search

luxonis / depthai-core

DepthAI C++ Library
MIT License
221 stars 120 forks source link

[BUG] OAK-FFC-4P-POE Firmware Crash with Custom Camera Tuning Blob #915

Open justin-larking-pk opened 8 months ago

justin-larking-pk commented 8 months ago

Describe the bug Firmware crash on Oak-ffc-4p-poe and crash dump after enabling custom camera tuning blob.

Specific error: "Error creating output stream: Communication exception - possible device error/misconfiguration. Original message 'Couldn't read data from stream: 'detections_l' (X_LINK_ERROR)'"

A oak-ffc-4p-poe was setup with two ar0234 m12 cameras. Previously the setup was functioning fine and operating a pipeline. However we require low exposure times thus attempted to use the 500us camera tuning blob (https://docs.luxonis.com/projects/api/en/latest/tutorials/image_quality/#:~:text=Camera%20exposure%20limit)

Simply adding this line: pipeline.setCameraTuningBlobPath(camera_blob_path); to our pipeline setup caused a firmware crash and a crash dump to be produced.

crashDump_0_18443010111CCC0F00_f033fd9c7eb0b3578d12f90302e87759c78cfb36.json

Expected behavior The addition of the camera tuning blob would limit max exposure time to 500us on the oak-ffc-4p-poe with two aro234 cameras connected.

Pipeline Graph Pieline details as json here: https://drive.google.com/file/d/1n3JYvl-YDbfGb_6GX9v8m6vP6wpP6CnH/view?usp=sharing pipeline_graph

justin-larking-pk commented 7 months ago

Any update on this?

alex-luxonis commented 7 months ago

Those tuning blobs had some issues with certain sensors. Recommended now to use the branch ae_exp_limit (PR https://github.com/luxonis/depthai-python/pull/921 to be merged soon) where a configurable exposure limit can be set (and updated at runtime as well if desired). It will also use the proper image quality tuning for the specific sensor (AR0234 in this case).