Add command for camera setup prior capture loop

README.md

@@ -22,6 +22,21 @@ set -o allexport; source .env; set +o allexport
 
 ```
 
+Any linux host with ffmpeg and RTSP camera:
+
+```shell
+# go to Prusa Connect site, add camera, copy token, turn on your 3D printer
+
+cp ffmpeg-mediamtx-rtsp-tcp.dist .env
+# edit .env file, set token and fingerprint
+# change 'rtsp://raspberry-pi:8554/cam' to rtsp://<username>:<password>@<camera-ip-address>/<stream-name>
+
+set -o allexport; source .env; set +o allexport
+
+./prusa-connect-camera.sh
+
+```
+
 ## Docs
 
 ```shell

docs/config.for.camera.csi.legacy.md

@@ -11,3 +11,16 @@ Example for older operating systems with (command `raspistill`):
 - save edited file `.env`
 
 Next, [test config](./test.config.md).
+
+## Real world scenario
+
+Some older RPI 3 with older Debian with basic cam:
+
+```shell
+PRINTER_ADDRESS=127.0.0.1
+PRUSA_CONNECT_CAMERA_TOKEN=token-change-me
+PRUSA_CONNECT_CAMERA_FINGERPRINT=trash-cam-night-video-wide-1
+CAMERA_DEVICE=/dev/video0
+CAMERA_COMMAND=raspistill
+CAMERA_COMMAND_EXTRA_PARAMS="--nopreview --mode 640:480 -o"
+```

docs/config.for.camera.csi.libcamera.md

@@ -11,3 +11,18 @@ Example for newer operating systems (commands `libcamera` or `rpicam-still`):
 - save edited file `.env`
 
 Next, [test config](./test.config.md).
+
+## Real example
+
+My Rpi Zero W with Raspberry Pi Camera v2 with maximum resolution available
+
+<!-- markdownlint-disable line_length -->
+```shell
+PRINTER_ADDRESS=192.168.1.25
+PRUSA_CONNECT_CAMERA_TOKEN=redatcted
+PRUSA_CONNECT_CAMERA_FINGERPRINT=c10eb887-f107-41a4-900e-2c38ea12a11c
+CAMERA_DEVICE=/dev/video0
+CAMERA_COMMAND=rpicam-still
+CAMERA_COMMAND_EXTRA_PARAMS="--immediate --nopreview --mode 2592:1944:12:P --lores-width 0 --lores-height 0 --thumb none -o"
+```
+<!-- markdownlint-enable line_length -->

docs/config.for.camera.esphome.snapshot.md

@@ -35,3 +35,19 @@ Flash the device and wait until it boots and is available.
 - save edited file `.env`
 
 Next, [test config](./test.config.md).
+
+## Real world example
+
+I have esp32-wrover-dev board with camera + esphome + web ui for camera exposing
+snapshot frame on port `8081`.
+
+We can use curl to fetch it.
+
+```shell
+PRINTER_ADDRESS=127.0.0.1
+PRUSA_CONNECT_CAMERA_TOKEN=redacted
+PRUSA_CONNECT_CAMERA_FINGERPRINT=06f47777-f179-4025-bd80-9e4cb8db2aed
+CAMERA_DEVICE=/dev/null
+CAMERA_COMMAND=curl
+CAMERA_COMMAND_EXTRA_PARAMS=http://esp32-wrover-0461c8.local:8081/ -o
+```

docs/config.for.camera.esphome.stream.md

@@ -4,6 +4,8 @@ With esphome camera stream we can use the `ffmpeg` to fetch the image from the
 camera stream. It requires a bit more computing power from esp device and the
 host that runs the image processing.
 
+Notice that this is not recommended way due to the amout of consumed resources.
+
 ## Prepare esphome device
 
 Configure esphome device:
@@ -37,3 +39,18 @@ Flash the device and wait until it boots and is available.
 - save edited file `.env`
 
 Next, [test config](./test.config.md).
+
+## Real world example
+
+The same ESP device with stream, notice different port (`8080`).
+
+<!-- markdownlint-disable line_length -->
+```shell
+PRINTER_ADDRESS=127.0.0.1
+PRUSA_CONNECT_CAMERA_TOKEN=token-change-me
+PRUSA_CONNECT_CAMERA_FINGERPRINT=f68336b-8dab-42cd-8729-6abd8855ff63
+CAMERA_DEVICE=/dev/null
+CAMERA_COMMAND=ffmpeg
+CAMERA_COMMAND_EXTRA_PARAMS="-y -i 'http://esp32-wrover-0461c8.local:8080/' -vframes 1 -q:v 1 -f image2 -update 1 "
+```
+<!-- markdownlint-enable line_length -->

docs/config.for.camera.md

@@ -2,6 +2,13 @@
 
 Notice that each camera should have different fingerprint and token.
 
