Raspberry Pi HQ Camera 12MP

[ulf]

I think the new Raspberry Pi HQ Camera 12MP can be a base for a multispectral project with a small camera.

 

The used sensor's sensitivity graphs looks promising:

https://www.leopardimaging.com/uploads/LI-IMX477-MIPI-CS_datasheet.pdf

RPi HQ camera specification:

• Sensor – 12.3MP Sony IMX477R stacked, back-illuminated sensor; 7.9 mm sensor diagonal, 1.55 μm × 1.55 μm pixel size
  
• Output – RAW12/10/8, COMP8
  
• Back focus – Adjustable (12.5 mm–22.4 mm)
  
• Lens standards – C-mount and CS-mount (C-CS adapter included)
  
• Integrated IR cut filter (can be removed to enable IR sensitivity, but the modification is irreversible)
  
• 20 cm Ribbon cable
  
• Tripod mount – 1/4”-20
  
• Compliance – FCC, EMC-2014/30/EU, RoHS, Directive 2011/65/EU
  
• Long term availability – In production until at least January 2026
  

The BG-filter in front of the actual sensor-chip hopefully isn't difficult to remove.

 

Such a project can be interesting for someone less software challenged than me.

[dabateman]

Actually the new Rasperian software looks just like linux to me. Very easy.

The code to snap a photo is simple

 

raspistill -o test.jpg

 

The IR blocking filter is just held by glue. I would first test UV sensitivity prior to removing the filter. The instructions on removal just say to push down on it so it breaks and pops the glue bond. I would rather use an exacto blade and carefully cut it off.

 

https://www.raspberrypi.org/documentation/hardware/camera/hqcam_filter_removal.md

 

So $100 for raspberry computer.

$50 for hq camera.

Not bad and maybe worth a test.

[Stefano]

You could debayer it to improve UV sensitivity. The lens can be made of quartz, and it wouldn't be too expensive being small.

[colinbm]

There is already an IR camera for the Raspberry Pi, I wonder if that is full spectrum & mono ?

[Stefano]

There is already an IR camera for the Raspberry Pi, I wonder if that is full spectrum & mono ?

It should be full spectrum.

[dabateman]

The Raspberry pi NoIR sensor has fixed lens. So not sure how UV sensitive it would be.

It uses a 1/4 inch Sony IMX219 sensor which is BSI and reported to have low dark current. Its a 1.12um pixel and 8Mpixels.

 

The lens would limit you.

 

The HQ being CS mount is great. I hope it doesn't have a Sony UV blocking cover slip though. The blocking on Jonathan's A7m3 coverglass worries me for the future. It really kills all uv below 350nm.

 

The new Olympus Em1 mk3 and Em1x with Sony 20Mpixel sensors are reported in the press releases to have better coatings on the sensor. This to me is bad as might kill off the deep UV.

The Olympus Em1mk1 might just be the last all star UV camera.

[colinbm]

Thanks Dave

[colinbm]

This was written before the new Raspberry Pi HQ Camera 12MP that Ulf mentions, but some extra background....

http://www.volcano-blog.com/blog/low-cost-ultraviolet-camera-measurements-using-the-raspberry-pi

https://www.mdpi.com/1424-8220/16/10/1649

[dabateman]

Yep I remember those. Fun researching all the chemicals to see if I could strip a sensor for Monochrome.

Those papers didn't talk about base sensitivity in the unstripped sensors though.

So you could buy an old fixed lens pi camera. Crack off the lens. Remove the blocking filter and 3D print a S-mount but be disappointed still by the very low UV signal.

They may need to be stripped to monochrome.

I was very disappointed by the UV sensitivity of my Omnivision sensor in my Axis212 camera. Basically nothing. I didn't bother stripped to monochrome.

 

They briefly mentioned the Sony IMX219 sensor, but take it no further. I wonder why?

 

This new Raspberry pi HQ sensor is interesting though. $50, then add the basically Debian linux main computer for $80, add a 4" touch screen for $50 and you have a full camera.

 

So under $200, you have a programmable camera. You can copy and past current code or download other camera applications. Almost seems like a real option. Might try out one day. Wonder if a monochrome version could become available. Also since the camera module just plugs in can swap out normal, UV, and monochrome cameras modules to the same board based camera. Or for $600 just build all 3.

 

 

[dabateman]

Ulf,

Here is a great tutorial with fails to help you build and program your camera.

https://www.hackster.io/TeCoEd/night-vision-101c0f

 

The HQ look to have rapidly sold out though.

 

 

Even better the full camera programming guide as a free pdf:

https://magpi.raspberrypi.org/books/camera-guide/pdf

[WiSi-Testpilot]

Thanks for the information about the new camera.

I have a Nvidia Jetson Nano with two R-Pi cameras V2. I will order the new camera and hope that it runs out of the box. Here are a few sample pictures. It shows my nano with 4 cameras including an output, an example for an averager and the result of my chroma keying software.

Best regards,

Wilhelm

[colinbm]

Hi Wilhelm

I will look forwards to you results.

Raspberry Pi High Quality Camera

Works with...

https://raspberry.piaustralia.com.au/collections/cameras/products/raspberry-pi-hq-camera

Col

[WiSi-Testpilot]

Thank you Col.

The nano is still not tested. However I guess it will work. I am sure, otherwise Nvidia will update the nvarguscamerasrc plugin.

