I tried to remove the Bayer layer on some digital cameras back in about 2010 without success. I tried both mechanical removal and chemicals, and like you, ended up with a box of broken cameras. It's kind of strangely satisfying to know that I was right not to waste any more money buying cameras, since the methods I was using were dead ends. Glad someone finally figured out how to do it.
Les! I am just now approaching this in 2023. I saw that you published your findings in IEEE... but it's behind a paywall. I was wondering if you would consider publishing the paper on your website or in an open journal format? That way more people would be able to read and interact with it.
I knew you would eat this shit up. You can just go ahead and mail the OSMU to me, I promise I’ll finish it up nice and neat. 😁 We need the open source measure unit more than you need another eight digit meter! 😁 Regardless, I greatly enjoy all of your contact no matter what you’re working on. Rock on!
@@LesLaboratory I've been keeping an eye on cryocooler prices and they're still way out of my budget, I think these would be a lot easier than messing with raw liquid helium but nowhere near as much fun ;)
Very cool! I'm impressed how "gentle" it is. IIRC silicon will eat that wavelength up to at least 100nm or so, you must have been riding a tight margin between ablating the resist and starting to melt/ablate the top surface. Such a clean and elegant setup, and the results speak for themselves! Great work!
Thanks! Yeah, it is tricky, and quite a critical setup. Most materials happily absorb UV, so yes right on the edge of ablating the Silicon. The job has to be good on the first pass, as subsequent passes will indeed damage the sensor, so yield isn't quite as good as I want it to be... yet! That said, it is much safer than using solvents like NMP, which is particularly nasty stuff.
@@LesLaboratory How about getting the dust off by washing with something like refrigerant. Air duster cans generally are R134A, but not too pure, so probably best to get a refill can for vehicles, the nozzle assembly, and use the liquid to gently wash it clean. Going to leave a water film behind, so do in a sealed vented box, and after use warm air to evaporate the condensed water film, or have a 50W bare halogen lamp running off 6-9V to heat the entire assembly up to drive it off, with the residual gas keeping the dust away.
@@LesLaboratory well, i actually bought one Lambda Physik LPX-200 for my lab for 2500€, 1% of the price of a new one, it was 22 year old. Took me half a year to fix everything. At least we got a lot of space in our garage xD
That was an amazing video ! A few details like the wavelegnth of your laser, the pulse energy, repetition rate, and focal length of the lens would be a nice addition if you'd like to see people reproducing your process.
I'm not sure what chemicals you have tried, but in general photoresists handle strong bases like 5% NaOH pretty badly. Maybe add a bit of isopropanol to that as well. The advantage of a base over strong acids is that you don't need a fume hood. But you should wear gloves and rinse thoroughly afterwards.
I tried Methanolamine as a weak base, but no dice. I was afraid that NaOH would attack the interconnects, but it could be worth a go. I have tried boiling DMSO, DCM, soaking for weeks in acetone, brake fluid, paint stripper etc etc. They are all very good at attacking the rest of the organics though!
@@LesLaboratory Weak bases als well as weak acids (DMSO is only weakly acidic) do not work for dissolving photoresist. Because the resist in color filters is generally crosslinked and has adhesion promotors, strong solvents also do a bad job. The material will swell a bit but will not dissolve. An aggressive acid or a strong base might react with / break down the polymer chains, so that is your best chance if you would like to pursue the chemical route in the future. Btw, the ablation you demonstrate also shows impressive results, never expected it to work this well. Thanks for making this video!
@@HuygensOptics Well, I have a box full of ruined cameras, so it might be worth racking them all up in beakers, and try something more aggressive :-) Me neither! The first test by hand under the microscope was a disaster, but was promising enough to pursue. The pay off in the end was great though!
@@LesLaboratory Do you have some ultrasonic nebulizer / humidifier / cleaning bath ? Inhabers / humidifiers : they produce a tight cone the water comes up, to spray off 3-5µm bobbles from the end, where they reach real hot temperatures, I once melted plastic foil though it was constantly in touch with the water cone. 20W. Just mentioning it to dose the wave to some fraction or pulse it also.
This is giving me flashbacks to my job in laser medical manufacturing. Routinely did work with excimer units and at scales like this. I still have a thorlabs mug I pilfered from the break room!
If there was ever a reason to build Les' nitrogen laser this is it. Now I have even more projects to build, thanks for the great video Les, outstanding work!!!
The real question is why they don't sell a version of the camera without the color filter. I mean the people use these cameras for experimental projects and it is handy of they have an option to select between the two.
I guess there just isn't a huge market. the same issue exists with DSLR sensors as well. The Astronomy folks have been forever scratching away at those.
In machine vision cameras, monochrome sensors without the color filter is quite normal. They are available with different interfaces like USB and Ethernet. Should be no problem using one of them with RasPi. I wonder the whole point of this project is the stripping part itself?
@@hirohyu0220 there are very very limited choices when it comes to monochrome sensors. For example, for the PI there are modules with the OV9281 monochrome sensor that require writing to the I2C bus directly to achieve relatively long exposures. Stand alone monochrome cameras that come at a reasonable price tag usually have their own very limiting UVC driver with no long exposure, so they are only really useful for certain machine vision applications, but not spectroscopy or hyperspectral imaging. All of those typically offer 1 to 2 Megapixel resolutions at best and limited sensitivity, hence the need for long exposures. If you want anything better, be ready to spend important amounts of money. At that point, paying to have a DSLR sensor debayered professionally might be an interesting proposal.
This is amazing! There have been people trying for years to create a reliable and "safe" way of debayering DSLRs for astrophotography use (there's a 10 year old and 100 page thread on stargazers lounge forum dedicated to attempts at this). Would this work for removing the lenses/bayer matrix from Canon DSLR CMOS chips, or is this process specific to Pi cameras? My understanding is some of them use materials other than photoresist. I am not 100% sure of the details. Would the same process work with a CO2 laser (that's all I have access to)?
while CO2 Laser would probably be absorbed without passing through the filter, it can only "burn“ the layer away (CW laser, not pulsed) the extremely short pulses are required for this process. it’s the difference between burning your eye out with a laser, and performing laser eye surgery
Amazing Process! I heard of physical removal, chemical sounded neat, but then *lasers* !. On the note of this being potential the first to try this / document it well on 13:28, would it be of use to write a paper, or get someone to help you to do so?
