Conversation
…de Gate add_h264_encoder now builds NV12 -> Gate -> VideoEncoder and keeps the gate's control queue. OakSource::set_video_streaming(on) switches the stream at runtime (off = the encoder idles and no video bytes cross the link, the camera keeps serving RGB/depth); OakSource::video_burst(frames, fps) passes a clip then closes again. The gate starts open, so existing callers see no change; OAK_VIDEO_GATED=1 starts it closed (Gate::initialConfig) for consumers that want nothing on the link until they ask. oak_rgbd_probe --gate exercises off / on / off / burst(10) windows. Pins depthai-rs at 51c6d02 (Gate, InputQueue, GateControl, set_initial_config). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XpkL1ayku6dKhdSTstGJJX
…ure, trims
Review pass on the gate change:
- OakSource keeps the H264 { queue, control } the builder returns instead of
splitting it into two Options with a hand-documented invariant.
- add_h264_encoder removes the gate and encoder nodes when any later step fails,
so the degrade path (no video, everything else runs) cannot leave a half-wired
node that fails Pipeline::start for the whole open. Same pattern as the stereo
attacher.
- The gate's input takes the encoder's queue shape (3 deep, blocking) so the
camera is back-pressured as before rather than dropping frames at the gate's
default 1-deep overwrite queue.
- Docs say each thing once; CLAUDE.md folds the gate into the modalities bullet;
the probe's window loop is a function with one duration constant.
- Pins depthai-rs b50c3ae (DAI_GATE macro, no ABI change).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TmCRdrYoyj2zokHSv5EvvM
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TmCRdrYoyj2zokHSv5EvvM
…eaders in calib.rs Simplify + review pass over the whole rewrite (#13 + #14) before merging: - Q<M>: one output queue with its own "first poll failure logged" latch, replacing the Which bitmask enum, the Cell<u8> on OakSource and the queue clones that dodged the borrow checker. OakSource holds the Queues the builder returned instead of re-flattening them. - Session::finish is the one shared tail of both open paths (attach IMU last, start, read the reference camera's intrinsics); Built/from_parts are gone. Knobs are passed into Session::connect, so the session is a function of its arguments. - Calibration readers (intrinsics, stereo calibration, IMU rotation) move from graph.rs to calib.rs next to the types they fill; one pinhole extractor. - One row-pitch normaliser feeds the stereo tightness check and the RGB/depth repacks; depth repacks straight into u16 (one allocation instead of two). - eye_image uses kornia's Image::from_borrowed_host_readonly directly; the kornia-tensor dependency goes. - add_stereo_depth tracks the nodes it made and removes them all on failure (one shape with add_h264_encoder); the no-CAM_A degrade lives in try_add_h264_encoder for both modalities; clamp_imu_hz and the device-id normalisation live in policy.rs with the other rules. - Docs say each thing once; README's open_stereo example has the right arity; CLAUDE.md points at the right files. No public API change; behaviour unchanged (the IMU keeps its per-drain clock offset as the shim always had). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TmCRdrYoyj2zokHSv5EvvM
2 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Stacked on #13. The on-device H.264 path is now
NV12 → Gate → VideoEncoder, andOakSourcecan drive the gate at runtime:set_video_streaming(on): off = the encoder idles and no video bytes cross the link; the colour camera keeps serving RGB/depth. For a saturated PoE/USB2 link, or streaming only while something is worth watching.video_burst(frames, fps): pass a clip, then close again ("record on detection").OAK_VIDEO_GATED=1starts closed viaGate::initialConfig(no frames leak before the first control).oak_rgbd_probe --gateruns off / on / off / burst(10) windows and prints the frame count per window.Pins depthai-rs at
51c6d02(kornia/depthai-rs#1: Gate, InputQueue, GateControl, set_initial_config). Policy (knob, defaults) stays in sensor-oak; the wrapper only mirrors depthai.Test plan
OAK_VIDEO_GATED)cargo run --release --example oak_rgbd_probe -- --gate→ off window ≈ 0 frames, on window ≈ 2·fps, burst ≈ 10. Also confirms the Gate node runs on RVC2 firmware.oak_stereo_probeunchanged (stereo path gains the gate too via the shared builder)🤖 Generated with Claude Code
https://claude.ai/code/session_01XpkL1ayku6dKhdSTstGJJX