So where can I get a 640x480 LWIR camera for under $1000?

[Ben321]

I know that these are usually SUPER expensive (about $20000 for FLIR T620, and even some of their cheaper 640x480 cores/modules are still like around $5000), but I'm wondering what is really driving up the cost. Would it be possible for a company to manufacture a 640x480 LWIR camera for a very low price that would let them then sell it for under $1000? I can't imagine that the manufacturing process for a 640x480 vanadium oxide microbolometer focal plain array, actually costs the company over $1000 per chip. I also can't imagine the manufacturing costs for making a small (1cm diameter) lens out of germanium, would be over $1000 per lens. It feels like FLIR and all other thermal imager manufacturers are just inflating costs well beyond what they need to to make a profit, in order to make an INSANELY HUGE profit. I'm guessing their profit margin is something like 5000% to 10000% based on the prices they are selling these things at, compared to what I believe their manufacturing costs are.

So back to this thread's title question. Is there any company out there, who's set out to make the most dirt-cheap 640x480 thermal imager possible, with the intent of providing the public with a with a decent resolution thermal imager, at a reasonable price? Is there any company who's managed to get the price of a 640x480 LWIR camera below $1000? If so, I'd love to know what that company is, so I can check out their website.

[Chanc3]

You can get really lucky in the used thermal imaging camera market. We recently picked a 640x480 camera for a fraction over £1000.

I'm waiting for a FLIR A15 to arrive as if it's anything like the Ex range, you maybe able to hack it to 640x480. (from 160x120)

Sent from my ONE A2003 using Tapatalk

[Kilrah]

I know that these are usually SUPER expensive (about $20000 for FLIR T620, and even some of their cheaper 640x480 cores/modules are still like around $5000)

As was already pointed out to you but you seem to keep ignoring, you can find offerings for around $3500.

Would it be possible for a company to manufacture a 640x480 LWIR camera for a very low price that would let them then sell it for under $1000?

Probably, but now put yourself in the manufacturer's seat - why would you do it? Your typical customers are highly trained, have "tactical" needs and will happily shell out anything you ask for. No point loweing the entry point, making less money for more work, and ending having to support the average Joe who has no clue what he's doing.

Nobody will move from their comfortable seats unless a new player manages to catch up and forces them to. Hasn't happened in that segment yet.

[Ben321]

You can get really lucky in the used thermal imaging camera market. We recently picked a 640x480 camera for a fraction over £1000.

I'm waiting for a FLIR A15 to arrive as if it's anything like the Ex range, you maybe able to hack it to 640x480. (from 160x120)

Sent from my ONE A2003 using Tapatalk

1GBP > 1USD. Last time I checked, 1GBP ~= 1.5USD. So 1000GBP is obviously well over the price of $1000 I was looking for.

You guys here on this forum must be super rich hobbyists. For me, $1000 is a MAJOR PURCHASE. I wouldn't ever consider spending > $1000 on a thermal imager. Maybe an oscilloscope. Maybe a programmable function generator. Maybe a new laptop. But certainly not on a camera who's resolution is 640x480 (I don't care how exotic of a wavelength it's sensing). You can get a webcam with a silicon CMOS visible light sensor array, which has a resolution of 640x480, for about $50.

In general, technology gets cheaper. Some of the oldest consumer level digital cameras (when they came out in the early 1990s) did cost about $1000, and did have a resolution of about 640x480, or maybe a 800x600 if you were lucky. However as time went on, prices have come down. Now days, a simple point-and-shoot digital camera, has 640x480 is a MINIMUM resolution (selectable in a settings menu), and usually has like 20megapixels as its default resolution, and the price is usually like only $200. So as you see, the longer that a technology exists on the market, the more prices tend to come down. That is a general trend in the consumer electronics industry.

