Public Lab Research note

Fixed wing over Cocodrie

by cfastie | November 21, 2014 23:10 21 Nov 23:10 | #11381 | #11381

Charles Malveaux brought a very big model airplane to the Public Lab Barnraising last week, and programmed it to fly around and take infrared photos. Charles is a research engineer and graduate student at Louisiana State University studying the use of UAVs to monitor sugar cane and other crops. The UAV Charles flew was a model Piper Cub with a wingspan of about five feet. .

Waiting for Mission Planner to program the plane to fly a predetermined route and take photos.

Charles controlled the plane manually as it taxied and then took off from the ground. Then he transferred control to the onboard computer to fly a pattern and take 12 photographs in a grid. Landing was manually controlled.

Four photos taken by the IR converted Canon S100 onboard the model Piper Cub. Shutter speeds (L to R from upper left) are 1/640, 1/640, 1/1000, and 1/400.
I think the IR block filter in the Canon S100 had been replaced with a Schott BG3 glass filter, or something very similar. Charles purchased the converted camera from Event 38, and that seems to be what they are selling. But the photos above are direct from the camera with no post processing, which is very confusing. These photos resemble false color IR images which require swapping channels after the photos are transferred to a computer. So I am not sure how the images where produced.

I also don't know how the exposure was being determined for each photo. The shutter speed varies among photos from 1/160 to 1/1250 second, with the ISO (800) and aperture (f/2.0) remaining constant. Only a few of the photos are free of motion blur. The sun was setting during the flight and it was very windy, so conditions were not ideal for this kind of photography.

I would be interested to learn whether a CHDK script was running in the camera during the flight, and what the white balance settings were. And more generally, How can these reddish photos be produced without post processing?


I had the same question regarding how Charles' camera produced those images, he mentioned it was the lens but I'm not sure. I'm pretty sure CHDK was running. Whatever the case it was an awesome landing!

Here's a mapknitter view.

Reply to this comment...

This does look like raw imagery from our (Event 38's) NGB modified cameras. We use custom made filter glass to get the best separation possible between infrared and visible channels... I can't be sure but I think this imagery appears a little more vibrant than usual because of the late day lighting conditions with the sun almost at the horizon. Attached is a sample of how they look with more even lighting conditions.


If you're going to fly again, I would suggest trying the latest CHDK and KAP_LUA script which has excellent control over the shutter speed and metering settings. 1/1600th of a second is enough to get rid of motion blur in the majority of conditions. I'm very glad to see the cameras being put to good use, keep up the good work!

Reply to this comment...

Thanks Jeff. I guess a handy feature of using a blue filter like this is that the NIR data is captured in the red channel. So with the right white balance setting, the photo from the camera can resemble false color IR images, as these do. Do you recommend a particular preset or custom white balance setting?

I think you are right that the setting sun punched up these photos some, but the versions above were also passed through Lightroom and I think they got some saturation love. I really like getting the CIR-like photos direct from the camera. If your goal was NDVI, would you recommend a different white balance setting?

Thanks, Chris

Is this a question? Click here to post it to the Questions page.

Reply to this comment...

We typically use the cloudy white balance or AWB. With this generation of filters it's not as sensitive to white balance issues as our previous formulations were, not sure about Public Labs' or other filters.

Ah I see, yes that could certainly add to the poppy look! Cloudy WB works well for NDVI with these filters.

Reply to this comment...

Login to comment.