.com/rerun-io/rerun/pul...

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.19.1/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.19.1/rerun_cpp_sdk.zip

This release fixes an error thrown when the web viewer is closed.

🔎 Details

🕸️ Web

• Fix wasm-bindgen patch #7970

📦 Dependencies

• Add wasm-bindgen version check to CI #7983

Rerun is an easy-to-use visualization toolbox and data platform for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.19.0/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.19.0/rerun_cpp_sdk.zip

📖 Release blogpost: https://rerun.io/blog/dataframe

🧳 Migration guide: http://rerun.io/docs/reference/migration/migration-0-19

✨ Overview & highlights

This release introduces two powerful features: a dataframe API (and view), as well as video support.

☰ Dataframe API

We now have an API for querying the contents of an .rrd file. This integrates with popular packages such as Pandas, Polars, and DuckDB.

You can read more in the Dataframe API how-to guide.

We have also added a matching dataframe view inside the Rerun Viewer.
Read more here.

🎬 Video

Rerun now supports logging MP4 videos using the new AssetVideo archetype.
This can greatly reduce bandwidth and storage requirements.

While the web viewer supports a variety of codecs, the native viewer supports only the AV1 codec for the moment, but we plan to support H.264 in the near future as well.
Read more about our video supports (and its limits) in our video docs.

⚠️ Breaking changes

• 🗾 Blueprint files (.rbl) from previous Rerun versions will no longer load automatically
• 🐧 Linux: Rerun now requires glibc 2.17+
• 🦀 Rust: The minimum supported Rust version is now 1.79

🧳 Migration guide: http://rerun.io/docs/reference/migration/migration-0-19

🔎 Details

📑 Raw changelog: 0.18.2...0.19.0

🪵 Log API

• BGR(A) image format support #7238
• Tensor & depth image value ranges can now be configured, from UI & code #7549
• New planar pixel formats: Y_U_V24/Y_U_V16/Y_U_V12 - _LimitedRange/FullRange #7666
• Add ShowLabels component, which controls whether instances’ labels are shown #7249 (thanks @kpreid!)
• Refactor MediaType guessing #7326

🌊 C++ API

• Add nullptr check when forwarding from component to datatype #7430

🐍 Python API

• Add missing show_labels and draw_order arguments in Python API #7363 (thanks @kpreid!)
• Allow logging to a recording without first calling rr.init() #7698
• Add support for NumPy arrays to the arrow serializer for string datatypes #7689

🦀 Rust API

• Update MSRV to Rust 1.79 #7563
• Update ndarray to 0.16 and ndarray-rand to 0.15 #7358 (thanks @benliepert!)
• Replace host_web_viewer method with WebViewerConfig::host_web_viewer #7553
• Fix Rust's TimeColumn::new_seconds/new_nanos creating sequence timelines #7402

🪳 Bug fixes

• Purge the query cache to prevent GC livelocks #7370
• Bug fix: always show latest data in follow-mode #7425
• Fix encoded image being suggested for non-image blobs (like video) #7428
• Chunk store: support for overlapped range queries #7586
• Fix image & video cache creating new entries when selecting data without explicit media type #7590

🌁 Viewer improvements

• The viewer will tail an .rrd that's is being written to #7475
• Native video support for AV1 #7557
• Allow splitting entity path expressions with whitespace #7782

🚀 Performance improvements

• Improve performance for scenes with many entities & transforms #7456
• Caches per recording #7513
• Automatic removal of unreachable static chunks #7518
• Invalidate hub-wide caches on deletions and overwrites #7525
• Do not cache static entries in the query-time latest-at cache #7654
• Make sure Arrow filter and take kernels early out where it makes sense #7704

🧑‍🏫 Examples

• Add drone LiDAR example #7336
• Add instant_splat example #7751 (thanks @pablovela5620!)

📚 Docs

• Add video reference docs #7533
• Document that Rerun does not support left-handed coords #7690
• Add a How-to guide for the dataframe API #7727
• Docs: move "roadmap" down to "development" #7775
• Add a "Getting started" guide for the dataframe API #7643
• Docs: clean up reference menu #7776
• Updating "Navigating the viewer" #7757

🖼 UI improvements

• Add a hook for views to add additional UI in the tab title bar #7438
• Text fields in the selection panel now span the available width #7487
• Do not deselect on ESC when it was used to close some other UI element #7548
• Add UI for precisely picking an exact sequence time #7673
• Remove the feature flag for plot query clamping #7664

🎨 Renderer improvements

• Introduce image data conversion pipeline, taking over existing YUV conversions #7640

🧑‍💻 Dev-experience

• Add a command palette action to reset egui's memory (debug build only) #7446
• Add NOLINT block to lint.py #7720

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.18.2/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.18.2/rerun_cpp_sdk.zip

Rerun 0.18.1 was (soft) yanked as we realized a bit too late that it was incomplete: we're jumping straight from 0.18.0 to 0.18.2.
0.18.2 is simply 0.18.1 with #7308 on top.