That was pretty neat! Cool that you could etch it off so effectively. Did you experiment with different amounts of overlap? Might help with the low-light striations.
Wow This is what I missed , I never realized that there's also RGB which absorbs Uv not only the glass lense and Hot mirrors That is very nice but even after removing The filter Layers , will uv pass through it's Glass lense ??
What an awesome video! Thank you so much for sharing! I definitely don't have a laser like this hanging around but I think there is real merit in the use of a stepper-controlled XYZ table like you have set up here for mechanical removal methods as well. I don't trust my shaky hands so I'm thinking I might have a crack at building a tiny CNC for debayering my SLR sensor. Cheers!
Thanks! Mechanical removal may well be achievable with an XYZ stage, and would be an impressive accomplishment. I wonder if perhaps you made the removal tool vibrate, rather than just drag it across the surface whether it might do a better job.
Was surprised you didn't control the laser from the code to have control as you turned around the path etc, but I guess the tolerances on laserexposure were large enough to not worry.
Nice work! I wonder if those artifacts were caused by debris from an earlier pass, masking some of the filter from ablating on a subsequent pass. Is there any worry about the pulsed heating diffusing the dopants in the sensor junctions? Would adding an air jet directed at the surface be helpful for cooling it and removing debris?
Thanks! There is not really muchg heat to deal with as the laser is pulsed. UV may cuase damage to the junctions, but so far so good. Yeas version 2 will have an air jet, and an upgraded laser system.
One thing I might simply oversee. The Pie camera is constructed as a color camera an therefore "thinks" that the array of 4 sub-pixels form one pixel with color information. By removing the filter you basically quadrupled the possible resolution of the sensor but can you get the information also transferred to the pie since the CMOS chip is designed to read in a certain way. Or is the output of the camera simply a row by row information of each pixel and all color definition is done on the pie itself?
You are correct. It is possible to pull raw Bayer data from the picam (the sub pixels) and in effect should as you say, quadruple the resolution. This will take some interesting programming to get right (see Zippering artifact), still it will be an interesting excursion if nothing else!
@@faxezu yes, exactly. At some point, I will get round to writing a program that pulls the raw Bayer data. There are some examples on the picamera site under advanced recipes.
I was thinking about a CNC mod of a similar XYZ stage for some time, so thanks a lot for sharing you build :D Using belts to compensate for the movement of the knobs seems to be obvious in hindsight but I guess everything is. Nice work!
Thanks It just so happened that the ends of the micrometers had a nice bit of steel trim attached that was threaded making attaching the drive pulleys very easy. It's nice how stuff works out sometimes.
Typically fantastic video ! was thinking about Plasma Barrell etchers or ashers could be easy to make and are for photoresist removal, I used to use them with oxygen, but then ya gone and done it in a way cool way ! please keep 'em coming....cheers.
Excellent video. Definitely earned my sub, so much interesting stuff on this channel. When you destroy the bayer mask with this method, it sheds some sort of material, which has me thinking do you destroy the microlenses as well in the process? Very much considering giving this a shot with a cheap cmos sensor as I also have a nitrogen laser. Lot of uses for it I didn’t realize were possible!
Some ESP-32 camera have a separate IR filter. However after seeing this, it's probably only for IR and there is probably a UV filter as well. Also it's not reliable as the camera shipped with ESP-32-Cam vary over the years.
If I am not mistaken I tried removing that layer using simple NaOH solution and it sptripped it imediately But as i read in the datashed there are few lines of dark pixels that are supposed to be covered for calibration purposes.
Awesome! If you do a small production run of these cameras, where should I look to check for availability? You've definitely earned a Wayne's World "we're not worthy."
Absolutely brilliant man! Very awesome! Wow! I've ordered two cheap 2K webcams to play with where one I removed the IR cut filter with hopes sometime this decade to utilized with a hard drive actuator to control the finer diffraction grating (BluRay, etc.) sweep along with a laptop CD/DVD drive sled to control the razor blades collimator slit(s) width. Thinking I can use the hard drive actuator also for controlling a polarizing lens for sweeping polarity too for more dimensions of spectral info of a sample. Man, this is really impressive to me and awesome to see such a simple design and you doing a lot of work to refine. Thank you for sharing! Now, wondering how this debayered Pi Camera mini spectrometer design performs? On a tangent regarding laser ablation... any thoughts of making a Laser Induced Breakdown Spectrometer (LIBS)? That would be neat. I can see you detailing really well a wide range of sample presentation DIY cost effective devices. You're more laser focused (pun intended) however, ICP seems feasible for you I'm guessing. Neat!
Thanks! That sounds like a cool setup for sure! The debayered camera works really well in the spectrometer, but debayering is still a faff, so I have been looking at other sensors. Ah, now LIBS that would be fun! The optical breakdown part is easy, measuring it would be the hard part. IIRC, you ignore the initial flash and measure the afterglow. I will have to do a feasibility study on it.
@@LesLaboratory Very impressive and awesome to see on UA-cam detailed by you. Yeah, about all I was thinking was more algorithms like one to perform background scans and another for subtracting from the sample scans. Seems I left off thinking about the possibility to modify Thermino and now will have to take the time to learn some Py and review your software sometime this season. Was thinking possibly some more calibration and/or performance verification scans, depending on presentation design as well. Yeah... LIBS was just like... uhm... that and ICP seemed right up your alley.
apparently photoresist has polarity so i propose a simple magnetic &/or static field near the laser work area would attract the charged partials as they are lasered off, give it a go with the dead cams to test
Positive and Negative in this case refers to the "image" left behind on the silicon after processing. See here for more info: www.microsi.com/blog/the-difference-between-positive-and-negative-photoresist I should probably at least use a laminar flow of dry air to get rid of the bits. I suspect that is why I seem to end up with "stuck" buts of Bayer filter left.
@@LesLaboratory i guess a simple flexible vax tube behind the work area with a bunch of narrow banned plastic straws poked in to it at 45% to flow direction would pull a reasonable laminar flow where needed
Do you precisly adjust the focus on the lenses and matrix in order to have the silicone out of the focus zone ( and so reduce the energy density at the silicone level), or is is it too hard to have that kind of control ?
