Optimization of pool to achieve VideoEncoder consistent frame rate

loufylouf

Hello all,

I'm currently testing an OAK-1 camera with Python and DepthAI 3.0.0rc4 (per pip list). I have a pretty straightforward pipeline where I use camera A to feed a VideoEncoder node with the following settings :

ImgFrame type NV12
Profile : H265_MAIN
Resolution : 3296x2464
FPS : 30 (requested)

I mostly get the required frame rate but I have consistently frame drops as you can see from the following log :

[2025-10-22 16:42:42,609][DEBUG] Frame info : instance num=0, sequence num=0, timestamp=25900.699519, delta first frame=0.0000
[2025-10-22 16:42:42,651][DEBUG] Frame info : instance num=0, sequence num=1, timestamp=25900.732886, delta first frame=0.0334
[2025-10-22 16:42:42,682][DEBUG] Frame info : instance num=0, sequence num=2, timestamp=25900.766254, delta first frame=0.0667
[2025-10-22 16:42:42,714][DEBUG] Frame info : instance num=0, sequence num=3, timestamp=25900.799621, delta first frame=0.1001
[2025-10-22 16:42:42,749][DEBUG] Frame info : instance num=0, sequence num=4, timestamp=25900.832988, delta first frame=0.1335
[2025-10-22 16:42:42,782][DEBUG] Frame info : instance num=0, sequence num=7, timestamp=25900.933094, delta first frame=0.2336
[2025-10-22 16:42:42,819][DEBUG] Frame info : instance num=0, sequence num=8, timestamp=25900.966459, delta first frame=0.2669
[2025-10-22 16:42:42,850][DEBUG] Frame info : instance num=0, sequence num=9, timestamp=25900.999829, delta first frame=0.3003
[2025-10-22 16:42:42,892][DEBUG] Frame info : instance num=0, sequence num=10, timestamp=25901.033194, delta first frame=0.3337

I have tried playing around with number of frames in the encoder node, reducing the max output frame size but the frame drop is always there. Am I missing something ? The documentation about pools and max frame size is extremely limited.

The code I'm using is the following :

# ~#~ coding: utf-8
import depthai as dai
import av

from fractions import Fraction
import time
import datetime
import logging
import argparse
import os
import enum


class Encoder(enum.Enum):
    MJPEG = {"profile": dai.VideoEncoderProperties.Profile.MJPEG, "codec_name": "mpeg2"}
    H265 = {"profile": dai.VideoEncoderProperties.Profile.H265_MAIN, "codec_name": "hevc"}


ENCODER = Encoder.H265
FRAME_RATE = 30
LOW_RESOLUTION = (864, 640)
STANDARD_RESOLUTION = (1728, 1280)
HIGH_RESOLUTION = (3296, 2464)

# Create a logger
logger = logging.getLogger("record_video_logger")
logger.setLevel(logging.DEBUG)

# Create a formatter to define the log format
formatter = logging.Formatter("[%(asctime)s][%(levelname)s] %(message)s")

# Create a file handler to write logs to a file
file_handler = logging.FileHandler("record_video.log")
file_handler.setLevel(logging.DEBUG)
file_handler.setFormatter(formatter)

# Create a stream handler to print logs to the console
console_handler = logging.StreamHandler()
console_handler.setLevel(logging.INFO)
console_handler.setFormatter(formatter)

# Add the handlers to the logger
logger.addHandler(file_handler)
logger.addHandler(console_handler)


def record_video(path_recordings, duration=5):
    logger.info(
        "\n*************************************************\n***************** New recording *****************\n*************************************************"
    )

    with dai.Pipeline() as pipeline:
        try:
            # Create the pipeline on the device
            camRgb = pipeline.create(dai.node.Camera).build(dai.CameraBoardSocket.CAM_A)
            output = camRgb.requestOutput(HIGH_RESOLUTION, type=dai.ImgFrame.Type.NV12)
            encoded = pipeline.create(dai.node.VideoEncoder).build(output, frameRate=FRAME_RATE, profile=ENCODER.value["profile"])
            # Use variable bit rate to benefit from situations where the image doesn't change much from frame to frame
            encoded.setRateControlMode(dai.VideoEncoderProperties.RateControlMode.VBR)
            encoded.setMaxOutputFrameSize(int(12540928 / 2))
            encoded.setNumFramesPool(8)
            logger.debug("Number images pool = {0}".format(encoded.getNumFramesPool()))
            encodedQueue = encoded.bitstream.createOutputQueue()
            pipeline.start()

