Can't get frigate cameras entities to play webrtc

Hello,

I’m looking for help to get frigate or advanced camera card to work with my frigate streams inside HA.

The problem is when I select the camera entity inside any card, it only plays a mjpeg stream, which it shows a “1 FPS” preview, I mean, only 1 snapshot of the camera per second and not the video itself:

When I create a webrtc card and selecting directly the go2rtc stream name, the video streams correctly, so, I believe the entity that the frigate integration is creating, does not have the video streams within:

type: custom:webrtc-camera
url: frontis_main
mode: webrtc

This is my frigate config which is in a separated proxmox CT running docker:

mqtt:
  enabled: true
  host: 192.168.1.15
  user: user
  password: password

go2rtc:
  streams:
    ninas_main:
      - rtsp://user:[email protected]:554/stream1
    ninas_sub: rtsp://user:[email protected]:554/stream2
    #Camara frontis
    frontis_main: 
      rtsp://user:[email protected]:554/Streaming/Channels/101
    frontis_sub: 
      rtsp://user:[email protected]:554/Streaming/Channels/102
  webrtc:
    listen: :8555
    candidates:
      - 192.168.1.3:8555
######## Aceleración de hardware (DECODE GPU) ########
ffmpeg:
  hwaccel_args:
    - -hwaccel
    - vaapi
    - -hwaccel_device
    - /dev/dri/renderD128
    - -hwaccel_output_format
    - yuv420p

detectors:
  ov:
    type: openvino
    device: GPU

model:
  width: 300
  height: 300
  input_tensor: nhwc
  input_pixel_format: bgr
  path: /openvino-model/ssdlite_mobilenet_v2.xml
  labelmap_path: /openvino-model/coco_91cl_bkgr.txt

######## Configuración de cámaras ########
cameras:
  frontis:
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/frontis_main
          roles:
            - record
            - audio
        - path: rtsp://127.0.0.1:8554/frontis_sub
          roles:
            - detect
    record:
      enabled: true
      retain:
        days: 7
        mode: motion

      alerts:
        retain:
          days: 7
          mode: motion
      detections:
        retain:
          days: 7
          mode: motion

    snapshots:
      enabled: true
      retain:
        default: 7

    detect:
      enabled: true
      width: 640
      height: 360

    live:
      streams:
        Main Stream: frontis_main
        Sub Stream: frontis_sub

    objects:
      track:
        - person
        - car
        - motorcycle
        - bus
        - bicycle
        - dog


    motion:
      mask: 
        0.351,0.566,0.337,0.518,0.323,0.436,0.32,0.302,0.331,0.236,0.347,0.156,0.366,0.116,0.398,0.117,0.415,0.174,0.413,0.219,0.409,0.247,0.405,0.271,0.404,0.312,0.398,0.358,0.379,0.392,0.37,0.411,0.362,0.491
  ninas:
    enabled: true

    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/ninas_sub
          roles:
            - detect
        - path: rtsp://127.0.0.1:8554/ninas_main
          roles:
            - record
            - audio
    lpr:
      enabled: true
      expire_time: 5
      min_area: 2000
      enhancement: 1


    record:
      enabled: true
      retain:
        days: 7
        mode: active_objects

      alerts:
        retain:
          days: 7
          mode: active_objects
      detections:
        retain:
          days: 7
          mode: active_objects

    snapshots:
      enabled: true
      retain:
        default: 7

    detect:
      enabled: true
      width: 640
      height: 360

    live:
      streams:
        Main Stream: ninas_main
        Sub Stream: ninas_sub

    objects:
      track:
        - person

    motion:
      mask: 
        0.19,0.982,0.161,0.544,0.162,0.432,0.164,0.16,0.17,0.054,0.355,0.054,0.353,0,0.172,0,0,0,0,0.455,0,0.6,0.002,0.758,0,0.908,0,1,0.202,1
      threshold: 40
      contour_area: 20
      improve_contrast: true



version: 0.16-0

camera_groups:
  Interior:
    order: 1
    icon: LuSquareArrowDown
    cameras: ninas
semantic_search:
  enabled: false
  model_size: small
face_recognition:
  enabled: false
  model_size: small
lpr:
  enabled: true
  device: GPU
classification:
  bird:
    enabled: false

The docker container is in host mode, and I’m able to play the streams using the go2rtc interface at 192.168.1.3:1984.

Anyone has experienced this before? I’ve seen some videos on YouTube showing that the integration works straight when selecting camera entity created by the frigate integration.

Any help will be appreciated!