rpicam-raw -n -t 5000 --segment 1 -o test_%04d.raw --mode 4608:2592:10:P --brightness 0.0 --contrast 0.7 --exposure normal --framerate 15 --gain 0 --awb auto --metering centre --saturation 1.0 --sharpness 1.5 --denoise off
- Packed in 10 Bits has the following Format:
- Raw stream: 4608x2592 stride 5760 format SBGGR10_CSI2P
- 1 Byte: MSB B
- 2 Byte: MSB G
- 3 Byte: MSB G
- 4 Byte: MSB R
- 5 Byte: LSB BGGR, each 2 Bits
cvlc -vvv v4l2://dev/video0 --v4l2-width 160 --v4l2-height 120 --sout "#transcode{vcodec=mp2v,acodec=mpga,fps=30}:rtp{mux=ts,sdp=rtsp://:8080/test.sdp}"- Documentation of Producer does not work correctly, only vlc with a desktop environment worked. Over SSH opening of /dev/video0 was not possible
- The documentation says, that over SPI metadata can be read out like current module temperature
- https://forums.openmv.io/t/lepton-3-5-disabling-enabling-shutter-ffc-manually/7413/15
- AGC must be deactivated to read exact temperatures See here
- Complete Hacker.io project logs: https://hackaday.io/project/159615/logs?sort=oldest
- https://cdn.sparkfun.com/assets/f/6/3/4/c/Lepton_Engineering_Datasheet_Rev200.pdf
- https://media.digikey.com/pdf/Data%20Sheets/GroupGets%20PDFs/PURETHERMAL-3_Rev2_Oct2022.pdf
- https://github.com/rob-coco/leptonic/tree/bookworm-update
- https://github.com/groupgets/LeptonModule/wiki
- https://learn.sparkfun.com/tutorials/flir-lepton-hookup-guide/all --> Not working (SPI of board faulty?)
- https://github.com/groupgets/purethermal1-uvc-capture Showing images with Guvcview working on bookworm --> https://github.com/jaswdr/guvcview
- https://github.com/groupgets/purethermal1-firmware Firmware can be modified to change the behaviour of the camera
- https://www.youtube.com/watch?v=eAxDrWSUIRc --> Mechanical information of how to assemble Flir Lepton
#Code to reformat 1280*720 12bit raw SRGGB12_CSI2P file into an 8bit BGR file for OpenCV
#Note... discards least significant 4 bits from each 12bit pixel value to get 8 bit format
import cv2
import numpy as np
rows = 720
cols = 1280
stride =1920
f = open('/home/youruser/test.raw','rb')
rawf =np.fromfile (f, dtype=np.uint8,count=rows*stride)
f.close()
#Add an additional single element at array start as the next operation will
#discard the zeroth element and we want to keep it!
rawf= np.concatenate(([rawf[0]],rawf))
#create new array with every third element from original deleted.
#This discards each third composite byte which contains two sets of 4 LSBs from previous two elements
raw8bit= np.delete(rawf,np.arange(0,rawf.size,3))
#Create final image format
bayer8=raw8bit.reshape(rows,cols)
#and populate with appropriate sub-pixels from bayer pattern
r = bayer8[0::2, 0::2]
g0 = bayer8[0::2, 1::2]
g1 = bayer8[1::2, 0::2]
b = bayer8[1::2, 1::2]
#Do some ad-hoc scaling to get quick and dirty but 'presentable' image.
#No gamma or any accurate colour space corrections!
R=(r)*3
B=(b)*4
G=((g0+g1)/2)*2.4
#Stack the colour planes into format used by OpenCV
BGR8=np.dstack((B,G,R)).astype(np.uint8)
#and display.....
cv2.imshow('raw to BGR',BGR8)
cv2.waitKey()
cv2.destroyAllWindows()