However, that trend seems to have ZERO effect on the prices of thermal imagers. 640x480 thermal imagers cost $20000 about 15 years ago, and they still tend to cost that much now (T620 for example). Some have slightly dropped their price, such as the $3500 one you were talking about. But that's still not a significant improvement. That only represents a price per pixel that has gone from 6.5cents per pixel to 1.2cents per pixel. This is an 82% decrease in price. Compare that though to what has happened with consumer digital cameras. It's gone from about 0.3cents per pixel to 0.001cents per pixel. That's a  99.7% decrease in price, which is a FAR GREATER decrease in price, than the 82% decrease in price that FLIR is offering for LWIR cameras. And that assumes it's even a decrease at all. More likely it's just that the $3500 thermal imager you were speaking of, simply has fewer features than the T620. The T620 cost about $20000 several years ago, and it still does.

Yes companies deserve to make a profit, but should it really be such an obscene profit as this? I'm guessing that it costs FLIR no more than maybe $100 to $200 to manufacture a single T620 unit. Should they really then be selling it for $20000? They could still make a profit if they sold it for $500. And by the way, this would actually bring in MORE CUSTOMERS. I predict, that this surge in customers would not only offset the decrease in profit-per-unit, but over time they would actually make MORE MONEY this way. You see, not only would people in military or industry be buying them, the average person would ALSO be buying them.

[Fraser]

The cold but realistic response to your question is why should a 640 X 480 pixel thermal camera be made available to you for less than $1000. 320 X 240 is still the standard for most applications. There are people with very deep pockets willing to pay significant sums of money for higher resolution cameras. Until that situation changes, there is little incentive to supply 640 X 480 resolution cameras at anything but premium prices.

Also be aware that 640 X 480 resolution cameras do not always have 640 X 480 true pixels. As you may be aware, TESTO used a feature called Super Resolution to create 640 X 480 pixel images from a 320 X 240 pixel resolution detector array. They used the unavoidable movement when hand holding the camera to create a stacked image and tidied it all up in the image processing stage. When mounted on a tripod, Super Resolution does not work.

I have to ask the question.... Why do you need 640 X 480 pixel resolution ?
There are applications for such in areas like the military, science and broadcast programs but 320 X 240 is sufficient for most applications. This is not visible light imaging. Thermal, sources are by nature, less well defined due to a combination of conducted, radiated and convection heat transfer within the DUT. Sharp Delta T transitions usually only occur when a thermal insulator is in play or a DUT is contrasting against a background of differing temperature.

In the early days of thermal imaging the resolution could be quite low and this could make life difficult for the user as they found it hard to spot references against which to interpret the thermal image of an area.  On some cameras Visible light camera image merging was used to provide a reasonable context for the thermal image detail. As camera resolution improved, it was possible to see enough detail on the surface of the DUT to establish the location of heat sources and Delta T across surfaces. 320 X 240 pixels are adequate in most circumstances for such image interpretation.

Visible light Image merging was of some benefit but the visible light image could cause confusion with regard to the thermal information in the combined image. FLIR decided to use edge detection techniques to capture edge detail only from the visible light image and overlay only that detail onto the thermal image. The edge detail provides the user with very useful context within an image without producing contrasts that could be confused with thermal data. The MSX system works well in good light but can add noise to the images in low light conditions as the visible light camera AGC increases the gain and produces noise.

Higher thermal resolutions are used in military weapons aiming systems that needed excellent target acquisition and aiming capability at long ranges without a very small FOV. The minimum detectable target size at long range is an issue for defensive weapons systems. Some very high resolution cooled thermal camera systems are deployed on ships etc, but they cost a small fortune and are tightly controlled technology.

Do remember that 640 X 480 pixels just allows you to be twice the distance from the DUT using the same FOV and obtain a similar level of detail in the image. Conversely, if you can obtain the required image at half the original distance you get similar thermal detail of the DUT as when using a 640 X 480 camera with the same FOV at the original distance. I hope that makes sense.

I own X2 and X3 Auxilliary telescope lenses for some of my thermal cameras. These may be fitted with Close Up lenses of 6" and 12" that I also own. These allow me to effectively use my 320 X 240 pixels very close to the DUT and at X2 or X3 magnification. The FOV is effectively reduced by a half or two thirds providing greater detail of a smaller area of the DUT.

