Flir1 G2 lens experiments

[Bill W]

A 'frankenstein' camera is alive !  This took a F1G2 module and the idea is to be able to fit various alternative lenses that I have to the F1G2.

The lens plate is a prototype routed PCB that carries a connector for the USB from the F1G2, took a couple of goes to get the right hole spacings and pillar heights.  Ground has had to be done by soldered wire - the F1G2 USB connector is only 4 pins and no ground, ground is via a spring finger of all stupidity 

Anyway it is now quite solid and the phone can be remoted down the USB cable.

Still to look at a better battery setup, really down to how well I can connect instead of the awful connector that FLIR used but this serves for now, 

Also the camera tripod mount is upside down 

Bill

[Bill W]

View with original FLIR lens, but focussed correctly.

These are about 44 degrees horizontally, based on fl of 1.2mm

[Bill W]

Now a 4.3mm Ge / ZnS lens at f/1.1

These are about 20 degrees horizontally

[Bill W]

Now the 6.3mm f/1, all Ge

These are about 14 degrees horizontally

[Bill W]

Finally for now the 8.6mm f/1 all Ge triplet

These are about 10 degrees horizontally

[Fraser]

Great work Bill   

A decent lens really improves the imagery coming out of the Lepton. FLIR were clever in using a very inexpensive silicon lens in the LEPTON but I believe that there is good reason why printed silicon lenses are not common in thermal imaging designs.... they are a significant compromise in terms of the image quality that they produce. It is a pity that very small GE or  Chalcogenide IR glass lens assemblies are not commonly available to experiment with. They are fitted to many new miniature cores coming out of China so we may yet see such affordable miniature lenses appearing on the market as accessories etc. With the Lepton there is then the issue of creating the FFC shutter but that is just electro-mechanics and, as we have seen previously on this forum, there are COTS filter switching assemblies made for visible light CCTV board cameras that may be used. These are easily adapted to become an FFC shutter.

[Vipitis]

I think this greatly shows that the Lepton isn't horrible. But having a proper lens that is focussed correctly and not blurring the image really badly and using MSX as a crutch, you get somewhat useable image due to the third party app.

When I started out in 2016 the shitty 80x60 Lepton and it's default lens was all that I had, but the app made a big difference. Sadly I wasn't skilled enough back than to do any computational stuff with the raw data, but it's something I want to revisit. I now have the 160x120 Lepton and the same horrible default lens, so there is a chance I might look into exiftool again.

I originally found out that TPL was doing Lepton based weapon sights (and also firefighting monoculars), but they had 30hz versions and better lenses. They did quote me like 250$ for one of those lenses and that was really excessive. It would also require some serious modifications to my phone back that to get it working. And knowing how I am struggling for like 3 years now to built a custom camera lens mount for my large cores and lenses - it was a good thing to avoid.

I can't find the old thread right now as I should be asleep already, but when Bill sold me the 18mm f/0.7 prototype lens I had a lot of fun to figure out how to convert it into a attachment for the phone and work with the given Lepton lens assembly. I sadly broke one of the three elements, but it was a big success for my modeling, printing and thermal endeavours. Of the many images I took, a Zine has been made which is still not printed and distributed (somewhat pandemic related, but once I figure this out towards the end of the year, I will make a post on this forum and send some out). Main point being, I also got a 4.5mm lens from bill that supposedly covers about a 12mm image circle for the 160x120 sensors that were like 45μm pitch. Given that the old Lepton I have is much smaller area, it should really fit between some of the shots above. And it would be a great project to try. My old phone is just laying about and the Lepton could be recovered and put onto one of those breakout boards. I definitely have the time and really want to do such a project as EE experience and I am also excited for working on the coding side - but I doubt I would take the time, given that I have much larger cores and lenses laying around to do the same project with: stacking for resolution, noise, focus and stiching panoramas.


There is lots of threads from the last 5 years that could be linked here including lens mods other people made.

[Bill W]

I can't find the old thread right now as I should be asleep already, but when Bill sold me the 18mm f/0.7 prototype lens I had a lot of fun to figure out how to convert it into a attachment for the phone and work with the given Lepton lens assembly.

That lens is one of the other hole sets on that PCB, so will be tried out soon.  I have a few other lenses on that 20mm thread format to try as well.
What is good is that most of those lenses were designed for much higher pitch [I mean larger, as in 50, 35 and 25um] sensors than the Lepton but as you say still giving a decent image.

Attached a (very crude) test out attempt at a panorama with the 8.6mm lens. 
A bit of sharpening and a x2 downsample applied.

[Fraser]

Bill,

Did you retain an FFC shutter or are you running the Lepton in dynamic scene software FFC mode ?

[Bill W]

Fraser,

Yes the shutter is fitted for the lens images, helps keep the Lepton in place. 

For the 'stock lens' image the shutter was missing so I could focus it.  Even so the scene is pretty reasonable.

Bill

[Bill W]

OK v2 PCB (the right way up) and started at the silly end for lenses.