            # Create the file path
            filename = str(datetime.datetime.now().isoformat(sep=" ", timespec="seconds")) + "s.mp4"
            filename = os.path.join(path_recordings, filename.replace(":", "h", 1).replace(":", "m", 1))
            logger.info("Nouveau fichier vidéo crée " + filename)

            output_container = av.open(filename, "w")
            stream = output_container.add_stream(ENCODER.value["codec_name"], rate=FRAME_RATE)
            stream.width = HIGH_RESOLUTION[0]
            stream.height = HIGH_RESOLUTION[1]
            stream.framerate = FRAME_RATE
            stream.time_base = Fraction(1, 1000 * 1000)  # Microseconds

            start = time.time()
            save_frame = False
            number_frames_saved = 0

            # /!\ Timestamps associated with a ImgFrame are in fact a datetime.timedelta
            first_frame_timestamp = None
            first_frame_sequence_number = float("nan")
            while pipeline.isRunning() and time.time() - start < duration:
                image_frame = encodedQueue.get()
                frame_timestamp = image_frame.getTimestamp()

                # Initialize the first frame data and save the frame
                if first_frame_timestamp is None:
                    first_frame_timestamp = frame_timestamp
                    first_frame_sequence_number = image_frame.getSequenceNum()
                    save_frame = True
                # Check if a new frame is present
                elif image_frame.getSequenceNum() != first_frame_sequence_number:
                    save_frame = True

                logger.debug(
                    "Frame info : instance num={0}, sequence num={1}, timestamp={2}, delta first frame={3:.4f}".format(
                        image_frame.getInstanceNum(),
                        image_frame.getSequenceNum(),
                        (frame_timestamp.seconds + frame_timestamp.microseconds / 1000000),
                        (
                            (frame_timestamp - first_frame_timestamp).seconds
                            + (frame_timestamp - first_frame_timestamp).microseconds / 1000000
                        ),
                    )
                )

                # Save the frame
                if save_frame:
                    number_frames_saved = number_frames_saved + 1
                    packet = av.Packet(image_frame.getData())
                    timedelta_between_frames = frame_timestamp - first_frame_timestamp
                    packet.pts = timedelta_between_frames.seconds * 1000000 + timedelta_between_frames.microseconds
                    output_container.mux_one(packet)
                    save_frame = False
        except Exception as e:
            logger.exception(e)

        pipeline.stop()
        pipeline.wait()
        logger.info("DepthAI pipeline stopped")
        logger.info("number of frames saved : {0}".format(number_frames_saved))

        output_container.close()


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--path")
    parser.add_argument("--duration", type=int)
    args, _ = parser.parse_known_args()

    record_video(args.path, args.duration)

I also get the following logs for you to see the CPU stats and memory stats :

[2025-10-22 17:27:14.525] [depthai] [info] Environment variable DEPTHAI_DEBUG is not set. Using default value: ''

[2025-10-22 17:27:14.526] [depthai] [info] Environment variable DEPTHAI_INSTALL_SIGNAL_HANDLER is not set. Using default value: ''

[2025-10-22 17:27:14.527] [depthai] [info] Environment variable DEPTHAI_LIBUSB_ANDROID_JAVAVM is not set. Using default value: ''

[2025-10-22 17:27:14.527] [depthai] [debug] Python bindings - version: 3.0.0-rc.4 from 2025-07-31 19:48:20 +0200 build: 2025-07-31 20:28:20 +0000

