Moon from NIR to mid-SWIR

[Andy Perrin]

Don't get too excited. Despite the grand title, the results are a little disappointing. Part of this is the fault of my equipment being insufficient for the job, but a lot of it is apparently the moon's fault. It just doesn't vary that much with wavelength, seemingly!

 

But a negative result is still a result as they say, so here you are...

 

Equipment

-TriWave camera (germanium-on-CMOS sensor), which has sensitivity from 350-1600nm

-Thorlabs 1" 100mm mounted achromatic doublet lens, AR-coated for 1050-1700nm

-Thorlabs SM1 Lever-actuated Iris Diaphragm (for controlling aperture)

-Thorlabs Filter Mount with Sliding Modular Inserts (with a bunch of the inserts for holding my filters. These are very convenient. You slide the filters back and forth for quick swaps.)

-Various Thorlabs SM1 tubes and C-mount adaptors for hooking things together and holding them at the correct distance.

-The filters are a mix of Omega seconds from eBay for NIR, and high quality Thorlabs filters in the SWIR (1200nm+). 

-INOGENI USB 3.0 NTSC video capture card (because the TriWave is analog output)

 

Software

- Custom written MATLAB code for snapping bursts of images and saving them

- Lynkeos astronomy software for aligning images and weeding out low quality ones

- Photoshop

 

Resolution was severely limited by the optics in this case, although the TriWave is only 640x480. I was using only about 128x128px of that, however, due to the 100mm lens. If I get a longer lens, I may try again with higher resolution.

 

Each of these individual images is boiled down from a stack of 600 photos each, but the Lynkeos software throws out many of the bad frames before the rest are averaged together.

 

[Nate]

Nice comparison Andy, I do know the Moon up in the NIR range is much sharper than using just the visible range, but I haven't gone past 850nm due to gear limitations.

I love photographing the Moon, but you can plan all week for an outing, then have perfectly clear skies and the atmosphere turns out to be too turbulent. I think I used best 10% of 2000 once.

[dabateman]

I think its a hair out of focus. 

There might actually be differences if you can nail focus. The three bright spots on the bottom and slight left look different. But I wouldn't know until you nailed focus. 

[Stefano]

The Moon looks the same in the range you have shown, but if you go further the "colors" will invert (this happens when you see mostly emission instead of reflection).

 

Example:

[Andy Perrin]

4 hours ago, dabateman said:

I think its a hair out of focus. 

There might actually be differences if you can nail focus. The three bright spots on the bottom and slight left look different. But I wouldn't know until you nailed focus. 

Except for the 1500nm one, the focus is fine, dabeteman. It's just that you're seeing the average of several hundred 128 pixel wide photos and even with alignment, averaging brings some fuzziness. What you really mean is that you want me to have a higher resolution, and that can only be done by increasing the focal length so I can at least use the full 640 pixels of the TriWave instead of 128!

 

If I get a longer lens, I may repeat this.

[Stefano]

Andy, do you have any longer normal camera lens? They should perform more or less well in SWIR, considering also your low resolution.

[Andy Perrin]

22 minutes ago, Stefano said:

Andy, do you have any longer normal camera lens? They should perform more or less well in SWIR, considering also your low resolution.

Not that I can attach to the TriWave easily or that would perform well. The thing is, this a C-mount camera with a 1/2" sensor. Things have to be at the right distance to focus and then there's also hot spot and focal shift issues that become even more extreme in SWIR. Ordinary lenses actually are not that great in SWIR, with significant loss of light (one stop usually) and major focal shift, never mind the issues of hooking them up to a C-mount camera. One exception is the Velostigmat, that one did ok, but it's not a long lens.

[dabateman]

If the fuzzyness is due to misalignment averaging,  what does a single frame look like after being passed through Topaz Gigapixel AI?

 

[Andy Perrin]

15 minutes ago, dabateman said:

If the fuzzyness is due to misalignment averaging,  what does a single frame look like after being passed through Topaz Gigapixel AI?

 

With averaged image: Terrible! The gigapixel seemed to pick up on the shifted image edges (of the averaged version) and amplified the artifacts instead of the moon. 
 

I forgot to test it on a single frame and unfortunately I already deleted them. Single frame was noisy and distorted by atmospheric turbulence (“seeing” in Astro jargon). 

[Doug A]

Neat experiment. Surprisingly, the moon looks similar in the different wavelengths. I expected more variation.

 

Thanks for sharing,

Doug A

[Andy Perrin]

12 minutes ago, Doug A said:

Neat experiment. Surprisingly, the moon looks similar in the different wavelengths. I expected more variation.

 

Thanks for sharing,

Doug A

Me too, especially since I went out to over twice the wavelength of red light (2*720nm = 1440nm). One thing's for sure: if there were any surface moisture, I would have seen the difference. Guess there's no water on the moon's surface! Of course, underneath is another story.

[Stefano]

Here you can see a UV image of the Moon taken in deep UV (82-184 nm) by the Hopkins Ultraviolet Telescope, although I can't find the image in the linked paper. As you can see, the Moon looks pretty much the same.

 

If you could only see reflected light, I'm pretty sure it would have that look very far into the infrared, past LWIR.

 

Generally I don't expect rocks, concrete etc. to vary much in the optical band. Note that the Moon is darker in UV (they measured its albedo to be about 0.038 compared to 0.136 for visible light (Wikipedia)), simply the "basins" are still darker than the remaining surface.

[Andy Perrin]

That top link seems to indicate there is some variation in NIR/SWIR but it’s false colors which may mean it’s fairly subtle. It might be interesting to try PCA or ICA like I did on the fields (see my old posts) to bring out the small variations. 

[Stefano]

I remember reading your topic about reading an old writing on a wall. So you could try that.

 

The problem I see is that atmospheric "seeing" and other imperfections might be extracted too.

 

The Moon has some faint colors even in visible light, so the same is likely true between 780 and 1500 nm.

[Andy Perrin]

Stefano, seeing has already been averaged out in the above photos, which are averages of several hundred photos each. 
 

The “seeing” distortions occur on a very short timescale of fractions of a second and are random. Highly distorted images (chosen by image entropy score) are removed by the software prior to averaging. 

[Bill De Jager]

Nice work and informative results, Andy.  Sometimes a negative result can be informative!

[dabateman]

Stefano,  that link was very interesting.  Thanks.