Pretty much. There is a fine line between effectively removing the Bayer layer and damaging the sensor. When damage occurs it latches the pixels permanently white.
Hello, I'm really impressed by your setup. It works really well. Do you think the laser could work with epoxy? I'm looking for a safe way to decap chips, but it's really hard not to damage the surface of the die. You're the first person to use a laser on a die I've seen that doesn't destroy it. Nice job in any case. I would really like to see more experiments like this, with integrated circuits. Regards.
Thanks! that is an interesting idea, perhaps I could de-cap with a CO2 laser first, then remove the final layers with the nitrogen laser. I might give this a go and see how it turns out!
This has gotta be one of the coolest home-projects ever. Frickin' laser beams! I'd love to try this on a larger sensor for general photography but I feel like it would be more cost effective to buy a monochrome Leica unless I got really lucky finding deals surplus equipment. ;w;
Leica is over-inflated-ly expensive, and when you know the brand supports the Chinese government, now you have another reason not to buy that brand. Rather, get yourself a monochrome industrial camera from e.g. FLIR (ex point grey) or pixelink, these I know you can get the IR cut filter off easily. They are not full frame but much larger than the pi camera, at 2/3" or 1" diagonal. I've got them to work on the raspberry pi.
This is a very interesting project. I was also thinking about building and coding an application for a spectrometer with an RPI camera for my future bachelor's thesis but I have always been bothered by the poor UV performance of the camera due to the Color Filter Array. Removing it like in your video seems to be really effective but getting your hands on such lasers and the other stuff needed is beyond the scope of my project. Could it also be possible to "bleach" the CFA with UVC germicidal lamps emitting UVC light at 254nm or using ozone producing UVC lamps which which have a quartz tube that is not doped and therefore emit at 254nm and 185nm? I guess that the ozone could also help to attack the color filters.
That is an interesting thought, and one I considered briefly. I 'baked' one once in an eprom eraser, but saw no detectable degradation of the Bayer layer, but to be fair I did not try too hard, because by that time, this project was already in the works! In principle it may be achievable, it might just be a matter of time and processes...
Thanks! No, it is on the cards though, I just need to find time to read the picam Python manual to do it correctly (too much day-job work on lately!) Smoothing will need to be done as well, but I am expecting that great things can be done!
might i ask how did you focus the 337nm beam into the fiber? how did you connect it to the ml100? how did you achive focus as everyy glass blocks 337nm, did you had to use a special fiber core?
You should consider selling this as either a service, or as value-added merch. I have interest in a CCD without a bayer filter, and I'm sure others do as well. So... am I right the to assume the the NoIR raspberry pi cameras still have the Bayer filter? They just have the IR screen removed....right?
Neat work, Les. I am particularly interested in how you automatically control the translation stage using step motors. Could you let me know the step motor model number and the coupler installed on the micrometer knobs to work with the step motor belts? Thanks! P.s. I checked your IEEE paper but didn't find it.
Wow, when you do a job you don't mess around. Is piranha pretty effective? It doesn't seem like it would be _that_ much of a hazard given you'd be mixing it _dropwise._ 3 drops H2SO4 to 1 drop H2O2.
Thanks, you might as well go "all in" with these things. I haven't used Piranha, no fume hood or poper handling equipment at the moment. It is however highly corrosive to all organics including flesh and if misused can explode. I think "Nile Red" demos it on his channel. At some point, it would be nice to acquire a place to do stuff like this. I miss chemistry.
@@LesLaboratory I find in chemistry the dangers are often quite over-exaggerated. Look for NileRed's video on pouring various acids on skin. It's reposted somewhere. People often repeat the warnings from SDS papers, but have you seen the SDS for water? Or salt? Piranha doesn't really produce vapors, but just to be safe you can use it outdoors or under your cooker exhaust. Just don some gloves and safety glasses. Explosions/eruptions are typically the result of unintentional mixing of waste solutions, piranha and acetone in particular. Don't get me wrong, even unmixed the two chemicals deserve respect, but again, using such a small volume doesn't really present a danger unless gross incompetence or carelessness is involved. Just dilute it when you're done, and keep a container of baking soda nearby if you're paranoid.
@@WaffleStaffel For sure. Same goes with electricity, radiation etc, etc. If I'm honest, it is present circumstances that put the brakes on creative chemistry shall we say. It wasn't that long ago I was chemically extracting gold from "things" , I'm certainly not afraid of it, though I do have a healthy respect s with all things. I just lack the facilities at the moment to really get into it.
@@LesLaboratory I never thanked you for putting this up and sharing the program. Thanks man, this is a real treasure. How do you find the cameras performance with UV? I saw the video with the assortment of UV bulbs, but it seemed to bottom out at 365.
Hi Les, I wonder if this would work also with laser engravers in the 450nm range. I also requested your paper on research gate, but would have prefered to write a more personal message. Just sent it of by mistake. Best, Michael
Primarily because Monochrome versions (5 Megapixels+) are expensive ~500 USD for a Debayered Picam! The other motivation was the challenge. Though the project was quite expensive, I have learned things and of course I can make as many debayered cams as I like. Now that is is demonstrably possible, I can refine the process and perhaps try other sensors as well.
Hi Les, interesting and impressive experiment! For a similar result could a laser used in cnc engravers be used? Power 5W - 10W? Maybe with more passes, the laser being set to a lower power for better control? Thank you!
Unfortunately, probably not. The Laser used in this video produces 10's of Kilowatts per pulse, enough to explosively disassemble the Bayer layer. A CW Laser would likely melt the layer through thermal effects, and will likely damage the sensor. That said, I am a huge fan of actually trying things out and being persistent, so who knows!
@@LesLaboratory Hi Les, thank you very much. Looks like there is a very fine line betwen success and error. Your work is on that edge! Spectacular accuracy!
This was my thought as well. It would be a total game changer if this could be done with easily available laser engravers. I kind of want to have a stab at it... Even though Les' reasoning for why it shouldn't work makes total sense.