[2025-10-22 17:27:14.527] [depthai] [debug] Library information - version: 3.0.0-rc.4, commit: 4adc1dfe3fccb11184ccd05a673673e74e0f30d9 from 2025-07-31 19:48:20 +0200, build: 2025-07-31 20:28:20 +0000, libusb enabled: true

[2025-10-22 17:27:14.548] [depthai] [info] Environment variable DEPTHAI_PROFILING is not set. Using default value: ''

[2025-10-22 17:27:14.548] [depthai] [info] Environment variable XLINK_LEVEL is not set. Using default value: ''

[2025-10-22 17:27:14.548] [depthai] [debug] Initialize - finished

[2025-10-22 17:27:14.604] [depthai] [debug] Resources - Archive 'depthai-bootloader-fwp-0.0.28.tar.xz' open: 4ms, archive read: 69ms

[2025-10-22 17:27:14,656][INFO]

*************************************************

***************** New recording *****************

*************************************************

[2025-10-22 17:27:14.657] [depthai] [info] Environment variable DEPTHAI_SEARCH_TIMEOUT is not set. Using default value: ''

[2025-10-22 17:27:14.657] [depthai] [info] Environment variable DEPTHAI_PROTOCOL is not set. Using default value: ''

[2025-10-22 17:27:14.657] [depthai] [info] Environment variable DEPTHAI_PLATFORM is not set. Using default value: ''

[2025-10-22 17:27:14.734] [depthai] [info] Environment variable DEPTHAI_DEVICE_MXID_LIST is not set. Using default value: ''

[2025-10-22 17:27:14.735] [depthai] [info] Environment variable DEPTHAI_DEVICE_ID_LIST is not set. Using default value: ''

[2025-10-22 17:27:14.735] [depthai] [info] Environment variable DEPTHAI_DEVICE_NAME_LIST is not set. Using default value: ''

[2025-10-22 17:27:14.736] [depthai] [info] Adding device to the filtered list: DeviceInfo(name=1.1, deviceId=19443010516C172F00, X_LINK_UNBOOTED, X_LINK_USB_VSC, X_LINK_MYRIAD_X, X_LINK_SUCCESS)

[2025-10-22 17:27:14.736] [host] [debug] Device - OpenVINO version: universal

[2025-10-22 17:27:14.737] [depthai] [info] Environment variable DEPTHAI_WATCHDOG is not set. Using default value: ''

[19443010516C172F00] [1.1] [1761146834.737] [host] [warning] DEPTHAI_WATCHDOG_INITIAL_DELAY value invalid: invalid stoi argument

[19443010516C172F00] [1.1] [1761146834.739] [host] [debug] Device - BoardConfig: {"camera":[],"defaultImuExtr":{},"emmc":null,"gpio":[],"imu":null,"logDevicePrints":null,"logPath":null,"logSizeMax":null,"logVerbosity":null,"network":{"mtu":0,"xlinkTcpNoDelay":true},"nonExclusiveMode":false,"pcieInternalClock":null,"sysctl":[],"uart":[],"usb":{"flashBootedPid":63037,"flashBootedVid":999,"manufacturer":"","maxSpeed":4,"pid":63035,"productName":"","vid":999},"usb3PhyInternalClock":null,"uvc":null,"watchdogInitialDelayMs":null,"watchdogTimeoutMs":null}

libnop:

0000: b9 13 b9 07 81 e7 03 81 3b f6 81 e7 03 81 3d f6 04 bd 00 bd 00 b9 02 00 01 ba 00 be be bb 00 bb

0020: 00 be be be be be be be 00 bb 00 be be bb 00

[2025-10-22 17:27:14.909] [depthai] [debug] Resources - Archive 'depthai-device-fwp-fef7dc0beab3e77bb7f4a38823c968388c1c5eb9.tar.xz' open: 2ms, archive read: 378ms

