A question about the "frontend" sample error

[jbchoi]

Open two terminals and run the frontend sample simultaneously on each terminal.

1. Running the frontend sample on the first terminal succeeds.

====================================================
nvidia@nvidia-desktop:~$ ./frontend
[INFO] (NvEglRenderer.cpp:110) Setting Screen width 640 height 480
PRODUCER: Creating output stream
PRODUCER: Launching consumer thread
Opening in BLOCKING MODE
Opening in BLOCKING MODE
NvMMLiteOpen : Block : BlockType = 8
===== NVMEDIA: NVENC =====
NvMMLiteBlockCreate : Block : BlockType = 8
892744264
842091865
NVMEDIA: H265 : Profile : 1
[6304136.880325] enc0: Waiting until producer is connected…
Opening in BLOCKING MODE
Opening in BLOCKING MODE
NvMMLiteOpen : Block : BlockType = 8
Opening in BLOCKING MODE
===== NVMEDIA: NVENC =====
NvMMLiteBlockCreate : Block : BlockType = 8
892744264
842091865
Opening in BLOCKING MODE
NvMMLiteOpen : Block : BlockType = 8
===== NVMEDIA: NVENC =====
NvMMLiteBlockCreate : Block : BlockType = 8
892744264
842091865
[6304137.005511] enc2: Waiting until producer is connected…
NVMEDIA: H265 : Profile : 1
[6304137.005708] enc1: Waiting until producer is connected…
NVMEDIA: H265 : Profile : 1
PRODUCER: Starting repeat capture requests.
[6304137.016602] enc0: Producer has connected; continuing.
[6304137.017171] enc1: Producer has connected; continuing.
[6304137.020688] enc2: Producer has connected; continuing.
NVMEDIA_ENC: bBlitMode is set to TRUE
NVMEDIA_ENC: bBlitMode is set to TRUE
NVMEDIA_ENC: bBlitMode is set to TRUE

2. Running the frontend sample on the second terminal fails.

========================================================================
nvidia@nvidia-desktop:~$ ./frontend
[INFO] (NvEglRenderer.cpp:110) Setting Screen width 640 height 480
Error generated. main.cpp, runArgusProducer:97 Failed to get ICaptureSession interface

3. JetPack info

========================================================================
nvidia@nvidia-desktop:~$ cat /etc/nv_tegra_release
R32 (release), REVISION: 5.1, GCID: 26202423, BOARD: t186ref, EABI: aarch64, DATE: Fri Feb 19 16:50:29 UTC 2021

4. In the second frontend sample run, both NANO+IMX219 and TX2 show the same failure message.
   How can I run more than one frontend sample?

[DaneLLL]

Hi,
If you have multiple cameras, yo would need to modify the sample to launch another sensor. Please refer to how the option is implemented in 10_camera_recording:

  -i        Set camera index [Default 0]

[jbchoi]

I use only one sensor on both NANO+IMX21 and TX2.
I want to run multiple frontend samples with one sensor.

[DaneLLL]

Hi,
This case is not supported in default frontend sample. You would need to customize the sample to create multiple OutputStream/consumer, and apply the TensorRT engine to each consumer.

[jbchoi]

We disabled inference with “ENABLE_TRT := 0”.
The OutputStream/consumer you mentioned needs customization is provided by libargus.
Which parts or classes of libargus do I need to customize?

[DaneLLL]

Hi,
Not sure about your use-case. Do you need multi encoding threads? After disabling ENABLE_TRT, there are 3 encoding threads in running frontend sample:

    VideoEncodeStreamConsumer consumer1("enc0", "output1.h265", Size2D<uint32_t>(640, 480));
    VideoEncodeStreamConsumer consumer2("enc1", "output2.h265", Size2D<uint32_t>(1280, 720));
    VideoEncodeStreamConsumer consumer3("enc2", "output3.h265", Size2D<uint32_t>(1920, 1080));
    consumers.push_back(&consumer1);
    consumers.push_back(&consumer2);
    consumers.push_back(&consumer3);

[jbchoi]

We are using multiple processes (Encoding/AutoFocus/Recording/other purposes) for different functions.
Could you explain how to customize it so that it can be used by multiple processes.

[DaneLLL]

Hi,
For this use-case we would suggest use Jetpack 4.6(r32.6.1) so that you can use functions for passing NvBuffer through processes. Please take a look at the discussion in
How to share the buffer in process context？