This patch release comes with the following fixes:

🌊 C++ API

• Install sdk_info.h even if RERUN_INSTALL_RERUN_C option is OFF #7246 (thanks @traversaro!)

🐍 Python API

• Fix VisualizerOverrides serializer and improved error handling #7288

🦀 Rust API

• Add rerun::external::ndarray #7259
• Handle proper half-size splatting semantics in from_mins_and_sizes #7291

🪳 Bug fixes

• Fix error when trying to clear non-existent component #7215
• Fix gamma (srgb EOTF) for GLTF via Asset3D embedded rgb(a) textures #7251
• Fix Chunk::component_batch_raw not checking the bitmap first #7286
• Fix and test all known HybridResults issues from 0.18 #7297
• Fix secondary plot components ignoring blueprint defaults #7302
• Fix relayout on tab background click #7283
• Update time crate to 0.3.36, fixing compilation on newer Rust versions #7308

🚀 Performance improvements

• Speed up data density graph by rendering them more coarsly #7229
• Default RERUN_CHUNK_MAX_BYTES to 384kiB instead of 4MiB #7263
• Speed up handling of large numbers of transform entities #7300
• Fix memory leak by updating to re_arrow2 0.17.5 #7262

🖼 UI improvements

• Hide time controls if there is only one time point on a timeline #7241

📦 Dependencies

• Correct dependency on puffin to 0.19.1, preventing a possible build failure #7221 (thanks @kpreid!)
• Update time crate to 0.3.36, fixing compilation on newer Rust versions #7228

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.18.1/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.18.1/rerun_cpp_sdk.zip

This patch release comes with the following fixes:

🌊 C++ API

• Install sdk_info.h even if RERUN_INSTALL_RERUN_C option is OFF #7246 (thanks @traversaro!)

🐍 Python API

• Fix VisualizerOverrides serializer and improved error handling #7288

🦀 Rust API

• Add rerun::external::ndarray #7259
• Handle proper half-size splatting semantics in from_mins_and_sizes #7291

🪳 Bug fixes

• Fix error when trying to clear non-existent component #7215
• Fix gamma (srgb EOTF) for GLTF via Asset3D embedded rgb(a) textures #7251
• Fix Chunk::component_batch_raw not checking the bitmap first #7286
• Fix and test all known HybridResults issues from 0.18 #7297
• Fix secondary plot components ignoring blueprint defaults #7302
• Fix relayout on tab background click #7283

🚀 Performance improvements

• Speed up data density graph by rendering them more coarsly #7229
• Default RERUN_CHUNK_MAX_BYTES to 384kiB instead of 4MiB #7263
• Speed up handling of large numbers of transform entities #7300
• Fix memory leak by updating to re_arrow2 0.17.5 #7262

🖼 UI improvements

• Hide time controls if there is only one time point on a timeline #7241

📦 Dependencies

• Correct dependency on puffin to 0.19.1, preventing a possible build failure #7221 (thanks @kpreid!)
• Update time crate to 0.3.36, fixing compilation on newer Rust versions #7228

Rerun is an easy-to-use visualization toolbox for multimodal and temporal data.
Try it live at https://rerun.io/viewer.

📖 Release blogpost: http://rerun.io/blog/column-chunks
🧳 Migration guide: http://rerun.io/docs/reference/migration/migration-0-18

• Python: pip install rerun-sdk
• Rust: cargo add rerun and cargo install rerun-cli --locked
• Online demo: https://rerun.io/viewer/version/0.18.0/
• C++ FetchContent: https://github.com/rerun-io/rerun/releases/download/0.18.0/rerun_cpp_sdk.zip

✨ Overview & highlights

Rerun 0.18 introduces new column-oriented APIs and internal storage datastructures (Chunk & ChunkStore) that can both simplify logging code as well as improve ingestion speeds and memory overhead by a couple orders of magnitude in many cases (timeseries-heavy workloads in particular).

These improvements come in 3 broad categories:

• a new send family of APIs, available in all 3 SDKs (Python, C++, Rust),
• a new, configurable background compaction mechanism in the datastore,
• new CLI tools to filter, prune and compact RRD files.

Furthermore, we started cleaning up our data schema, leading to various changes in the way represent transforms & images.

New send APIs

Unlike the regular row-oriented log APIs, the new send APIs let you submit data in a columnar form, even if the data extends over multiple timestamps.

This can both greatly simplify logging code and drastically improve performance for some workloads, in particular timeseries, although we have already seen it used for other purposes!