[2025-10-22 17:27:14.909] [depthai] [info] Environment variable DEPTHAI_DEVICE_BINARY is not set. Using default value: ''

[2025-10-22 17:27:14.932] [depthai] [info] Environment variable DEPTHAI_CONNECT_TIMEOUT is not set. Using default value: ''

[2025-10-22 17:27:14.933] [depthai] [info] Environment variable DEPTHAI_BOOTUP_TIMEOUT is not set. Using default value: ''

[2025-10-22 17:27:16.639] [depthai] [debug] Searching for booted device: DeviceInfo(name=1.1, deviceId=19443010516C172F00, X_LINK_BOOTED, X_LINK_USB_VSC, X_LINK_MYRIAD_X, X_LINK_SUCCESS), name used as hint only

[2025-10-22 17:27:18.214] [depthai] [info] Environment variable DEPTHAI_CRASHDUMP is not set. Using default value: ''

[19443010516C172F00] [1.1] [1.565] [system] [info] Memory Usage - DDR: 0.12 / 333.26 MiB, CMX: 2.04 / 2.50 MiB, LeonOS Heap: 7.24 / 81.17 MiB, LeonRT Heap: 2.89 / 39.24 MiB / NOC ddr: 0 MB/s

[19443010516C172F00] [1.1] [1.565] [system] [info] Temperatures - Average: 33.33C, CSS: 34.40C, MSS 33.21C, UPA: 32.73C, DSS: 32.97C

[19443010516C172F00] [1.1] [1.565] [system] [info] Cpu Usage - LeonOS 17.13%, LeonRT: 1.22%

[2025-10-22 17:27:18.234] [depthai] [info] Environment variable DEPTHAI_RECORD is not set. Using default value: ''

[2025-10-22 17:27:18.235] [depthai] [info] Environment variable DEPTHAI_REPLAY is not set. Using default value: ''

[19443010516C172F00] [1.1] [1761146838.251] [host] [debug] Schema dump: {"connections":[{"node1Id":0,"node1Output":"0","node1OutputGroup":"dynamicOutputs","node2Id":1,"node2Input":"in","node2InputGroup":""},{"node1Id":1,"node1Output":"bitstream","node1OutputGroup":"","node2Id":2,"node2Input":"in","node2InputGroup":""}],"globalProperties":{"calibData":null,"cameraTuningBlobSize":null,"cameraTuningBlobUri":"","eepromId":0,"leonCssFrequencyHz":700000000.0,"leonMssFrequencyHz":700000000.0,"pipelineName":null,"pipelineVersion":null,"sippBufferSize":18432,"sippDmaBufferSize":16384,"xlinkChunkSize":-1},"nodes":[[0,{"alias":"","id":0,"ioInfo":[[["","inputControl"],{"blocking":true,"group":"","id":0,"name":"inputControl","queueSize":3,"type":3,"waitForMessage":false}],[["dynamicOutputs","0"],{"blocking":false,"group":"dynamicOutputs","id":3,"name":"0","queueSize":8,"type":0,"waitForMessage":false}],[["","mockIsp"],{"blocking":true,"group":"","id":1,"name":"mockIsp","queueSize":8,"type":3,"waitForMessage":false}],[["","raw"],{"blocking":false,"group":"","id":2,"name":"raw","queueSize":8,"type":0,"waitForMessage":false}]],"logLevel":1,"name":"Camera","parentId":-1,"properties":[185,16,185,33,0,3,0,136,0,0,0,0,48,128,224,185,3,129,250,1,130,8,127,6,218,129,252,127,185,5,129,160,223,129,23,218,129,252,127,0,130,240,3,152,19,185,5,129,250,1,0,129,128,28,129,7,218,129,252,127,112,128,193,128,151,0,0,19,128,250,1,0,185,3,0,128,176,53,185,3,133,250,1,134,176,193,151,19,129,250,1,130,48,198,143,19,0,0,120,73,23,128,218,128,252,0,128,168,52,0,186,0,0,189,0,255,255,255,255,255,136,0,0,128,191,0,3,3,4,4,4,186,1,185,5,185,1,184,0,186,2,129,224,12,129,160,9,185,1,190,22,0,190]}],[1,{"alias":"","id":1,"ioInfo":[[["","in"],{"blocking":true,"group":"","id":4,"name":"in","queueSize":3,"type":3,"waitForMessage":false}],[["","bitstream"],{"blocking":false,"group":"","id":5,"name":"bitstream","queueSize":8,"type":0,"waitForMessage":false}],[["","out"],{"blocking":false,"group":"","id":6,"name":"out","queueSize":8,"type":0,"waitForMessage":false}]],"logLevel":1,"name":"VideoEncoder","parentId":-1,"properties":[185,11,0,30,0,0,8,134,0,174,95,0,3,80,0,1,136,0,0,240,65]}],[2,{"alias":"","id":2,"ioInfo":[[["","in"],{"blocking":true,"group":"","id":7,"name":"in","queueSize":3,"type":3,"waitForMessage":false}]],"logLevel":1,"name":"XLinkOut","parentId":-1,"properties":[185,3,136,0,0,128,191,189,15,95,95,120,95,49,95,98,105,116,115,116,114,101,97,109,0]}]]}