If you tell me why you need a resolution higher than the industry nominal of 320 X 240 pixels, I may be able to suggest options that are open to you.

If it just that you want more bang for your buck........ Well I cannot help you there, but can say that you have never had it so good on the thermal imaging front and maybe should just be grateful for what is already available for less than $1000

Fraser

[Ben321]

The cold but realistic response to your question is why should a 640 X 480 pixel thermal camera be made available to you for less than $1000. 320 X 240 is still the standard for most applications. There are people with very deep pockets willing to pay significant sums of money for higher resolution cameras. Until that situation changes, there is little incentive to supply 640 X 480 resolution cameras at anything but premium prices.

Also be aware that 640 X 480 resolution cameras do not always have 640 X 480 true pixels. As you may be aware, TESTO used a feature called Super Resolution to create 640 X 480 pixel images from a 320 X 240 pixel resolution detector array. They used the unavoidable movement when hand holding the camera to create a stacked image and tidied it all up in the image processing stage. When mounted on a tripod, Super Resolution does not work.

I have to ask the question.... Why do you need 640 X 480 pixel resolution ?
There are applications for such in areas like the military, science and broadcast programs but 320 X 240 is sufficient for most applications. This is not visible light imaging. Thermal, sources are by nature, less well defined due to a combination of conducted, radiated and convection heat transfer within the DUT. Sharp Delta T transitions usually only occur when a thermal insulator is in play or a DUT is contrasting against a background of differing temperature.

In the early days of thermal imaging the resolution could be quite low and this could make life difficult for the user as they found it hard to spot references against which to interpret the thermal image of an area.  On some cameras Visible light camera image merging was used to provide a reasonable context for the thermal image detail. As camera resolution improved, it was possible to see enough detail on the surface of the DUT to establish the location of heat sources and Delta T across surfaces. 320 X 240 pixels are adequate in most circumstances for such image interpretation.

Visible light Image merging was of some benefit but the visible light image could cause confusion with regard to the thermal information in the combined image. FLIR decided to use edge detection techniques to capture edge detail only from the visible light image and overlay only that detail onto the thermal image. The edge detail provides the user with very useful context within an image without producing contrasts that could be confused with thermal data. The MSX system works well in good light but can add noise to the images in low light conditions as the visible light camera AGC increases the gain and produces noise.

Higher thermal resolutions are used in military weapons aiming systems that needed excellent target acquisition and aiming capability at long ranges without a very small FOV. The minimum detectable target size at long range is an issue for defensive weapons systems. Some very high resolution cooled thermal camera systems are deployed on ships etc, but they cost a small fortune and are tightly controlled technology.

Do remember that 640 X 480 pixels just allows you to be twice the distance from the DUT using the same FOV and obtain a similar level of detail in the image. Conversely, if you can obtain the required image at half the original distance you get similar thermal detail of the DUT as when using a 640 X 480 camera with the same FOV at the original distance. I hope that makes sense.

I own X2 and X3 Auxilliary telescope lenses for some of my thermal cameras. These may be fitted with Close Up lenses of 6" and 12" that I also own. These allow me to effectively use my 320 X 240 pixels very close to the DUT and at X2 or X3 magnification. The FOV is effectively reduced by a half or two thirds providing greater detail of a smaller area of the DUT.

If you tell me why you need a resolution higher than the industry nominal of 320 X 240 pixels, I may be able to suggest options that are open to you.

If it just that you want more bang for your buck........ Well I cannot help you there, but can say that you have never had it so good on the thermal imaging front and maybe should just be grateful for what is already available for less than $1000

Fraser

I'm not so much interested even in temperature measurement, as I am being able to simply produce an image that is in the LWIR wavelength range. I don't care if it's not calibrated for temperature measurement (using blackbody curves, and fancy equations to turn pixel values into temperature readings). I just want to be able to get an image in the LWIR part of the spectrum, and the bigger the better (640x480 being the SMALLEST that I would consider to be a "reasonable sized image"). Bigger, simply to cover a larger field of view, not to improve resolving fine details on objects. 160x120 (like the FLIR One does) is not exactly something that you'd submit to a web photo gallery like Flickr, or something that you'd use for your Windows desktop wallpaper. If it can resolve details as well as a 160x120 or 320x240 thermal imager, and only used the increased 640x480 resolution to increase the total field of view (not to resolve finer details), that would be all I need.