+Fingerprint can be easily generated using command:
+
+```shell
+uuidgen
+```
+just copy/paste the output as fingerprint value into the config.
+
 **Do not** change fingerprint after launching the script - thus camera is registered
 and you may need to revert the change or delete and readd camera again and start
 from scratch.

docs/config.for.camera.rtsp.md

@@ -33,3 +33,43 @@ If that works, then configuration should be pretty straightforward:
 You can try with `UDP`, but you may not get it ;-)
 
 Next, [test config](./test.config.md).
+
+## Real world example
+
+My another Rpi Zero W named `hormex` has two cameras:
+
+- CSI
+- endoscope on /dev/video
+
+and I'm running `mediamtx` server to conver those to RTSP streams.
+More about mediamtx is [here](./stream.mediamtx.md).
+
+So I can have two configs:
+
+`.stream-csi` over UDP:
+
+<!-- markdownlint-disable line_length -->
+
+```shell
+PRINTER_ADDRESS=127.0.0.1
+PRUSA_CONNECT_CAMERA_TOKEN=redacted
+PRUSA_CONNECT_CAMERA_FINGERPRINT=62e8ab72-9766-4ad5-b8b1-174d389fc0d3
+CAMERA_DEVICE=/dev/null
+CAMERA_COMMAND=ffmpeg
+CAMERA_COMMAND_EXTRA_PARAMS="-loglevel error -y -rtsp_transport udp -i "rtsp://hormex:8554/cam" -f image2 -vframes 1 -pix_fmt yuvj420p "
+```
+<!-- markdownlint-enable line_length -->
+
+`.stream-endo` over TCP:
+
+<!-- markdownlint-disable line_length -->
+```shell
+PRINTER_ADDRESS=127.0.0.1
+PRUSA_CONNECT_CAMERA_TOKEN=redacted
+PRUSA_CONNECT_CAMERA_FINGERPRINT=01a67af8-86a3-45c7-b6e2-39e9d086b367
+CAMERA_DEVICE=/dev/null
+CAMERA_COMMAND=ffmpeg
+CAMERA_COMMAND_EXTRA_PARAMS="-loglevel error -y -rtsp_transport tcp -i "rtsp://hormex:8554/endoscope" -f image2 -vframes 1 -pix_fmt yuvj420p "
+
+```
+<!-- markdownlint-enable line_length -->

docs/config.for.camera.usb.md

@@ -25,3 +25,18 @@ Notice not every device is an actual camera.
 - save edited file `.env`
 
 Next, [test config](./test.config.md).
+
+## Real world example
+
+Raspberry Pi Zero W with endoscope camera over USB, registered as `/dev/video1`:
+
+<!-- markdownlint-disable line_length -->
+```shell
+PRINTER_ADDRESS=192.168.1.25
+PRUSA_CONNECT_CAMERA_TOKEN=redacted
+PRUSA_CONNECT_CAMERA_FINGERPRINT=7054ba85-bc19-4eb9-badc-6129575d9651
+CAMERA_DEVICE=/dev/video1
+CAMERA_COMMAND=fswebcam
+CAMERA_COMMAND_EXTRA_PARAMS="--resolution 1280x960 --no-banner"
+```
+<!-- markdownlint-enable line_length -->

docs/configuration.tuning.md

@@ -18,3 +18,81 @@ For Raspberry Cam v2 you could use `csi.dist` as ssource and add
 