[19443010516C172F00] [1.1] [1761146838.252] [host] [debug] Asset map dump: {"map":{}}

[2025-10-22 17:27:18.267] [depthai] [info] Environment variable DEPTHAI_ENABLE_ANALYTICS_COLLECTION is not set. Using default value: ''

[2025-10-22 17:27:18.268] [depthai] [info] Logging disabled

[19443010516C172F00] [1.1] [1.622] [Camera(0)] [debug] Full FoV configs:

[19443010516C172F00] [1.1] [1.622] [Camera(0)] [debug] 1352x1012 @ 52fps

[19443010516C172F00] [1.1] [1.622] [Camera(0)] [debug] 2024x1520 @ 85fps

[19443010516C172F00] [1.1] [1.622] [Camera(0)] [debug] 4056x3040 @ 30fps

[19443010516C172F00] [1.1] [1.622] [Camera(0)] [debug] Output requests: Request 0: size: 3296x2464, fps: 30, type: 22, resizeMode: CROP

[19443010516C172F00] [1.1] [1.622] [Camera(0)] [debug] Best camera config selected: 4056x3040 @ 30fps

[19443010516C172F00] [1.1] [1.623] [system] [info] SIPP (Signal Image Processing Pipeline) internal buffer size '18432'B, DMA buffer size: '16384'B

[19443010516C172F00] [1.1] [1.658] [system] [[2025-10-22 17:27:18,330][INFO] Nouveau fichier vidéo crée C:\Projects\srel\2025-10-22 17h27m18s.mp4

info] ImageManip internal buffer size '393408'B, shave buffer size '36864'B

[19443010516C172F00] [1.1] [1.658] [system] [info] ColorCamera allocated resources: no shaves; cmx slices: [10-15]

ImageManip allocated resources: shaves: [15-15] no cmx slices.

[19443010516C172F00] [1.1] [1.659] [Camera(0)] [debug] Camera scale factors: hNum: 0, hDen: 0, vNum: 0, vDen: 0

[19443010516C172F00] [1.1] [1.659] [Camera(0)] [debug] Patching intrinsics for socket 0

[19443010516C172F00] [1.1] [1.660] [Camera(0)] [debug] Original intrinsics for 3840x2160:

3074.674 0.000 1908.219

0.000 3073.446 1030.946

0.000 0.000 1.000

Patched intrinsics for 4056x3040:

3074.674 0.000 2016.219

0.000 3073.446 1470.946

0.000 0.000 1.000

[19443010516C172F00] [1.1] [2.004] [VideoEncoder(1)] [debug] Bitstream frame size: 12540928, num frames in pool: 8

x265 [info]: HEVC encoder version 4.1+1-1d117be

x265 [info]: build info [Windows][GCC 15.1.0][64 bit] 8bit+10bit+12bit

x265 [info]: using cpu capabilities: MMX2 SSE2Fast LZCNT SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2

x265 [info]: Main profile, Level-5 (Main tier)

x265 [info]: Thread pool created using 12 threads

x265 [info]: Slices : 1

x265 [info]: frame threads / pool features : 3 / wpp(39 rows)

x265 [info]: Coding QT: max CU size, min CU size : 64 / 8

x265 [info]: Residual QT: max TU size, max depth : 32 / 1 inter / 1 intra

x265 [info]: ME / range / subpel / merge : hex / 57 / 2 / 3

x265 [info]: Keyframe min / max / scenecut / bias : 25 / 250 / 40 / 5.00

x265 [info]: Lookahead / bframes / badapt : 20 / 4 / 2

x265 [info]: b-pyramid / weightp / weightb : 1 / 1 / 0

x265 [info]: References / ref-limit cu / depth : 3 / off / on

x265 [info]: AQ: mode / str / qg-size / cu-tree : 2 / 1.0 / 32 / 1

x265 [info]: Rate Control / qCompress : CRF-28.0 / 0.60

x265 [info]: tools: rd=3 psy-rd=2.00 early-skip rskip mode=1 signhide tmvp

x265 [info]: tools: b-intra strong-intra-smoothing lslices=8 deblock sao

[19443010516C172F00] [1.1] [2.567] [system] [info] Memory Usage - DDR: 326.55 / 333.26 MiB, CMX: 2.49 / 2.50 MiB, LeonOS Heap: 21.68 / 81.17 MiB, LeonRT Heap: 7.72 / 39.24 MiB / NOC ddr: 3073 MB/s

[19443010516C172F00] [1.1] [2.567] [system] [info] Temperatures - Average: 38.88C, CSS: 40.28C, MSS 38.65C, UPA: 36.77C, DSS: 39.81C

[19443010516C172F00] [1.1] [2.567] [system] [info] Cpu Usage - LeonOS 31.54%, LeonRT: 45.02%

[19443010516C172F00] [1.1] [3.568] [system] [info] Memory Usage - DDR: 326.55 / 333.26 MiB, CMX: 2.49 / 2.50 MiB, LeonOS Heap: 21.68 / 81.17 MiB, LeonRT Heap: 7.72 / 39.24 MiB / NOC ddr: 5194 MB/s

[19443010516C172F00] [1.1] [3.568] [system] [info] Temperatures - Average: 40.09C, CSS: 41.43C, MSS 39.35C, UPA: 37.48C, DSS: 42.12C

[19443010516C172F00] [1.1] [3.568] [system] [info] Cpu Usage - LeonOS 42.03%, LeonRT: 36.01%

[19443010516C172F00] [1.1] [4.577] [system] [info] Memory Usage - DDR: 326.55 / 333.26 MiB, CMX: 2.49 / 2.50 MiB, LeonOS Heap: 21.68 / 81.17 MiB, LeonRT Heap: 7.72 / 39.24 MiB / NOC ddr: 4622 MB/s

[19443010516C172F00] [1.1] [4.577] [system] [info] Temperatures - Average: 40.91C, CSS: 42.58C, MSS 40.74C, UPA: 38.18C, DSS: 42.12C

[19443010516C172F00] [1.1] [4.577] [system] [info] Cpu Usage - LeonOS 41.34%, LeonRT: 24.47%

[19443010516C172F00] [1.1] [5.585] [system] [info] Memory Usage - DDR: 326.55 / 333.26 MiB, CMX: 2.49 / 2.50 MiB, LeonOS Heap: 21.68 / 81.17 MiB, LeonRT Heap: 7.72 / 39.24 MiB / NOC ddr: 4611 MB/s

[19443010516C172F00] [1.1] [5.585] [system] [info] Temperatures - Average: 40.56C, CSS: 42.35C, MSS 40.28C, UPA: 37.71C, DSS: 41.89C