API documentation:

• 🐍 Python send_columns docs
• 🌊 C++ send_columns docs
• 🦀 Rust send_columns docs

API usage examples:

rr.init("rerun_example_scalar", spawn=True)

for step in range(0, 64):
    rr.set_time_sequence("step", step)
    rr.log("scalar", rr.Scalar(math.sin(step / 10.0)))

rr.init("rerun_example_send_columns", spawn=True)

rr.send_columns(
    "scalars",
    times=[rr.TimeSequenceColumn("step", np.arange(0, 64))],
    components=[rr.components.ScalarBatch(np.sin(times / 10.0))],
)

const auto rec = rerun::RecordingStream("rerun_example_scalar");
rec.spawn().exit_on_failure();

for (int step = 0; step < 64; ++step) {
    rec.set_time_sequence("step", step);
    rec.log("scalar", rerun::Scalar(std::sin(static_cast<double>(step) / 10.0)));
}

const auto rec = rerun::RecordingStream("rerun_example_send_columns");
rec.spawn().exit_on_failure();

std::vector<double> scalar_data(64);
for (size_t i = 0; i < 64; ++i) {
    scalar_data[i] = sin(static_cast<double>(i) / 10.0);
}
std::vector<int64_t> times(64);
std::iota(times.begin(), times.end(), 0);

auto time_column = rerun::TimeColumn::from_sequence_points("step", std::move(times));
auto scalar_data_collection =
    rerun::Collection<rerun::components::Scalar>(std::move(scalar_data));

rec.send_columns("scalars", time_column, scalar_data_collection);

let rec = rerun::RecordingStreamBuilder::new("rerun_example_scalar").spawn()?;

for step in 0..64 {
    rec.set_time_sequence("step", step);
    rec.log("scalar", &rerun::Scalar::new((step as f64 / 10.0).sin()))?;
}

let rec = rerun::RecordingStreamBuilder::new("rerun_example_send_columns").spawn()?;

let timeline_values = (0..64).collect::<Vec<_>>();
let scalar_data: Vec<f64> = timeline_values
    .iter()
    .map(|step| (*step as f64 / 10.0).sin())
    .collect();

let timeline_values = TimeColumn::new_sequence("step", timeline_values);
let scalar_data: Vec<Scalar> = scalar_data.into_iter().map(Into::into).collect();

rec.send_columns("scalars", [timeline_values], [&scalar_data as _])?;

Background compaction

The Rerun datastore now continuously compacts data as it comes in, in order find a sweet spot between ingestion speed, query performance and memory overhead.

This is very similar to, and has many parallels with, the micro-batching mechanism running on the SDK side.

You can read more about this in the dedicated documentation entry.

Post-processing of RRD files

To help improve efficiency for completed recordings, Rerun 0.18 introduces some new commands for working with rrd files.

Multiple files can be merged, whole entity paths can be dropped, and chunks can be compacted.

You can read more about it in the new CLI reference manual, but to give a sense of how it works the below example merges all recordings in a folder and runs chunk compaction using the max-rows and max-bytes settings:

rerun rrd compact --max-rows 4096 --max-bytes=1048576 /my/recordings/*.rrd > output.rrd

Overhauled 3D transforms & instancing

As part of improving our arrow schema and in preparation for reading data back in the SDK, we've split up transforms into several parts.
This makes it much more performant to log large number of transforms as it allows updating only the parts you're interested in, e.g. logging a translation is now as lightweight as logging a single position.

There are now additionally InstancePoses3D which allow you to do two things:

• all 3D entities: apply a transform to the entity without affecting its children
• Mesh3D/Asset3D/Boxes3D/Ellipsoids3D: instantiate objects several times with different poses, known as "instancing"
  • Support for instancing of other archetypes is coming in the future!

All four tetrahedron meshes on this screen share the same vertices and are instanced using an InstancePoses3D archetype with 4 different translations

⚠️ Breaking changes

• .rrd files from older versions won't load correctly in Rerun 0.18
• mesh_material: Material has been renamed to albedo_factor: AlbedoFactor #6841
• Transform3D is no longer a single component but split into its constituent parts. From this follow various smaller API changes
• Python: NV12/YUY2 are now logged with Image
• ImageEncoded is deprecated and replaced with EncodedImage (JPEG, PNG, …) and Image (NV12, YUY2, …)
• DepthImage and SegmentationImage are no longer encoded as a tensors, and expects its shape in [width, height] order

🧳 Migration guide: http://rerun.io/docs/reference/migration/migration-0-18

🔎 Details

🪵 Log API

• Add Ellipsoids3D archetype #6853 (thanks @kpreid!)
• Dont forward datatype extensions beyond the FFI barrier #6777
• All components are now consistently implemented by a datatype #6823
• Add new archetypes.ImageEncoded with PNG and JPEG support #6874
• New transform components: Translation3D & TransformMat3x3 #6866
• Add Scale3D component #6892
• Angle datatype stores now only radians #6916
• New DepthImage archetype #6915
• Port SegmentationImage to the new image archetype style #6928
• Add components for RotationAxisAngle and RotationQuat #6929
• Introduce TransformRelation component #6944
• New LeafTransform3D, replacing OutOfTreeTransform3D #7015
• Remove Scale3D/Transform3D/TranslationRotationScale3D datatypes, remove Transform3D component #7000
• Rewrite Image archetype #6942
• Use LeafTranslation (centers), LeafRotationQuat and LeafRotationAxisAngle dire...