Raspberry Pi Camera: Difference between revisions

From wikiluntti
VisualWikitext
 
(13 intermediate revisions by the same user not shown)
Line 127: Line 127:


== SSH and Python ==
== SSH and Python ==
=== Virtual Environment ===
Note! To use a virtual environment with a PiCam2, the easiest way to use picamera2 in a virtual environment is to use system-site-packages (see https://forums.raspberrypi.com/viewtopic.php?t=361758)
<code>
python3 -m venv --system-site-packages env
</code>


=== Take a video ===
=== Take a video ===


The following code  
The following code  
Line 148: Line 156:
</code>
</code>


Thus, we might want to use GStreamer or QT, something. A great tutorial about QT can be found at [https://gist.github.com/docPhil99/ca4da12c9d6f29b9cea137b617c7b8b1 How to display opencv video in pyqt apps ].
Thus, we might want to use GStreamer or QT, something. A great tutorial about QT can be found at [https://gist.github.com/docPhil99/ca4da12c9d6f29b9cea137b617c7b8b1 How to display opencv video in pyqt apps] .


=== Python and PiCam2 ===
=== Python, QTPreview and PiCam2 ===


<syntaxhighlight lang="python">
<syntaxhighlight lang="python">
import time
import time
from picamera2 import Picamera2
from picamera2 import Picamera2
import cv2
picam2 = Picamera2()
picam2 = Picamera2()


Line 162: Line 169:
</syntaxhighlight>
</syntaxhighlight>


works.
works or, similarly defining to use QT
<syntaxhighlight lang="python">
from picamera2 import Picamera2, Preview
from time import sleep
from libcamera import Transform
 
picam2 = Picamera2()
picam2.start_preview(Preview.QT, transform=Transform(hflip=True, vflip=True))
picam2.start()
sleep(2)
picam2.close() 
</syntaxhighlight>
According to the manual (https://datasheets.raspberrypi.com/camera/picamera2-manual.pdf) ''The QtGL preview window is not recommended when the image needs to be shown on a remote display (not connected to the Pi).'' Also the DRM driver didn't work out of the box, as is indicated in the manual. See chapter 3.4 Remote preview windows.
 
=== Extracting Frames ===
 
<syntaxhighlight lang="python">
import cv2
import numpy as np
from picamera2 import Picamera2
 
height = 480
width = 640
middle =((width//2),(height//2))
 
cam = Picamera2()
cam.configure(cam.create_video_configuration(main={"format": 'XRGB8888',
                                                          "size": (width, height)}))
cam.start()
 
while True:
    frame = cam.capture_array()
    cv2.circle(frame, middle, 10, (255, 0 , 255), -1)
    cv2.imshow('frame', frame)
    cv2.waitKey(1)
</syntaxhighlight>
 
 
=== Frames & Yolo ===
 
The following code recognizes different things, but the preprocessing time is about 300 ms. Thus it is a bit slow for most applications.
 
<syntaxhighlight lang="python" line>
import cv2
from picamera2 import Picamera2
from ultralytics import YOLO
 
# Initialize the Picamera2
picam2 = Picamera2()
picam2.preview_configuration.main.size = (1280, 720)
picam2.preview_configuration.main.format = "RGB888"
picam2.preview_configuration.align()
picam2.configure("preview")
picam2.start()
 
# Load the YOLO11 model
model = YOLO("yolo11n.pt")
 
while True:
    # Capture frame-by-frame
    frame = picam2.capture_array()
 
    # Run YOLO11 inference on the frame
    results = model(frame)
 
    # Visualize the results on the frame
    annotated_frame = results[0].plot()
 
    # Display the resulting frame
    cv2.imshow("Camera", annotated_frame)
 
    # Break the loop if 'q' is pressed
    if cv2.waitKey(1) == ord("q"):
        break
</syntaxhighlight>
 
=== 1  ===
 
=== 2  ===
 
=== 3  ===


== I ==
== I ==

Latest revision as of 15:12, 12 November 2025

Introduction

Rpicam is based on libcamera rpicam-still

To a raw jpeg file rpicam-still -r -o path/to/file.jpg

Some settings:

  • --qt-preview allow ssh and hardware accelerated preview window remotely using X-forwarding.
  • --timeout 0
  • --shutter 100000 takes a 100 sec photo
  • -r
  • --encoding png
  • -n 1
  • --width 4056 --height 3040
  • -o path/to/file.jpg
  • --gain 1 --awbgains 1,1 --immediate Automatic Exposure/Gain Control (AEC/AGC) and Auto White Balance (AWB). Skip the preview phase entirely with the immediate option

White Balance

libcamera-still --awb custom --awbgains 0.9,2.0

In --awbgains 0.9,2.0, we have --awbgains red,blue

  • 0.9 is the gain for the red channel and
  • 2.0 is the gain for the blue channel.

In other words, if your image is:

  • too red -> decrease the first number
  • too blue -> decrease the second number
  • too yellow -> increase the second number (blue)
  • too green -> increase both numbers
  • too pink ->

RPi camera rev 1.3

Focal Length 3.60 mm.

