First Cambridge Church, Harvard Square LWIR

[Andy Perrin]

First Cambridge Church, Harvard Square, in Massachusetts, US. The image is a LWIR panorama. 

 

Technical details:

69 image panorama, each image was 320x240 pixels

Camera: FLIR E60

Lens: FLIR 15 degree AFOV germanium lens

 

The images were stitched together, ragged edges were filled in with Photoshop Content Aware Fill (so, that means some tree branches are not real), then details were brought out using the Aurora 2019 HDR software, and sharpened using Topaz Sharpen. 

 

 

[colinbm]

Very cool.

[lukaszgryglicki]

WOW - impressive.

 

[Doug A]

Fabulous image! I've never seen a FLIR image anywhere near this detailed and sharp. 69 shot pano is a lot of time & work. In this case it was worth it.

Thanks for sharing,

Doug A

[Andy Perrin]

It wasn’t too bad - I took all 69 photos in ten minutes that I was waiting for the bus, and they stitched very quickly because of the low resolution of each individual image. 

[Nate]

Nice Andy! Curious what the structure is made out of. I don't see any supporting beams showing through from temp difference.

[Andy Perrin]

1 hour ago, Nate said:

Nice Andy! Curious what the structure is made out of. I don't see any supporting beams showing through from temp difference.

What do you mean? I can see them clearly? Here's some of them marked for you:

[Kai]

A really astonishing result. Great work!!!

[Nate]

9 hours ago, Andy Perrin said:

What do you mean?

You're right I see them now. I was thinking the beams would be larger. 

[bobfriedman]

nice work.. a pano success.   does that cam cover the full 8-14 um band?

[Andy Perrin]

Bob - yes to 8-14 micron band roughly. Since it's a micro-bolometer-based sensor, which works by detecting the temperature change of a suspended bit of semiconductor, it technically will react to the entire spectrum up to UV even, but there is a germanium lens on the front which restricts the spectral range to begin at 2 microns and end at 16 microns. The temperature of the surroundings means that at 20deg C, 80% of the available light tends to be in the 7.5-30 micron band. Put the two together and you have the effective range of the camera, give or take a little, as 7.5-16 microns. Different manufacturers will list slightly different endpoints, and I'm not sure if there is any additional filtration, but those are the dominant considerations.

--

Now, speaking of supporting beams, let me point out that there is an entire "ghost" window if you look carefully. I have pushed the contrast a lot here with the Photoshop curves tool to make it more obvious, including dodge/burning it:

[Stefano]

1 hour ago, Andy Perrin said:

Bob - yes to 8-14 micron band roughly. Since it's a micro-bolometer-based sensor, which works by detecting the temperature change of a suspended bit of semiconductor, it technically will react to the entire spectrum up to UV even, but there is a germanium lens on the front which restricts the spectral range to begin at 2 microns and end at 16 microns. The temperature of the surroundings means that at 20deg C, 80% of the available light tends to be in the 7.5-30 micron band. Put the two together and you have the effective range of the camera, give or take a little, as 7.5-16 microns. Different manufacturers will list slightly different endpoints, and I'm not sure if there is any additional filtration, but those are the dominant considerations.

This is something I have been thinking about for a while. Microbolometers can be heated by light of essentially all wavelengths (at least in the optical region), so in theory it should be possibile to make a SWIR camera or even a visible light camera by just replacing the lens and/or using filters.

 

Microbolometer-based SWIR and MWIR cameras do exist: https://www.polytec.com/eu/optical-systems/products/special-cameras/ir-cameras-arrays/swir-microbolometer-cameras

 

Then of course microbolometers have their limitations, but in principle you can make them respond to any wavelength you wish, I think.

[lonesome_dave]

This is really nice Andy.

Inspires me to try something similar with the low-res imagers I have quite a few of.

I'll need to investigate all that post-processing you did to work this magic.

 

[Andy Perrin]

50 minutes ago, lonesome_dave said:

This is really nice Andy.

Inspires me to try something similar with the low-res imagers I have quite a few of.

I'll need to investigate all that post-processing you did to work this magic.

 

Well the most important thing to know is that the FLIR “radiometric JPEG” files in reality contain the raw data in 16bit, and this can be extracted using the Exiftool program.

[lukaszgryglicki]

I even have a program for this - it allows creating images/movies from CSQ files - I've written it myself: https://github.com/lukaszgryglicki/csqconv

[Andy Perrin]

That is not quite the same - I am referring to the radiometric JPEGs, which are .jpg files. 

[lukaszgryglicki]

OK, right, my program extracts image data from videos saved by camera and saves them as JPG.

 

[Andy Perrin]

Yes, I was referring to STILL images, not videos. 

[Andrea B.]

Well done, Andy!

 

Interesting subject which brings out the underlying structure. And impressive processing to get 69 separate images to behave themselves when concatenated.

 

I wonder why that window was covered over?