Photographers use visible light cameras to create works of art, but wavelengths outside the visible spectrum can also be used to create art. In fact, I'd argue that would be more artistic, because such wavelengths often produce unusual or exotic effects when visualized as an image. So I think that LWIR cameras can produce even more artistic images than visible light cameras. Yes, there's NIR (near infrared) photography, but that produces images that are only slightly different from visible light photography. The strange images that are produced by LWIR cameras have much more artistic value than simple NIR images, in my opinion. LWIR images make much better computer wallpaper, and make much better candidates for upload to Flickr, or at least they do if they aren't INSANELY SMALL images, like those taken by the FLIR One.


And it's not just art. The mere ability to see wavelenghts outside the visible spectrum is a very awesome experience. But it's not nearly as awesome an experience when the image is so tiny. It's like expanding the human senses beyond those that we were born with, at least if the image is of a reasonable size (640x480 or better).

I'm only looking for a 640x480 imager, because it's the only reasonable size (640x480 or larger) imager that I think I MIGHT be able to get for a reasonable (less than $1000) price. It's not because it's near the top of the imager size I'd like to get my hands on. In fact, 640x480 is the MINIMUM size that I'd consider buying.

Yes, I bought a FLIR One, but only because before that I never even had a thermal imager, and a FLIR One was better than nothing. I hope to be able to upgrade soon to a 640x480 thermal imager. 320x240 is just too small of an image to be able to be considered a "reasonable sized image". So I would never consider upgrading to anything less than a 640x480 thermal imager. My application is not a demanding application. It doesn't require that the factory put in $10000 worth of quality control and calibration. All it needs to do is be able to save an image to an SD card at the click of a button. A simple one-button device, with a battery-charger port, and an SD card slot, is all I need. It doesn't have to be anything fancy. All it needs to do is to be able to take a 640x480 (or better) picture using LWIR radiation. It doesn't matter to me if it has a few dead pixels or hot pixels. I wouldn't be using it for demanding applications (like scientific, industrial, fire department, or military applications). It doesn't matter to me if it's made with the cheapest parts available, from some super-cheap generic brand that you've never heard of before, from a company located some place in China. I just need the price to be less than $1000, and the image size to be at least 640x480.

[Bill W]

Lets cross check a few of those numbers.

The CMOS webcam 640x480 has 3um pixels.
A thermal 640x480 is at least 12um pixels, more likely 17um pixels.
Therefore you get 16 to 25 times the number of sensors from the same amount of CMOS processing.
The thermal sensor needs to be in a vacuum, an awkward costly process that has not got cheaper since 2002.
Thermal sensors need multi-step 3D processes, not simple 2D like a visual sensor.
VOx is a very unpleasant process, so you do not get the economies of scale of using a big fabrication facility.

So a $5 visual chip can be multiplied by 16, then by 4 then by 3 then by 2.  That makes $1920 before you even think about lenses and calibration time.  The visual camera might have the same $20 of electronics content which is barely relevant

Bill.

[encryptededdy]

Even if we ignore manufacturing costs, I doubt it makes any business sense to sell a 640x480 camera for <$1000 because I doubt there's major demand for cameras of that resolution - the majority of those who do demand cameras of that resolution can already afford to pay the current price.

with the intent of providing the public with a decent resolution thermal imager, at a reasonable price

Here's the problem - while us, as enthusiasts, would love to have a 640x480 LWIR camera for <$1000, the public does not - there's no demand for it. Unfortunately for the general public thermal cameras have a limited use case and they are not willing to pay even $500 for one. Heck, when you visit posts about Seek Thermal or FLIR One on tech news sites, there's still plenty of people complaining that $250 is too expensive for any thermal camera whereas we are very happy to see a 160x120 camera for about a 1/4 of the price we would have paid 2-3 years ago.