 

Programming is not difficult. I always take examples from the internet / nvidia forum and modify them.

For example, this is the Python3 / Gstreamer / Opencv code for the Averager.

The variables csi or summe are arrays with the dimensions of the video format. As usual for arrays one has access to every dot and can apply graphical commands too.

The nv plugins are running on the hardware image signal processor and some Opencv commands are CUDA accelerated.

Best regards,

Wilhelm

 

import numpy as np
import cv2
import time

def willi():

cap = cv2.VideoCapture("nvarguscamerasrc sensor_id=1 maxperf=true ! video/x-raw(memory:NVMM), width=1280, height=720, format=(string)NV12, framerate=60/1 ! nvvidconv flip-method=2 ! video/x-raw, width=1280, height=720, format=(string)BGRx ! videoconvert ! video/x-raw, format=(string)BGR ! appsink", cv2.CAP_GSTREAMER)
if cap.isOpened():
	cv2.namedWindow("willi", cv2.WINDOW_AUTOSIZE)
	counter = 0
	while cv2.getWindowProperty("willi", 0) >= 0:

		ret, csi = cap.read()
			  
		cv2.imshow("CSI Kamera", csi)

		summe = np.asfarray(csi)
		summe = summe * 0.0

		for i in range (0,100):
			ret, csi = cap.read()	  
			summe = summe + np.asfarray(csi)

		summe = summe / (255 * 100)
		cv2.imshow("Average", summe )
		counter = counter + 1
		print("Sekunde: ", time.strftime("%S"),"100 x Acc:",counter)

		keyCode = cv2.waitKey(30) & 0xFF
		# Stop the program on the ESC key
		if keyCode == 27:
			break

	cap.release()
	cv2.destroyAllWindows()
else:
	print("Unable to open camera")

if __name__ == "__main__": willi()

[colinbm]

Thanks Wilhelm

I have never done any programing, so this is is all Chinese to me, sorry....:-(

Col

[colinbm]

Looks like everyone is into it.....?

https://www.asus.com/au/Single-Board-Computer/Tinker-Board/

[ulf]

Ulf,

Here is a great tutorial with fails to help you build and program your camera.

https://www.hackster...t-vision-101c0f

Thanks David, but I do not have any such camera and no desire to get one either.

My life is too filled up with other projects and lacking time to learn programming from start again.

[WiSi-Testpilot]

I made a quick test with the Raspberry Pi CSI V2 Cameras with and without my Baader U filter.

Best regards,

Wilhelm

 

Setup

 

UV and Vis

 

Full spectrum and Vis

[Herman1705]

Pi HQ camera connected to Raspberry Pi 3

 

VIS image; ISO 40; 1/690 s

 

Baader U filter; ISO 250; 1/8 s; recorded as jpeg image, no further processing.

 

Hoya R72 filter; ISO 250; 1/8 s; recorded as jpeg image, no further processing.

 

The IR camera filter not yet removed.

The software dumps the RAW data too as attachment to the JPEG; still not any succesfull use of this data.

[Andy Perrin]

I use EXIFTOOL to extract the raw data from the JPEGs produced by the FLIR cameras. I wonder if that would work for you?

[colinbm]

Thanks for doing these Herman

[WiSi-Testpilot]

Pi HQ camera connected to Raspberry Pi 3

...

 

Thank you, Herman.

What is the diameter of the filter thread?

I ordered one, but is backordered.

Best regards,

Wilhelm

[Herman1705]

The removal of the Ir-blocking filter was quite easy.

Extraction of the embedded RAW data with Matlab: https://www.strollsw...ty-camera-raw/.

With GNU Ocatve the Matlab code has to be rewritten: the DigitalCamera struct is incomplete.

Not all of the Matlab code is needed; is does white balancing (extracted from the jpeg EXIF) and composition to sRGB too.

 

Ir:

 

RAW composite; filter Haida 950nm; ISO 400; 1/50s; f/5.6;

 

> 800 nm the sensor should produce monochrome images I think.

adjusted the B and R channel to fit with the G channel (called it white balancing )

 

 

UV:

 

RAW composite; filter Baader_U; ISO 800; 1 s; f/5.6

 

same adjustments on B and R as the Ir image

 

 

Next challenge: single image IrRG with the Tiffen yellow 15 (or 12) filter.

First have to improve this before IrRG recomposition.

RAW composite RGB (not IrRG); filter Tiffen 15; ISO 400; 1/4673 s; f/5.6

[Herman1705]

Thank you, Herman.

What is the diameter of the filter thread?

I ordered one, but is backordered.

Best regards,

Wilhelm

 

The filter thread is 37 mm.

[WiSi-Testpilot]

Thank you, Herman.

Did you use the telelens?

It seems for me, that the lens absorbs a lot of UV.

Meanwhile I got the camera and an adapter, but have too many projects.

Best regards,

Wilhelm

[Herman1705]

UV 'transmission' 16mm CCTV lens for the Raspberry Pi HQ camera.:

 

Setup for the shot: Sony A580; EL-Nikkor f/5.6 105mm; Baader-U filter; ISO 400; 1 s; f/8. UV reflector: PTFE board. Light source: sun.

 

Steinhel München Cassar S f/2.8 50mm; --------------------- CCTV f/1.4 16mm;------------------------- ENNA München Lithagon f/3.5 35mm

 

Poor UV transmission of the 16mm CCTV lens.