The v1 camera is based on the Omnivision OV5647.

RPi Cam ver 2

Specs

  • Size ~ 25 × 24 × 9 mm
  • 8 MP, 1080p47, 1640 × 1232p41 and 640 × 480p206 / 3280 × 2464 pixels
  • Sony IMX219
  • Sensor area: 3.68 × 2.76 mm (4.6 mm diagonal); Pixel size 1.12 µm × 1.12 µm
  • Depth of field: Approx 10 cm to ∞
  • f = 3.04 mm
  • H Fov 62.2 deg, v Fov 48.8

Fixed focus module. Adding a +2D lens allows you to focus at about 25cm from the “target”. https://raspi.tv/2013/adapt-your-raspberry-pi-camera-for-close-up-use

Drawings with measures: https://datasheets.raspberrypi.com/camera/camera-module-2-mechanical-drawing.pdf

M12 lenses recommendation: https://www.gaojiaoptotech.com

Macro photos with V2

A V2 camera and screw out the lens until it focuses at 10cm.

M12 lenses

With any M12 mount, you do have to remove the tiny factory-stock lens from the sensor block first.

Hold https://www.thingiverse.com/thing:3347061

Focusing options

Before using these tools, you should take a hobby knife and carefully remove the glue locations so that the lens will rotate freely. You don't need to remove the glue at all: a slight twist of the tool, and the glue gives way with a little snap sound.

Tools for changing the lens (to change the focus distance):

Spectroscope

https://github.com/leswright1977/PySpectrometer2?tab=readme-ov-file

  • A CCTV Lens with Zoom (M12 Thread) (Search eBay for F1.6 zoom lens)

HQ Cameras

a

b

Microscopes

Some microscopes

Macro photography

Reproduction ratio: The ratio of the subject size on the sensor plane to the actual subject size.

d

SSH and Python

Virtual Environment

Note! To use a virtual environment with a PiCam2, the easiest way to use picamera2 in a virtual environment is to use system-site-packages (see https://forums.raspberrypi.com/viewtopic.php?t=361758)

python3 -m venv --system-site-packages env

Take a video

The following code

rpicam-still --qt-preview --timeout 0

works and gives a small window over the ssh port.

Python & Matplotlib

Matplotlib works, and thus eg plt.show() works charmly even with YOLO.

Python & QT

Using Pillow or cv2 makes life harder; v2.imshow("Camera", img) gives an error message

[ WARN:0@0.095] global cap_gstreamer.cpp:1173 isPipelinePlaying OpenCV | GStreamer warning: GStreamer: pipeline have not been created

Thus, we might want to use GStreamer or QT, something. A great tutorial about QT can be found at How to display opencv video in pyqt apps .

Python, QTPreview and PiCam2

import time
from picamera2 import Picamera2
picam2 = Picamera2()

picam2.start(show_preview=True)
time.sleep(5)

works or, similarly defining to use QT 

from picamera2 import Picamera2, Preview
from time import sleep
from libcamera import Transform

picam2 = Picamera2()
picam2.start_preview(Preview.QT, transform=Transform(hflip=True, vflip=True))
picam2.start()
sleep(2)
picam2.close()

According to the manual (https://datasheets.raspberrypi.com/camera/picamera2-manual.pdf) The QtGL preview window is not recommended when the image needs to be shown on a remote display (not connected to the Pi). Also the DRM driver didn't work out of the box, as is indicated in the manual. See chapter 3.4 Remote preview windows.

Extracting Frames

import cv2
import numpy as np
from picamera2 import Picamera2

height = 480
width = 640
middle =((width//2),(height//2))

cam = Picamera2()
cam.configure(cam.create_video_configuration(main={"format": 'XRGB8888',
                                                           "size": (width, height)}))
cam.start()

while True:
    frame = cam.capture_array()
    cv2.circle(frame, middle, 10, (255, 0 , 255), -1)
    cv2.imshow('frame', frame)
    cv2.waitKey(1)


Frames & Yolo

The following code recognizes different things, but the preprocessing time is about 300 ms. Thus it is a bit slow for most applications. 

import cv2
from picamera2 import Picamera2
from ultralytics import YOLO

# Initialize the Picamera2
picam2 = Picamera2()
picam2.preview_configuration.main.size = (1280, 720)
picam2.preview_configuration.main.format = "RGB888"
picam2.preview_configuration.align()
picam2.configure("preview")
picam2.start()

# Load the YOLO11 model
model = YOLO("yolo11n.pt")

while True:
    # Capture frame-by-frame
    frame = picam2.capture_array()

    # Run YOLO11 inference on the frame
    results = model(frame)

    # Visualize the results on the frame
    annotated_frame = results[0].plot()

    # Display the resulting frame
    cv2.imshow("Camera", annotated_frame)

    # Break the loop if 'q' is pressed
    if cv2.waitKey(1) == ord("q"):
        break

1

2

3

I

Retrieved from "https://wiki.luntti.net/index.php?title=Raspberry_Pi_Camera&oldid=14778"