This is positively brilliant! Do you think you could make a pushbroom hyperspectral camera for use with microscopes? I have been slowly chipping away at my own, but wondering how you might go about it given the unique solutions you come up with
@@LesLaboratory I made for document authentication. It has a monochromator that illuminates just before the tube lens. At the ocular I have a slit followed by a beam splitter. The camera has its lens removed and the imaging sensor exposed. The grating had a lens in front of it and the grating reflects back through the lens and into the beam splitter and directly images the diffracted slit onto the sensor. I have a scanning table to fully translate the specimen to get the full document.
Wow, fantastic results! - Is there any chance you could get high enough peak pulse power from a tightly-focused UV laser diode to ablate the photoresist? - You can get UV metal-marking machines (galvo-driven mirrors) for under $5K, I wonder if you could throttle one of those down enough to work without damaging the sensor? - Per another thread below, I suspect blasting with UVC for a long while, followed by a hot solvent (high-proof IPA? Acetone? TCE?) might do the trick. The UV may or may not bleach the CFA dyes, but enough UV might break the polymer bonds, and the resulting shorter chains could be removed with a solvent. Fantastic work though, maybe you should make a little side-business of stripping camera sensors! :-)
Cheers! The response is flatter, and in principle they should be more sensitive, since the filter in a red sub-pixel for example removes green and blue light, but I have not quantitatively measured this.
@@stefanmayer444 I probably still have a half stripped one from testing. It is interesting to see one half in color and one half in B&W. I wonder if there is an actual use for a sensor that does that?
@@LesLaboratory Cool, that would be very interesting to see! I don't know any applications for a half half sensor, other than demonstrational purposes. But what about a striped pattern? An application could maybe be digital night vision. One line would be the sensitive but B&W, and the other would be in colour for the indentification of colors. Traditional night vision tubes are monochrome, but are fairly sensitive. Digital night vision isn't as sensitive yet, but has the advantage of differentiating colors. With your method, it would be possible to strip the sensor only partially in a controlled manner. For such a niche product it might be more economical to use your method, but I'd use a different sensor for that application. Do you know if other sensors behave the same way as the PI cam? Sensitive sensors are expensive to experiment with ;-)
Wow this is a great application to remove bayer filter. I was thinking of using the bluray tracking from a drive or similar setup to use the nitrogen laser output to remove the layers. Still wondering if you used a quartz lens or a high purity plastic lens on the focusing etc..... Some thoughts.... Would be intriguing to use the imaging of x-ray or soft x-ray detection and monitoring etc.
Thanks! I do not think Blu Ray laser diodes would be anywhere near powerful enough, you need brief, high intensity pulses. The lens is a 'glass' of some description that passes UV. Plastic won't transmit UV very well (if at all). Sure the sensors will probably detect soft x-rays, but there are better ways of doing that too...
@@LesLaboratory I guess i am not very clear in what I was trying to convey, what I meant was to the use the structured support of the blu ray laser setup to scan and remove the coating using another laser( Uv laser) instead of the existing blu ray semiconductor etc....sorry for the confusion.
Shortly... I never expected that that many people would want one. I will be doing OV5647 sensors to begin with. Depending on yields, I may consider debayering the Pro version.
Hello Les, why not using a monochrome sensor? If the specific version of the monochrome sensor is not available, would you be keen to provide this as a service?
There aren't any monochrome pi-cams. Yes, I might well do. I am trying to get the yield up and am in the process of upgrading the whole machine into something more reliable and repeatable.
@@LesLaboratory I have not yet done so. There is no reason why it would not work as basically they just create the breakoutboard/drivers for Sony (and other) manufacturers. The sony sensor with higher dynamic range would have a better linear response to radiance (which you could test in your integrating sphere).
@@gtalckmin let know how it goes. When I tried it I found the driver support to be poor, and the resolution was not spectacular. I have tried a nice black and white is cam, but again, drivers and resolution. In contrast the support for the picam is vast. All that said, it would be cool to analyse and compile a list of astronomy/spectroscopy capable cams for the pi!
Have you taken any sample images with the filter removed and a normal lens to show that the total pixel resolution is like 4x greater? (32megapixels b&w image)?
I have taken pictures with it, but what you are really asking for, is to pull the raw pixel data off with all the sub-pixels visible. I have not tried this, but the picam should support it...
Have you ever tried this with a Dslr camera sensor, there's quite a few amateur astronomers tried to debayer their cameras (mostly canon) with varying success. You could probably make a few quid if it works
This method also removes the micro lens layer so there will probably be no sensitivity gain in the visible range at all or even reduced sensitivity. Therefore this method might not be suitable for astro cameras.
@@martinbruckner2109 not really to do with sensitivity, its to allow narrowband imaging using different filters to mitigate light pollution. dedicated mono astro cameras are hideously expensive in comparison. i've stripped and re-cased a canon 1100D ready to add peltier cooling to the sensor but at the moment it's still basically a colour camera with the various pitfalls of using colour one-shot imaging
Hi, I am a PhD student at ETH Zurich. I need a camera with as small pixel size as possible to image an optical lattice. The Pi camera v2 would be perfect for this except for the bayer filter. I tried removing it with various chemicals without success. I was wondering if I could pay you to debayer a few of these sensors
It probably is 16MW, I used to play with a KrF laser when not getting a PhD, it had a 1ns pulse length at 10Hz at which point 16MW during the pulse is less than 1W on average.
Makes sense, that would mean output of 10's of millijoules. Would have been nice if the paper had given figures in MW/cm3, and given some more data on stripping speed, etc. The N2 Laser puts out just a few 10's of kW as best, and takes a good half hour to strip the filter.
@@LesLaboratory yeah, it is coming back to me, the pulses were up to 1J, so 1GW for a ns. It was later ablation of YBCO to deposit in think films so needed to make rather more of a plumber than your application. Very cool stuff by the way. I played with webcams and spectroscopes attached to a pi 1 for a museum exhibit a in long time ago and have been meaning to try again with a picam ever since. You are doing it a lot more thoroughly than I would have.
Absolutley. I am in the process of building a part 2 for this, with a fan and an upgraded laser system for fast stripping! It takes far too long with a niotrogen laser!
The Laser engraver is unlikely to do the job, given its wavelength. To disassemble organics like this you really need UV to break chemical bonds, without damaging the silicon. I have tried with a CO2 Laser, but all it does it heat up the Silicon, it won't do anything at all to the Bayer layer. The photoresists used in these in very tough stuff.