[encryptededdy]

If so, I'd love to know what that company is, so I can check out their website

Be assured that if such a thing ever appears, it will be the top of the Thermal Imaging section of this forum in a matter of seconds.

[Redshift1340]

The quality and the clarity of the thermal image being captured isn't only, or even mostly, to do with the array size.  Comparing the resolution of a visual light camera with a thermal camera is an apples to oranges comparison.  You can get a hell of a lot of detail out of a 320x240 array, even 160x120. For example, compare a FLIR one G2 image (160x120) with a FLIR E40 image (160x120).  Don't think that just because your FLIR one is 160x120 and produces a blur fest, that it means you have to find a thermal camera that's near the top of the resolution range (640x480). I have a FLIR one G2 as well, and while I'd love to have a 640x480 array, I know I can get ten times the clarity of the FLIR one even staying at the same resolution with an E30, E40, or a slight bump up like the Scout PS24(240x180), etc.  Look around YouTube and the web for sample vids on some of those, you might be surprised at the quality.

[Ben321]

Lets cross check a few of those numbers.

The CMOS webcam 640x480 has 3um pixels.
A thermal 640x480 is at least 12um pixels, more likely 17um pixels.
Therefore you get 16 to 25 times the number of sensors from the same amount of CMOS processing.
The thermal sensor needs to be in a vacuum, an awkward costly process that has not got cheaper since 2002.
Thermal sensors need multi-step 3D processes, not simple 2D like a visual sensor.
VOx is a very unpleasant process, so you do not get the economies of scale of using a big fabrication facility.

So a $5 visual chip can be multiplied by 16, then by 4 then by 3 then by 2.  That makes $1920 before you even think about lenses and calibration time.  The visual camera might have the same $20 of electronics content which is barely relevant

Bill.

Did you read my intent on how I would have used it?

I remember reading about a year ago, that there's a new (and much cheaper) process for making LWIR microbolometers that was recently invented. I can't find the article now, but I remember thinking at the time "Will this usher in a new era of LWIR camera technology, where multi-megapixel LWIR cameras can be sold to the general public at the same price as visible light point-and-shoot cameras?". Unfortunately, even if it has the ability to do so, I don't see any manufacturing companies taking advantage of the new manufacturing process, and cranking out dirt cheap high resolution LWIR cameras to be sold to the general public. FLIR isn't, nor is any other company taking advantage of this new manufacturing process I read about. Though I can't find the specific article now, I did find this one website for a company called ULIS, that makes LWIR microbolometers focal plane arrays out of silicon instead of VOx. I'm not sure if that's the new manufacturing technique I read about before, but it definitely seems to be different from the standard VOx design. And since silicon is a much more readily available material, I can see the possibility of it driving down prices significantly.

Here's a link to a product page of this company, specifically for a product called the Pico640P.
http://www.ulis-ir.com/index.php?infrared-detector=17--m-640x480
Its a 640x480 "amorphous silicon" microbolometer focal plane array chip.
It has a wavelength range is 8um to 14um. It is protected from sun damage and overexposure.
Unfortunately their link to the PDF datasheet http://www.ulis-ir.com/uploads/Products/ulis-pico640p-ul04272.pdf is more like a link to the product brochure, and really provides no more info than the product page for this chip. It doesn't have any pinouts or electrical specifications. I also can't find the price of this chip anywhere.
Fortunately they also have a button to contact the company for more info.
I have contacted them with this, and have asked for the unit-price for this chip. I'll post back what I find out about it, if they bother to even reply (since I'm an individual, not a business, I think I'll probably be lucky if I even get a response).

[Redshift1340]

Though I can't find the specific article now, I did find this one website for a company called ULIS, that makes LWIR microbolometers focal plane arrays out of silicon instead of VOx. I'm not sure if that's the new manufacturing technique I read about before, but it definitely seems to be different from the standard VOx design. And since silicon is a much more readily available material, I can see the possibility of it driving down prices significantly.

