1W should not be necessary and may give you problems with heat.
If you are on version 2.10.0.b2 this is surprising. Could you check the version number, and also if you have an image before running the calibration for the first time. Is that coloured? or dark? or saturated?
Running version 2.10.0b2. I am very sure the 5mm white LED is bright enough and I have illumination len. You can see the iimage as attached supposed to be a brown dried leaf but it is green and fresh lol.
I think from the photo you can tell the LED is bright enough.
Could you click disable flat-field correction and take another image of with the uncorrected camera? You may need to do the auto exposure as well. It is expected to be brighter in some parts and probably pinkish around the edges.
Part of the problem could be that the lens_spacer_picamera_2_pilens.stl is not printed in black. If it is letting light through or allowing light to scatter around inside, it will give an orange background (with orange plastic) on the blank image that you are using for calibration. The calibration is then trying to cancel out the orange so stepping up the green and blue channels. If it is fluorescent orange plastic, that would be a bigger effect.
This is the image of a dried leaf before any calibration. It looks alright to me, But once i start Full auto calibration, everything just become weird. I will try to print a black lens_spacer_picamera_2_pilens.stl and try again tomorrow. The orange is matt and doesnt reflect.
That looks fine. Importantly it is not orange and not too dark. You are removing the leaf before calibration? I am clutching at straws, because that looks exactly fine and should calibrate well.
Yes I always remove the sample before i click on full auto calibration, But trust me, i had done countless time and it always end up with green color.
I will try to print a dark brown and see what happen.
This is puzzling. I am out of ideas now. 
I have been testing for over a week plus and 3d printed many part to retry too. Almost give up but still I am trying. Just finished printing the dark brown lens_spacer_picamera_2_pilens.stl. Will remove and install and try again. I had saw video from others are using orange color filament and doesnt have the problem I am facing. Wondering too.
Still the same, full calibration will turn green. If i ignore the calibration, it looks normal to me.
Very strange indeed. I would go with it uncalibrated for now.
Yes that’s what i am doing now. I am getting a pi 4 later, I wanna try and see if it is a pi 3 issue.
Hi William are you able to show me the before and after calibration screenshot of yours?
I mean the screenshot below, I am notice after calibration, exposure, analogue and digital gain doesnt change at all. For the pre build also doesnt change at all after calibration.
I check the log and saw this error. Anything to do with that?
PiCameraStreamer.camera is deprecated. Replace with PiCameraStreamer.picamera
That warning is just reminding us that done of the code needs updating - I’d be very surprised if it was causing this, though you’re right to mention it. If you change the log level to “info” and run the auto calibrate again it ought to give you more details. In particular I am surprised the gain isn’t changing and the logs should tell us why. When you start the full auto calibrate, does the camera feed go black then get gradually brighter?
It may also be worth trying the steps individually - start with it uncalibrated, then do the auto gain/exposure, then try the lens shading table, then try the white balance.
If you are happy to share your settings JSON file with and without the problem that would be helpful. It’s possible the code is generating an invalid LST and looking at the JSON and logs might provide a clue as to why.
Taking an image with raw Bayer data before and after would also be really useful to debug it.
Thanks,
R
Yes the camera feed did go black and gradually brighter. But when I put a sample to view, it will be greeny. I am very sure all the calibration doesn’t change the value on the app interface at all.
Auto gain/exposure wouldn’t turn the sample greeny only, I will try out what you just mentioned and see how I can gather those information for debugging.
I don’t see a settings named lens shading table and also how do I take an image with raw Bayer data?
I am doing a fresh installation again now.