I destroyed sensors initially whist I was working out the damage threshold for the silicon. The pixels latch up white when they are damaged. There are a number of factors to take into account when doing this to get it right, such as peak power, spot size and beam angle.
This is super interesting! Thank you for the video. Have you ever considered offering this as a service? It looks like you could do this without even having do dismantle the camera. I am a fellow creator with a channel for filmmakers… if you where interested to commercialise this and/or try some things with a little higher end gear (I will NOT give you my Alexa for this 😂), let me know
Hi, I would like to do some tests on corona discharge, using a pi camera without bayer filter. Unfortunately I don't have all the equipment you have. How can I do? Can i buy a pi cam from you without bayer filter? I thank you (Daniele)
Yes, I might well do. I am trying to get the yield up and am in the process of upgrading the whole machine into something more reliable and repeatable.
Yes, but is is getting hold of them at a reasonable price that is the issue. I did try a single lens element that transmits a good portion of UV to good effect though.
I know I'm late to the party, but are there any plans to do another small production run on this? I am very interested. I sent you an email with details but have not yet heard back.
There is interest so I will be doing a short run when more Picams arrive. I would be real interested to see ho they perform in astophotography applications...
@@LesLaboratory For astrophotography i would prefer old full frame Nikons, Canons. A bigger sensor with 12 or 16mp. There is a ton of these used cameras on aliexpress. Maxmax already sells new modified Sony A6000 but they jump from $600 to $2000. For $4000 you could buy a new cilíndricas full frame astro cam like ZWO. There is nothing on the cheap camera niche. It is a shame there is no new replacement cmos for these cameras. Sometimes you find some for $14 but it is a limited supply
I was interested in stripping a genuine picamera, because I wanted a sensible spectral response out of it. Third party cameras (at least the ones I tried) have poor driver support on the Pi too, and that was another factor. The other reason, was to see if it could even be done, as there are applications beyond what is demonstrated here. Have you used the FSM-AR0144, and is it any good? I am always keen to try new cameras!
I tried to remove the Bayer layer on some digital cameras back in about 2010 without success. I tried both mechanical removal and chemicals, and like you, ended up with a box of broken cameras. It's kind of strangely satisfying to know that I was right not to waste any more money buying cameras, since the methods I was using were dead ends. Glad someone finally figured out how to do it.
did you try using NaOH?
@@deltaxcd have you used it with any success? The usual recommendation is NMP... but even that has very limited success.
@@h.n.4060 Į already forgot how it worked but I think yes I used it and it instantly dissolved everything
Les! I am just now approaching this in 2023. I saw that you published your findings in IEEE... but it's behind a paywall. I was wondering if you would consider publishing the paper on your website or in an open journal format? That way more people would be able to read and interact with it.
awesome! what's next? liquid helium cooled raspi camera? 😁
or at least the RasPi HQ camera
I knew you would eat this shit up. You can just go ahead and mail the OSMU to me, I promise I’ll finish it up nice and neat. 😁
We need the open source measure unit more than you need another eight digit meter! 😁
Regardless, I greatly enjoy all of your contact no matter what you’re working on. Rock on!
Thanks! Now there is an idea! Liquid Helium would be great fun!
@@LesLaboratory I've been keeping an eye on cryocooler prices and they're still way out of my budget, I think these would be a lot easier than messing with raw liquid helium but nowhere near as much fun ;)
@@danriches7328 Oh I dunno, condensing you own gasses would be a hoot!
Very cool! I'm impressed how "gentle" it is. IIRC silicon will eat that wavelength up to at least 100nm or so, you must have been riding a tight margin between ablating the resist and starting to melt/ablate the top surface. Such a clean and elegant setup, and the results speak for themselves! Great work!
Thanks! Yeah, it is tricky, and quite a critical setup.
Most materials happily absorb UV, so yes right on the edge of ablating the Silicon. The job has to be good on the first pass, as subsequent passes will indeed damage the sensor, so yield isn't quite as good as I want it to be... yet! That said, it is much safer than using solvents like NMP, which is particularly nasty stuff.
@@LesLaboratory How about getting the dust off by washing with something like refrigerant. Air duster cans generally are R134A, but not too pure, so probably best to get a refill can for vehicles, the nozzle assembly, and use the liquid to gently wash it clean. Going to leave a water film behind, so do in a sealed vented box, and after use warm air to evaporate the condensed water film, or have a 50W bare halogen lamp running off 6-9V to heat the entire assembly up to drive it off, with the residual gas keeping the dust away.
Excimer laser is generally 2 meter long 1 meter wide 500kg unit with 200000$ price tag. Definetly something that you need in your garage xD
My thoughts exactly. There are smaller KrF Excimers, but they seem to e experimental units rather than commercial!
@@LesLaboratory well, i actually bought one Lambda Physik LPX-200 for my lab for 2500€, 1% of the price of a new one, it was 22 year old. Took me half a year to fix everything. At least we got a lot of space in our garage xD
@@RoosieBoomstick Beautiful! That would have been an excellent project!
Awesome work! The Pi camera with no Bayer filter and cover glass removed would make a great low cost laser beam profiler
For sure, although the sensor is quite small. I did some beam profiling some time ago, using ImageJ to visualise the data
That was an amazing video !
A few details like the wavelegnth of your laser, the pulse energy, repetition rate, and focal length of the lens would be a nice addition if you'd like to see people reproducing your process.
Especially if he broke many cameras finding the exact right specs to use
@@sirtimatbob You'd think you'd only need to break one camera, because you can test it on a small area of the sensor.
I'm not sure what chemicals you have tried, but in general photoresists handle strong bases like 5% NaOH pretty badly. Maybe add a bit of isopropanol to that as well. The advantage of a base over strong acids is that you don't need a fume hood. But you should wear gloves and rinse thoroughly afterwards.
I tried Methanolamine as a weak base, but no dice. I was afraid that NaOH would attack the interconnects, but it could be worth a go.
I have tried boiling DMSO, DCM, soaking for weeks in acetone, brake fluid, paint stripper etc etc.
They are all very good at attacking the rest of the organics though!