[19443010516C172F00] [1.1] [5.585] [system] [info] Cpu Usage - LeonOS 42.28%, LeonRT: 28.62%

[19443010516C172F00] [1.1] [6.587] [system] [info] Memory Usage - DDR: 326.55 / 333.26 MiB, CMX: 2.49 / 2.50 MiB, LeonOS Heap: 21.68 / 81.17 MiB, LeonRT Heap: 7.72 / 39.24 MiB / NOC ddr: 4630 MB/s

[19443010516C172F00] [1.1] [6.587] [system] [info] Temperatures - Average: 40.45C, CSS: 41.20C, MSS 40.51C, UPA: 38.18C, DSS: 41.89C

[19443010516C172F00] [1.1] [6.587] [system] [info] Cpu Usage - LeonOS 43.83%, LeonRT: 27.17%

[19443010516C172F00] [1.1] [1761146843.657] [host] [debug] Device about to be closed...

[19443010516C172F00] [1.1] [1761146843.666] [host] [debug] Shutdown OK

[2025-10-22 17:27:23.670] [depthai] [info] XLinkInHost node stopped - exception: Couldn't read data from stream: '__x_1_bitstream' (X_LINK_ERROR)

[19443010516C172F00] [1.1] [1761146843.670] [host] [debug] Timesync thread exception caught: Couldn't read data from stream: '__timesync' (X_LINK_ERROR)

[19443010516C172F00] [1.1] [1761146843.670] [host] [debug] Log thread exception caught: Couldn't read data from stream: '__log' (X_LINK_ERROR)

[19443010516C172F00] [1.1] [1761146844.289] [host] [debug] Watchdog thread exception caught: Couldn't write data to stream: '__watchdog' (X_LINK_ERROR)

[2025-10-22 17:27:24.292] [depthai] [debug] XLinkResetRemote of linkId: (0)

[2025-10-22 17:27:24.294] [depthai] [info] Environment variable DEPTHAI_CRASHDUMP_TIMEOUT is not set. Using default value: '10500'

[19443010516C172F00] [1.1] [1761146844.295] [host] [debug] Device closed, 637

[2025-10-22 17:27:24,295][INFO] DepthAI pipeline stopped

[2025-10-22 17:27:24,295][INFO] Number of frames saved : 147

OskarSonc

Hey @loufylouf
You’re right at the throughput limit (3296×2464@30 NV12 → H.265 with CROP and host muxing), so occasional skips are expected. Maybe you could try to write the encoder bitstream straight to a .h265 file (no PyAV, no per-frame DEBUG) with a large, blocking queue; then avoid CROP by using a native mode like 4056×3040@30 or standard 3840×2160@30 and see if drops disappear; and please also try with newer DepthAI 3.1.x which has pacing/pool improvements.
Thanks
Oskar

loufylouf

OskarSonc Thanks Oskar ! So I've tried the following :

Reducing the throughput with the original, uncropped ratio, and a size of 3072x2016, and I still have the same frame drops, though the debug logs are showing a significant improvment on the DDR usage
Updating to DepthAI 3.1.0 : it actually increased the number of frames drops
Removed the logging per frame and removed the PyAV call to mux_one : no impact
I have enabled the profiling through the environnement variable : doesn't provide much info, the top read (?!) speed reported is around 45MiB/s

I haven't been able yet to test setting a large blocking queue, how would I do that ? It is a bit unclear from the documentation

I also just saw that there seems to be different pools, can I also increase the size of the pool for the camera node ?

[19443010516C172F00] [1.1] [1.897] [Camera(0)] [info] Camera::Create: Number of frames for the Outputs frame pool was capped by the max size of the pool or increased to minPoolFramesOutputs (=2) to 2 (orig: 3) | maxSizePoolOutputs: 10485760 B | output frame size: 9289728 B
[19443010516C172F00] [1.1] [2.186] [VideoEncoder(1)] [debug] Bitstream frame size: 9503744, num frames in pool: 8