Three images taken with a 50mm f/1 lens originally designed for 15µm.  At the bottom the original FlirOne image with the 50mm image sizes rougly added.  With this field of view it is difficult to aim especially if there is little thermal contrast about as with these (3pm, a cloudy, still November day).
The images are straight off the Joe app, so upsampled x4 but nothing extra.

Will do a few more when there's some more interesting targets.

Unfortunately several of my other lenses for that mount do not reach infinity focus, once mounted up the detector is just too far back.

These are about 1.8 degrees horizontally

[Bill W]

Now 19mm f/1.3 lens originally designed for 25µm pitch

Feels like the smaller aperture is helping with resolution ?

These are about 4 degrees horizontally

[Bill W]

Now 35mm f/1 lens originally designed for 15µm pitch

These are about 2.4 degrees horizontally

[Bill W]

A 65mm lens, f/1.3.  This is a horizontal field of just 1.4 degrees, a right pain to aim !  I failed to pick up any aircraft with it.

Of interest this is getting some detail in the moon which is 43 pixels across on the sensor.  The image is an inverse of a visual one, so the 'dark' lunar maria are warmer which is no real surprise.

[Vipitis]

The more you post the stronger the nostalgia gets.

My current phone doesn't support the app and I really hate the myFLIR app. But it has the Lepton 3.5

Getting an image of the moon like this was a goal of mine for a very long time. I wonder if I should rip the Lepton 2.5 Out of my old phone and put it on a board. I got 4.5, 32mm and 150mm lenses to try it with as well as some remaining elements from the 150mm and ZnSe stuff.

[Bill W]

Photo of the current 'Frankencamera' build with the 35mm lens fitted

Bill

[Fraser]

I have many FLIR One G2 cameras that are sat in disgrace as they have the common ‘no boot’ failure.These lens experiments are stimulating the experimenter within me. I may just have to repair a camera or two in order to try some lens experiments of my own. It would be great if there was easy availability of some conventional design Chalcogenide miniature lenses that could be adapter to use on the Lepton core. I have approached Seek Thermal on the matter but they refuse to consider supplying any lenses as accessories, even for their own camera cores. They state that the camera has to be calibrated for each lens that they use and changing FOV lens would result in a poor image. The MOSAIC cores have some decent lenses fitted. I may be in the market for some small Argus lenses from you Bill 

Thanks for these very interesting lens tests    The Lepton 3 provides better imaging than I had expected.

Fraser

[Bill W]

Having gone to the lengths of designing that PCB, I would be happy to either share the data or get some made if the forum wanted enough of them. 
In doing that I could also replace or add some other lens fixings on request.  As it is now it has several Argus lens fixings, but some are fairly exotic like the P4430 Wreathall double screw fixing - doubt anyone else has one of them spare.

I have a few spare lenses to go with them, 15mm Argus 3 types and various 4.3 / 6.3mm (20mm threaded).  The 20mm thread mounts are the issue.


Bill

[Fraser]

I have a thought for another use of Bills FLIR One G2 experimentation PCB. I have some lenses for experimentation but no specifications to assist in their use. I have often wanted to create an optical test bench to determine the back focus distance, lens quality and field of view. The Lepton uses a truly tiny microbolometer and 12um pixels so the lepton FPA is actually very useful for testing lens assemblies as under illumination is unlikely ! As the active component of a thermal imaging optical bench, the adapted FLIR One G2 could be a very neat solution. There would also be the option to just heavily modify the FLIR One G2 standard casing but the casing may increase the minimum back focus distance obtainable. Hmmm, another project on my to-do list.

I attach a picture of a standard optical test bench for lens element experimentation. A thermal imaging version needs a thermal camera FPA in place of the white card screen and a hot resistor in place of the lamp house emitter.

[Bill W]

There would also be the option to just heavily modify the FLIR One G2 standard casing but the casing may increase the minimum back focus distance obtainable. Hmmm, another project on my to-do list.

The casing probably has to go.  The actual sensor plane on the Lepton is quite a way back.

In 'Frankencamera' the Lepton shutter unit is hard up against the 1.6mm PCB  The lens mounts are generally pretty close to, or simply on, the other side.  Even then some lenses will not focus at infinity and the Argus designs generally had a fairly large rear focus at around 8mm.

I should be able to work out where the Lepton focal plane sits with a suitable lens in place as I would know the real focal plane of the lens.

Bill

[newex]

https://youtu.be/j_8XbGDZz9Y

This is video of using a 25mm lens with a Flir Lepton 3. Distance 30-35m.

[Fraser]

Newex,

How did you mount your 25mm lens on your Lepton please ?

The Lepton design is a bit of a pain due to the placement of the FFC shutter assembly.

Fraser

[newex]

This is just an experiment at this time.
I had to design a board to easily and accurately mount the lens on the optical axis of the core.
This is a Pure Thermal 2 clone.

And a print case. It looks so like this.
And remove the original Lepton lens, of course.

[Fraser]

Nice work 

Thank you

Fraser

[Fraser]

I feel a FLIR One G2 disassembly coming on, followed by adaption to use a better lens