@@LesLaboratory Weak bases als well as weak acids (DMSO is only weakly acidic) do not work for dissolving photoresist. Because the resist in color filters is generally crosslinked and has adhesion promotors, strong solvents also do a bad job. The material will swell a bit but will not dissolve. An aggressive acid or a strong base might react with / break down the polymer chains, so that is your best chance if you would like to pursue the chemical route in the future. Btw, the ablation you demonstrate also shows impressive results, never expected it to work this well. Thanks for making this video!
@@HuygensOptics Well, I have a box full of ruined cameras, so it might be worth racking them all up in beakers, and try something more aggressive :-)
Me neither! The first test by hand under the microscope was a disaster, but was promising enough to pursue. The pay off in the end was great though!
@@LesLaboratory Do you have some ultrasonic nebulizer / humidifier / cleaning bath ? Inhabers / humidifiers : they produce a tight cone the water comes up, to spray off 3-5µm bobbles from the end, where they reach real hot temperatures, I once melted plastic foil though it was constantly in touch with the water cone. 20W. Just mentioning it to dose the wave to some fraction or pulse it also.
This is giving me flashbacks to my job in laser medical manufacturing. Routinely did work with excimer units and at scales like this. I still have a thorlabs mug I pilfered from the break room!
If there was ever a reason to build Les' nitrogen laser this is it. Now I have even more projects to build, thanks for the great video Les, outstanding work!!!
The real question is why they don't sell a version of the camera without the color filter. I mean the people use these cameras for experimental projects and it is handy of they have an option to select between the two.
I guess there just isn't a huge market. the same issue exists with DSLR sensors as well. The Astronomy folks have been forever scratching away at those.
In machine vision cameras, monochrome sensors without the color filter is quite normal. They are available with different interfaces like USB and Ethernet. Should be no problem using one of them with RasPi. I wonder the whole point of this project is the stripping part itself?
@@hirohyu0220 there are very very limited choices when it comes to monochrome sensors.
For example, for the PI there are modules with the OV9281 monochrome sensor that require writing to the I2C bus directly to achieve relatively long exposures.
Stand alone monochrome cameras that come at a reasonable price tag usually have their own very limiting UVC driver with no long exposure, so they are only really useful for certain machine vision applications, but not spectroscopy or hyperspectral imaging.
All of those typically offer 1 to 2 Megapixel resolutions at best and limited sensitivity, hence the need for long exposures.
If you want anything better, be ready to spend important amounts of money.
At that point, paying to have a DSLR sensor debayered professionally might be an interesting proposal.
Wow. So casually doing something awesome. Subscribed.
Thanks! :-)
@@LesLaboratory do you think it would work like that with other sensors too (DSLRs etc)?
Oh man, that is super cool!
Amazing video Thank you for all details
You are absolutely AMAZING! I love your videos so much. 😍😍😍
Thank you for your fantastic work! 💕
Thank you! I will keep them coming! 😀
This is amazing! There have been people trying for years to create a reliable and "safe" way of debayering DSLRs for astrophotography use (there's a 10 year old and 100 page thread on stargazers lounge forum dedicated to attempts at this). Would this work for removing the lenses/bayer matrix from Canon DSLR CMOS chips, or is this process specific to Pi cameras? My understanding is some of them use materials other than photoresist. I am not 100% sure of the details. Would the same process work with a CO2 laser (that's all I have access to)?
while CO2 Laser would probably be absorbed without passing through the filter, it can only "burn“ the layer away (CW laser, not pulsed)
the extremely short pulses are required for this process. it’s the difference between burning your eye out with a laser, and performing laser eye surgery
Very nice work mate
Amazing Process! I heard of physical removal, chemical sounded neat, but then *lasers* !. On the note of this being potential the first to try this / document it well on 13:28, would it be of use to write a paper, or get someone to help you to do so?
Thanks for the excellent video and explanation. However, you should have done the sunscreen experiment at the end 🙂
I should! I think I would need daylight for that though, and a better lens...
That was pretty neat! Cool that you could etch it off so effectively. Did you experiment with different amounts of overlap? Might help with the low-light striations.
Wow This is what I missed , I never realized that there's also RGB which absorbs Uv not only the glass lense and Hot mirrors That is very nice but even after removing The filter Layers , will uv pass through it's Glass lense ??
What an awesome video! Thank you so much for sharing! I definitely don't have a laser like this hanging around but I think there is real merit in the use of a stepper-controlled XYZ table like you have set up here for mechanical removal methods as well. I don't trust my shaky hands so I'm thinking I might have a crack at building a tiny CNC for debayering my SLR sensor. Cheers!
Thanks! Mechanical removal may well be achievable with an XYZ stage, and would be an impressive accomplishment. I wonder if perhaps you made the removal tool vibrate, rather than just drag it across the surface whether it might do a better job.
Was surprised you didn't control the laser from the code to have control as you turned around the path etc, but I guess the tolerances on laserexposure were large enough to not worry.
Nice work! I wonder if those artifacts were caused by debris from an earlier pass, masking some of the filter from ablating on a subsequent pass. Is there any worry about the pulsed heating diffusing the dopants in the sensor junctions?
Would adding an air jet directed at the surface be helpful for cooling it and removing debris?
Thanks! There is not really muchg heat to deal with as the laser is pulsed. UV may cuase damage to the junctions, but so far so good. Yeas version 2 will have an air jet, and an upgraded laser system.
Really awesome and inspiring.
Thanks!
Excellent! Love this videos
Thanks!
One thing I might simply oversee.
The Pie camera is constructed as a color camera an therefore "thinks" that the array of 4 sub-pixels form one pixel with color information. By removing the filter you basically quadrupled the possible resolution of the sensor but can you get the information also transferred to the pie since the CMOS chip is designed to read in a certain way. Or is the output of the camera simply a row by row information of each pixel and all color definition is done on the pie itself?
You are correct. It is possible to pull raw Bayer data from the picam (the sub pixels) and in effect should as you say, quadruple the resolution.
This will take some interesting programming to get right (see Zippering artifact), still it will be an interesting excursion if nothing else!
@@LesLaboratory Ah, thanks for the reply!
So the images we saw in the video average the 4 subpixels to get the brightness information per pixel?
@@faxezu yes, exactly. At some point, I will get round to writing a program that pulls the raw Bayer data. There are some examples on the picamera site under advanced recipes.
