UNSOLVED VEYE-MIPI-IMX385 hot pixels

JaCC

Hello,
during nighttime operation we noticed quite a large amount of hot pixels at gain values ≥ 25. Additionally, it seems like pixels below a certain brightness threshold get set to black? We believe this because objects in the foreground are clearly visible, but anything beyond approximately 2 meters becomes almost imperceptible.

This is a script used to set the i2c parameters:

./veye_mipi_i2c.sh -w -f videoformat -p1 PAL
sleep 1
./veye_mipi_i2c.sh -w -f sharppen -p1 0
sleep 0.2
./veye_mipi_i2c.sh -w -f irtrigger -p1 0
sleep 0.2
./veye_mipi_i2c.sh -w -f lowlight -p1 0x00 #fixed 25 fps
sleep 0.2
./veye_mipi_i2c.sh -w -f wdrmode -p1 0x00 #back light mode off
sleep 0.2
./veye_mipi_i2c.sh -w -f wdrtargetbr -p1 0xff #for wdrmode 0x03
sleep 0.2
./veye_mipi_i2c.sh -w -f denoise -p1 0x00
sleep 0.2
./veye_mipi_i2c.sh -w -f daynightmode -p1 0xfe #Black and White mode
sleep 0.2
./veye_mipi_i2c.sh -w -f new_expmode -p1 1
sleep 0.2
./veye_mipi_i2c.sh -w -f new_mshutter -p1 100000  # in µs
sleep 0.2
./veye_mipi_i2c.sh -w -f new_mgain -p1 45 #0.1 - 0.3
sleep 0.2
./veye_mipi_i2c.sh -w -f brightness -p1 0x00

Does anyone have some advice as to why this could be and how to fix it?

veye_xumm

@jacc said in VEYE-MIPI-IMX385 hot pixels:

new_expmode

I see that you're using the new_expmode mode, indicating that you're employing a fully manual control of exposure time and gain. May I inquire about the values for new_mshutter and new_mgain?
Would it be possible to upload an image from the camera output that you believe has an issue?

JaCC

@veye_xumm
Thank you for answering quickly.
We work with:
new_mgain: 25 ... 45
new_mshutter: 40.000 ... 100.000

Else we use the configuration posted above.

Hot pixels are there in every setting, but more prominent for higher sensitivity.
For ~5 minutes we got rid of most of the hot pixels with brightness set to 0x32, but with the next parameter setting (changing other parameters) they reappeared and afterwards we could not reproduce this behavior.

An actual image will be available tomorrow.

And, is there an internal gamma correction available?

JaCC

Attached is a test shot. This particular shot has been taken at new_mgain=30 and 1 sec exposure, we do see ~30 bright hot pixels already at new_mgain=25 and 40 ms exposure.

Since this is fish eye shot, stars are to be expected only in the brighter circular area in the middle.
blocking the lens, we see the stars vanish and only the hot pixels remain.

veye_xumm

@jacc
I think your gain might be set too high. Can you try reducing the "new_mgain" to below 10 dB and see if that helps?

JaCC

@veye_xumm

At mgain=10 we don't see anything, neither hot pixels nor stars.
We see other people using the cam with mgain = 25 and no hot pixels at all.
And we had one set of parameters, where we didn't have hot pixels at mgain = 25 either.
It only lasted 5 minutes though.