Here's a link to a product page of this company, specifically for a product called the Pico640P.
http://www.ulis-ir.com/index.php?infrared-detector=17--m-640x480
Its a 640x480 "amorphous silicon" microbolometer focal plane array chip.
It has a wavelength range is 8um to 14um. It is protected from sun damage and overexposure.
Unfortunately their link to the PDF datasheet http://www.ulis-ir.com/uploads/Products/ulis-pico640p-ul04272.pdf is more like a link to the product brochure, and really provides no more info than the product page for this chip. It doesn't have any pinouts or electrical specifications. I also can't find the price of this chip anywhere.
Fortunately they also have a button to contact the company for more info.
I have contacted them with this, and have asked for the unit-price for this chip. I'll post back what I find out about it, if they bother to even reply (since I'm an individual, not a business, I think I'll probably be lucky if I even get a response).

I believe the $960 Opgal Therm-App uses a Pico amorphous silicon microbolometer at 384x288 pixels, so it seems they are already being used and have made the prices more competitive.  Maybe they'll drive down the prices even more in the coming months or years.  Let's hope so

[Ben321]

I believe the $960 Opgal Therm-App uses a Pico amorphous silicon microbolometer at 384x288 pixels, so it seems they are already being used and have made the prices more competitive.  Maybe they'll drive down the prices even more in the coming months or years.  Let's hope so

Do you know of any device currently using the Pico640 (640x480) or the Pico1024 (1024x768) sensor chips?

[Redshift1340]

I believe the $960 Opgal Therm-App uses a Pico amorphous silicon microbolometer at 384x288 pixels, so it seems they are already being used and have made the prices more competitive.  Maybe they'll drive down the prices even more in the coming months or years.  Let's hope so

Do you know of any device currently using the Pico640 (640x480) or the Pico1024 (1024x768) sensor chips?

Not specifically, but a company called Thermoteknix has decided to use the Pico640 gen2

http://www.vision-systems.com/articles/2010/04/ulis-pico640-lwir-detector-selected-for-portable-thermal-imaging-equipment.html

[Redshift1340]

Actually I've found a model Thermoteknix is using the Pico640 in, it's the Thermoteknix MicroCAM 3.  I'm sure there's others, or will be soon. I'll keep an eye out.

Here's an excerpt from an article about the cost savings of the Pico gen2's new manufacturing process..

http://www.militarysystems-tech.com/suppliers/thermal-vision-specialists/ulis

"Cost-saving measures
ULIS has pioneered technologies to enable customers to have price-competitive end-products. Early on, ULIS switched from metallic to ceramic IR sensor packaging that brought significant cost-savings. In conjunction with the CEA-Leti, ULIS has developed PLP (Pixel Level Packaging), a new wafer-based vacuum packaging. With this disruptive technology, ULIS can manufacture its pixels using the same semiconductor process step as the packaging. This makes the whole production cycle more efficient. The new packaging technique is expected to bring an additional 30% in cost-savings for end-customers.

All of these advances progressively make ULIS’ IR products more affordable and accessible to a wide-range of industry applications and attractive to procurers of military equipment increasingly turning to products that have their roots in the commercial world."
For more information, visit: http://www.ulis-ir.com

[Kilrah]

However, that trend seems to have ZERO effect on the prices of thermal imagers. 640x480 thermal imagers cost $20000 about 15 years ago, and they still tend to cost that much now (T620 for example). Some have slightly dropped their price, such as the $3500 one you were talking about. But that's still not a significant improvement. That only represents a price per pixel that has gone from 6.5cents per pixel to 1.2cents per pixel. This is an 82% decrease in price. Compare that though to what has happened with consumer digital cameras. It's gone from about 0.3cents per pixel to 0.001cents per pixel. That's a  99.7% decrease in price, which is a FAR GREATER decrease in price, than the 82% decrease in price that FLIR is offering for LWIR cameras.