I was thinking about a CNC mod of a similar XYZ stage for some time, so thanks a lot for sharing you build :D Using belts to compensate for the movement of the knobs seems to be obvious in hindsight but I guess everything is. Nice work!
Thanks It just so happened that the ends of the micrometers had a nice bit of steel trim attached that was threaded making attaching the drive pulleys very easy. It's nice how stuff works out sometimes.
This would be a way to make a very sensitive camera sensor for an x ray imager or similar
Great Video Les. Thank you. What adjustable table do you use? is it DIY?
I know a few people with new UV galvo laser engravers.... I feel like this would be amazing for this
Is it became sensitive towards thermal camera spectrum ? And will its wavelength sensitivity portion move if cool down to - c using LN or peltier ?
Could use a nice Sterling model! (Our thermal lab has a pro grade IR cam with sterling cooling...)
Typically fantastic video ! was thinking about Plasma Barrell etchers or ashers could be easy to make and are for photoresist removal, I used to use them with oxygen, but then ya gone and done it in a way cool way ! please keep 'em coming....cheers.
Thanks! I did consider building a plasma etcher, I have a decent diffusion pump, and could have acquired or hacked together a chamber, but, Lasers!
Excellent video. Definitely earned my sub, so much interesting stuff on this channel. When you destroy the bayer mask with this method, it sheds some sort of material, which has me thinking do you destroy the microlenses as well in the process? Very much considering giving this a shot with a cheap cmos sensor as I also have a nitrogen laser. Lot of uses for it I didn’t realize were possible!
Some ESP-32 camera have a separate IR filter. However after seeing this, it's probably only for IR and there is probably a UV filter as well. Also it's not reliable as the camera shipped with ESP-32-Cam vary over the years.
If I am not mistaken I tried removing that layer using simple NaOH solution and it sptripped it imediately
But as i read in the datashed there are few lines of dark pixels that are supposed to be covered for calibration purposes.
Interesting, I tried this but killed the sensors as it ate the aluminium interconnects on the chip. Under what conditions was this done?
Awesome! If you do a small production run of these cameras, where should I look to check for availability? You've definitely earned a Wayne's World "we're not worthy."
Glad you like the video! Ping me an e-mail if you want one (details in about page)
For the purpose of spectrograph wouldn't it be easier to remove bayer layer etc just on the diagonal like 20px width equivalent ?
Absolutely brilliant man! Very awesome! Wow! I've ordered two cheap 2K webcams to play with where one I removed the IR cut filter with hopes sometime this decade to utilized with a hard drive actuator to control the finer diffraction grating (BluRay, etc.) sweep along with a laptop CD/DVD drive sled to control the razor blades collimator slit(s) width. Thinking I can use the hard drive actuator also for controlling a polarizing lens for sweeping polarity too for more dimensions of spectral info of a sample. Man, this is really impressive to me and awesome to see such a simple design and you doing a lot of work to refine. Thank you for sharing! Now, wondering how this debayered Pi Camera mini spectrometer design performs? On a tangent regarding laser ablation... any thoughts of making a Laser Induced Breakdown Spectrometer (LIBS)? That would be neat. I can see you detailing really well a wide range of sample presentation DIY cost effective devices. You're more laser focused (pun intended) however, ICP seems feasible for you I'm guessing. Neat!
Thanks! That sounds like a cool setup for sure! The debayered camera works really well in the spectrometer, but debayering is still a faff, so I have been looking at other sensors.
Ah, now LIBS that would be fun! The optical breakdown part is easy, measuring it would be the hard part. IIRC, you ignore the initial flash and measure the afterglow. I will have to do a feasibility study on it.
@@LesLaboratory Very impressive and awesome to see on UA-cam detailed by you. Yeah, about all I was thinking was more algorithms like one to perform background scans and another for subtracting from the sample scans. Seems I left off thinking about the possibility to modify Thermino and now will have to take the time to learn some Py and review your software sometime this season. Was thinking possibly some more calibration and/or performance verification scans, depending on presentation design as well. Yeah... LIBS was just like... uhm... that and ICP seemed right up your alley.
apparently photoresist has polarity so i propose a simple magnetic &/or static field near the laser work area would attract the charged partials as they are lasered off, give it a go with the dead cams to test
Positive and Negative in this case refers to the "image" left behind on the silicon after processing. See here for more info: www.microsi.com/blog/the-difference-between-positive-and-negative-photoresist
I should probably at least use a laminar flow of dry air to get rid of the bits. I suspect that is why I seem to end up with "stuck" buts of Bayer filter left.
@@LesLaboratory i guess a simple flexible vax tube behind the work area with a bunch of narrow banned plastic straws poked in to it at 45% to flow direction would pull a reasonable laminar flow where needed
Do you precisly adjust the focus on the lenses and matrix in order to have the silicone out of the focus zone ( and so reduce the energy density at the silicone level), or is is it too hard to have that kind of control ?
Pretty much. There is a fine line between effectively removing the Bayer layer and damaging the sensor. When damage occurs it latches the pixels permanently white.
Awesome work! Would you mind sharing the code you used to drive the steppers and gate the laser?
wow, this is great, love it.
I am work on spectrometer, I wish i could contact you for more ideas.
You can, e-mail is in the bio.
Hello,
I'm really impressed by your setup. It works really well.
Do you think the laser could work with epoxy? I'm looking for a safe way to decap chips, but it's really hard not to damage the surface of the die.
You're the first person to use a laser on a die I've seen that doesn't destroy it.
Nice job in any case. I would really like to see more experiments like this, with integrated circuits.
Regards.
Thanks! that is an interesting idea, perhaps I could de-cap with a CO2 laser first, then remove the final layers with the nitrogen laser. I might give this a go and see how it turns out!
@@LesLaboratory Can't wait to see your experiments!
This has gotta be one of the coolest home-projects ever. Frickin' laser beams! I'd love to try this on a larger sensor for general photography but I feel like it would be more cost effective to buy a monochrome Leica unless I got really lucky finding deals surplus equipment. ;w;
Thanks! Yeah, me too, but they are real expensive. I am hoping some old camera body turns up on eBay that I can have a crack at :-)
Leica is over-inflated-ly expensive, and when you know the brand supports the Chinese government, now you have another reason not to buy that brand. Rather, get yourself a monochrome industrial camera from e.g. FLIR (ex point grey) or pixelink, these I know you can get the IR cut filter off easily. They are not full frame but much larger than the pi camera, at 2/3" or 1" diagonal. I've got them to work on the raspberry pi.