 For certain USB cameras (such as Tracer Endoscope) you shoudl use `usb.dist` and
 you should be able to add `--resolution 1280x960` to the `CAMERA_COMMAND_EXTRA_PARAMS`.
+
+## Setting up video camera controls
+
+Get device capabilities, especially User controls
+
+```shell
+v4l2-ctl -d /dev/video0 --all
+```
+
+and set accordingly parameters you want in `CAMERA_SETUP_COMMAND` env var, for example:
+
+```shell
+CAMERA_SETUP_COMMAND="v4l2-ctl --set-ctrl brightness=64,gamma=300 -d $CAMERA_DEVICE"
+```
+
+remember to restart given camera service.
+
+You can try to use `guvcview` desktop application to check prams in realtime.
+
+## Image flip and rotation
+
+You can pass on params to rpicam-still or fswebcam as you want.
+
+### rpicam-still
+
+See `rpicam-still --help`
+
+```text
+  --hflip      Read out with horizontal mirror
+  --vflip      Read out with vertical flip
+  --rotation   Use hflip and vflip to create the given rotation <angle>
+```
+
+so for example:
+
+```shell
+CAMERA_COMMAND=rpicam-still
+CAMERA_COMMAND_EXTRA_PARAMS="--rotation 90 --immediate --nopreview --thumb none -o"
+```
+
+### fswebcam
+
+See `fswebcam --help`
+
+```text
+  --flip <direction>       Flips the image. (h, v)
+  --crop <size>[,<offset>] Crop a part of the image.
+  --scale <size>           Scales the image.
+  --rotate <angle>         Rotates the image in right angles.
+```
+
+so for example:
+
+```shell
+CAMERA_COMMAND=fswebcam
+CAMERA_COMMAND_EXTRA_PARAMS="--flip v --resolution 640x480 --no-banner"
+```
+
+## ffdshow
+
+With ffdshow you can things with filters, it will just require more computing power.
+
+See [here](https://superuser.com/questions/578321/how-can-i-rotate-a-video-180-with-ffmpeg)
+for basic ones.
+
+You probably want to use `-vf "transpose=1"` to rotate image 90 degrees clockwise:
+<!-- markdownlint-disable line_length -->
+```shell
+CAMERA_COMMAND=ffmpeg
+CAMERA_COMMAND_EXTRA_PARAMS="-y -i 'http://esp32-wrover-0461c8.local:8080/' -vf "transpose=1" -vframes 1 -q:v 1 -f image2 -update 1 "
+```
+<!-- markdownlint-disable line_length -->
+
+Frankly speaking you can do anything you want with ffdshow, for example
+
+`-vf transpose=1,shufflepixels=m=block:height=16:width=16`
+
+Why? why not :D

docs/index.md

@@ -5,6 +5,18 @@
 This project aims to make it easier to use any camera to be used as
 Prusa Connect camera.
 
+## Features
+
+- allows to read images from CSI cameras, USB cameras, RTSP streams, still images...
+- do not send pictures if the printer is offline
+- store data in memory to prevent MicroSD wear out
+- verbose error messages to see if the image capture works
+- ability to run multiple cameras in separate instances
+
+## Planned
+
+- run in container
+
 ## Architecture
 
 ```mermaid
@@ -25,8 +37,6 @@ sequenceDiagram
 
 ## Known limitations
 
-- TODO: camera pre-configuration is not yet supported
-  (such as fixed white balance etc)
 - this script performs processing of the single camera, if you need more cameras
   then just create multiple copies with different settings (see below)
 - Rpi Zero W or older devices may have CPU limitations to process remote streams

docs/stream.mediamtx.md

@@ -1,6 +1,7 @@
 # mediamtx
 
-Use mediamtx on antoher Raspberry Pi to create RTSP camera stream for test.
+Use [mediamtx](https://github.com/bluenviron/mediamtx) on another Raspberry Pi
+to create RTSP camera stream for test.
 
 Assuming you run [mediamtx with Raspberry Pi CSI camera](https://github.com/bluenviron/mediamtx#raspberry-pi-cameras)
 and that `raspberry-pi` is the hostname of your device and that you expose two cams:
@@ -32,7 +33,7 @@ Start mediamtx server:
 ```
 
 This should allow us to reach two streams, replace `rpi-address` with the name
-of your Raspberry Pi hostname or IP address
+of your Raspberry Pi hostname or IP address. The ports are default for mediamtx.
 
 ```shell
 ffplay rtsp://rpi-address:8554/cam

prusa-connect-camera.sh

@@ -19,6 +19,17 @@
 # camera device to use, if you use camera attached to the CSI via camera ribbon then leave as is
 : "${CAMERA_DEVICE:=/dev/video0}"
 
+# camera setup command and params
+# executed before taking image
+# defualt is empty, because some cameras do not support it
+# in general you want to use something like v4l2-ctl
+# <setup_command>
+# so for example:
+# setup_command=v4l2-ctl --set-ctrl brightness=10,gamma=120 -d $CAMERA_DEVICE
+# will translate to:
+#   v4l2-ctl --set-ctrl brightness=10,gamma=120 -d /dev/video0
+: "${CAMERA_SETUP_COMMAND:=}"
+
 # avaliable options:
 # rpicam-still - using CSI camera + modern Rasberry Pi operating systems since Debian 11 Bullseye
 # raspistill - using CSI camera + older Raspberyy Pi operating systems
@@ -137,6 +148,13 @@ if [[ -z "${command_capture}" ]]; then
   exit 1
 fi
 
+CAMERA_SETUP_COMMAND
+if [[ -n "${CAMERA_SETUP_COMMAND}" ]]; then
+  echo "INFO: Running CAMERA_SETUP_COMMAND command"
+  echo "${CAMERA_SETUP_COMMAND}"
+  eval ${CAMERA_SETUP_COMMAND}
+fi
+
 echo "Camera capture command: ${command_capture} ${CAMERA_COMMAND_EXTRA_PARAMS} ${TARGET_DIR}/camera_${camera_id}.jpg"