Is it that hard to understand that there are hundreds of millions of visible light sensors that have been sold each year for more than a decade now with the related cash flow and incentive to reduce cost, versus maybe a few thousands for high res thermal? 
When visible sensors were only used in super pro DSLRs costing $10-20k and thus sold in comparable quantities their prices weren't going down either.

The big difference is everybody can make use of a visible light camera which caused the volume to increase, while "nobody" comparatively cares about thermal so it will forever remain a niche product, even if it gets cheaper and becomes "widespread" it will still be 3-4 orders of magnitude less than visible light sensors.

You guys here on this forum must be super rich hobbyists. For me, $1000 is a MAJOR PURCHASE. I wouldn't ever consider spending > $1000 on a thermal imager.

No, they're hobbyists who understand they have no need for a 640x480 thermal cam just to look around for fun. And by definition if you need one then it's not a hobby anymore, and whatever you can make with it certainly can cover the cost.

[Fraser]

Ben321'

Thank you for the explanation.

I can totally understand your desire to capture the artistic aspects of thermal imaging. I have a thread on here containing many X-Ray images that I have produced. They contain 2048 X 2048 pixels and people have already suggested that I could produce art with such high resolution X-Ray equipment.

https://www.eevblog.com/forum/projects/the-x-ray-image-thread-by-aurora-various-electronics-via-x-ray-imaging/msg748531/#msg748531

The problem you have is the specialist nature of thermal imaging. There may be a few Artists working in the thermal image arena, but not many. As you suggest, the currently available resolutions do limit the size of the pictures that may be produced. The market does not exist for 640 X 480 imagers that have no calibration of refinement. That would have to be a DIY project.

I have had some thoughts on the matter though. Increased FOV requires wide angle lenses. The FLIR E4 and F1G2 already have wide angle lenses in thermal camera terms. Most of my FLIR professional cameras have an FOV of only 24 Degrees ! You can add an Auxiliary wide angle lens but they are very expensive.

Have you considered panoramic images created with image stitching software? There has been excellent work on this by members of this forum in the E4 threads.

With regard to the image size issue, have you considered interpolation to upscale the image? IRISYS used clever interpolation to upscale a 15 x 15 pixel image to 128 X 128 pixel output ! The result was far better than I expected.

The above suggestions will be far cheaper than any 640 X 480 thermal camera that you will find any time soon.

Art is about experimentation and making the best of what you have to hand. Artists are often financially challenged until they are 'discovered'. They need to be inventive as well as creative..... That is part of the Art. Please do look into computer programs that can provide image stitching and good quality interpolation.

Fraser

[Fraser]

Further to my last. I have been considering available options. I am not clear whether you are a student of ART or a hobbyist like me with an interest in the thermal domain. If you are a student some colleges own decent quality thermal cameras in the technology and engineering departments. If you could borrow such for your art related projects it would save you buying a higher resolution camera. Any 320x240 pixel modern camera will outperform the F1G2 and you can assess whether you really need the 640 X 480 resolution. A simple X2 interpolation of a 320x240 image is very effective and would be a definite option and you could go for higher interpolation multipliers.

Another alternative is to rent a suitable camera. It is expensive but not as expensive as buying one ! If you talk nicely to the thermal camera rental companies and explain your need, your student status (if applicable) you may get a good deal on a weeks rental ? People can be very kind when they know someone has limited funds and an interesting project to complete. I have helped those on limited budgets in the past. It's all good Karma in my view.

If you feel you must own a 640 X 480 resolution camera, please beg or borrow a DECENT 320 X 240 camera first and try interpolation. You should not judge camera resolution performance by the images produced by the F1G2. If you cannot get your hands on such a camera, just download some 320 X 240 thermal images from the Internet and apply interpolation to them and see if the quality is acceptable.

Fraser

[Bill W]

I did find this one website for a company called ULIS, that makes LWIR microbolometers focal plane arrays out of silicon instead of VOx. I'm not sure if that's the new manufacturing technique I read about before, but it definitely seems to be different from the standard VOx design.

Ulis have been around since at least 2004 making ASi sensors, and the ceramic package came out around 2008 with a 384x288@25um in it.  The 'new' Pico range are a move to 17um pixel pitch accompanied by some functional improvements, and their range of sensors is now from 80x80 to 1024x768.  Note that they are primarily a sensor, not camera, manufacturer.

Internally ASi is similar to VOx, still a 3-dimensional structure, but the silicon bolometer results in worse sensitivity vs non-uniformity although it is more process friendly.  Being French (not American or Israeli ) they are going to be more sensible when it comes to export controls.  UK camera manufacturers Thermoteknix and Argus 'seem........' to be major users of ULIS sensors.

I await hearing their response to you with interest.

[gardner]

Did you read my intent on how I would have used it?

Yes, and its a valid use-case.  I'd be more convinced, though, if you'd used your flir one and done some interesting stitching and image stacking to generate really cool high-res still life or landscape images, and then could demonstrate the limitations of that and why, as an artist, the resolution or image acquisition time is a limiting factor.

The rest of the discussion boils down to "I wish they were cheaper".  To which there is general agreement.

[mhosier]

Following on from the above points, I have used both interpolation and stitching successfully on thermal images from a variety of cameras and both have their place.  Interpolation produces pleasing looking images, and IMHO, works best when it scales to a direct multiple of the original pixel size (eg. 320x240, scaled up to 640x480).  It is weaker where hard edges or very fine details are involved, but as long as the original image was not too noisy, the result is normally good.  The best software I have found for interpolation of thermal images is BFIC, which has a download link in the video description here: 

Image stitching produces sharper images, but it is important to try and keep the images straight, as when they have to be rotated during the stitching process it tends to cause some deterioration of the image, especially if stitching more than a handful of images.  It is neccessary to have some overlap, and it helps if there are clear features to line up in the images.  Flir Tools+ can do this, and a free 30 day trial is available so that you can try it.  As an alternative, there are "non-thermal imaging" stitching programs available.  I use a small package called Rasterstitch, which is quite effective in most cases. 

[Redshift1340]

They're right, image stitching is a great idea for what you're trying to accomplish.  For instance, here's a stitched image encryptededdy shared a while back taken with his Opgal Therm-App (384x288) that looks amazing:)

A ~3000x1000 resolution, fully radiometric, panorama.

1. Stiched RAW images with ICE, output PSD
2. Filled in missing space with Photoshop content aware fill
3. Imagemagick converted image to 8bit scaled PNG.
4. Lightroom noise reduction on said 8bit PNG
5. Imagemagick convert lightroom output back into 16bit RAW PNG
6. Imagemagick convert RAW PNG into FLIR RAW PNG.
7. PHP splitjpg.php to embed FLIR RAW PNG into a FLIR radiometric JPEG.
8. FLIR Tools to add measurements and generate final output.

Noise reduction was necessary as the ~6 degrees C span had a bit of noise.


(non-radiometric preview)

Fully radiometric version, FLIR Radiometric JPEG: http://puu.sh/kgu8H/b1023306ce.jpg

For those wondering, here is the raw version built straight from raw files without noise reduction (this one has a lower 4 degrees C span):


Still usable, not as nice looking.

[Bill W]

The ULIS website has a 6MP video (~ 3000 x 2200) from a 6 up array of 1024x768 cameras.

http://www.ulis-ir.com/index.php?mact=News,cntnt01,detail,0&cntnt01articleid=65&cntnt01origid=15&cntnt01detailtemplate=details_news&cntnt01returnid=64

Some effects as the calibrations on each camera are not synchronised.

Bill

ww.fire-tics.co.uk

[encryptededdy]

They're right, image stitching is a great idea for what you're trying to accomplish.  For instance, here's a stitched image encryptededdy shared a while back taken with his Opgal Therm-App (384x288) that looks amazing:)

Thanks for posting that. Here's another one I made recently.

[Redshift1340]

Thanks for posting that. Here's another one I made recently.

Nice work , That looks incredible.  Thanks for sharing!  If you make any more sometime, I'd love to see em.  Which lens did you use when making this composite? The 6.8mm?