This is a very interesting project. I was also thinking about building and coding an application for a spectrometer with an RPI camera for my future bachelor's thesis but I have always been bothered by the poor UV performance of the camera due to the Color Filter Array. Removing it like in your video seems to be really effective but getting your hands on such lasers and the other stuff needed is beyond the scope of my project. Could it also be possible to "bleach" the CFA with UVC germicidal lamps emitting UVC light at 254nm or using ozone producing UVC lamps which which have a quartz tube that is not doped and therefore emit at 254nm and 185nm? I guess that the ozone could also help to attack the color filters.
That is an interesting thought, and one I considered briefly. I 'baked' one once in an eprom eraser, but saw no detectable degradation of the Bayer layer, but to be fair I did not try too hard, because by that time, this project was already in the works!
In principle it may be achievable, it might just be a matter of time and processes...
brilliantly done.
Truly amazing project. Did you turn off debayering on the Pi when capturing images afterwards?
Thanks!
No, it is on the cards though, I just need to find time to read the picam Python manual to do it correctly (too much day-job work on lately!) Smoothing will need to be done as well, but I am expecting that great things can be done!
might i ask how did you focus the 337nm beam into the fiber? how did you connect it to the ml100? how did you achive focus as everyy glass blocks 337nm, did you had to use a special fiber core?
You should consider selling this as either a service, or as value-added merch. I have interest in a CCD without a bayer filter, and I'm sure others do as well.
So... am I right the to assume the the NoIR raspberry pi cameras still have the Bayer filter? They just have the IR screen removed....right?
Literally DIY laser eye surgery, that’s amazing
Thanks! :-D
Sooo awesome!! Subscribed! ☺️😊
Thanks for your support!
Neat work, Les. I am particularly interested in how you automatically control the translation stage using step motors. Could you let me know the step motor model number and the coupler installed on the micrometer knobs to work with the step motor belts? Thanks! P.s. I checked your IEEE paper but didn't find it.
I missed this part-- what was the light source in the spectrophotometer?
Just an idea : UV light washes colors, why not expose the image sensor to strong UV light for a long time? I think the RGB colors will decay.
The color might, but the photoresist itself is likely strongly absorbing in the UV. Worth a try though!
amazing!So is possible to convert a standard Sony a7ii camera to a monochrome camera?
Got a 3018 with a 2.5W blue diode laser, wonder it that would work? It burns the paint off black business cards.
Wow, when you do a job you don't mess around. Is piranha pretty effective? It doesn't seem like it would be _that_ much of a hazard given you'd be mixing it _dropwise._ 3 drops H2SO4 to 1 drop H2O2.
Thanks, you might as well go "all in" with these things. I haven't used Piranha, no fume hood or poper handling equipment at the moment. It is however highly corrosive to all organics including flesh and if misused can explode. I think "Nile Red" demos it on his channel.
At some point, it would be nice to acquire a place to do stuff like this. I miss chemistry.
@@LesLaboratory I find in chemistry the dangers are often quite over-exaggerated. Look for NileRed's video on pouring various acids on skin. It's reposted somewhere. People often repeat the warnings from SDS papers, but have you seen the SDS for water? Or salt? Piranha doesn't really produce vapors, but just to be safe you can use it outdoors or under your cooker exhaust. Just don some gloves and safety glasses. Explosions/eruptions are typically the result of unintentional mixing of waste solutions, piranha and acetone in particular. Don't get me wrong, even unmixed the two chemicals deserve respect, but again, using such a small volume doesn't really present a danger unless gross incompetence or carelessness is involved. Just dilute it when you're done, and keep a container of baking soda nearby if you're paranoid.
@@WaffleStaffel For sure. Same goes with electricity, radiation etc, etc. If I'm honest, it is present circumstances that put the brakes on creative chemistry shall we say. It wasn't that long ago I was chemically extracting gold from "things" , I'm certainly not afraid of it, though I do have a healthy respect s with all things. I just lack the facilities at the moment to really get into it.
@@LesLaboratory I never thanked you for putting this up and sharing the program. Thanks man, this is a real treasure. How do you find the cameras performance with UV? I saw the video with the assortment of UV bulbs, but it seemed to bottom out at 365.
Hi Les, I wonder if this would work also with laser engravers in the 450nm range. I also requested your paper on research gate, but would have prefered to write a more personal message. Just sent it of by mistake.
Best, Michael
Great peace of work but why did you not just buy a monochrome sensor?
Primarily because Monochrome versions (5 Megapixels+) are expensive ~500 USD for a Debayered Picam!
The other motivation was the challenge. Though the project was quite expensive, I have learned things and of course I can make as many debayered cams as I like. Now that is is demonstrably possible, I can refine the process and perhaps try other sensors as well.
Hi Les, interesting and impressive experiment! For a similar result could a laser used in cnc engravers be used? Power 5W - 10W? Maybe with more passes, the laser being set to a lower power for better control? Thank you!
Unfortunately, probably not. The Laser used in this video produces 10's of Kilowatts per pulse, enough to explosively disassemble the Bayer layer. A CW Laser would likely melt the layer through thermal effects, and will likely damage the sensor.
That said, I am a huge fan of actually trying things out and being persistent, so who knows!
@@LesLaboratory Hi Les, thank you very much. Looks like there is a very fine line betwen success and error. Your work is on that edge! Spectacular accuracy!
This was my thought as well. It would be a total game changer if this could be done with easily available laser engravers. I kind of want to have a stab at it... Even though Les' reasoning for why it shouldn't work makes total sense.
So is it like a monochrome astrophotography camera?
It might be possible to use a xenon flash tube to deliver the required illumination.
An interesting Idea. I one accidentally removed all the color from a cloth by exposing it to a 100 Joule flash from a laser lamp.
This is positively brilliant!
Do you think you could make a pushbroom hyperspectral camera for use with microscopes? I have been slowly chipping away at my own, but wondering how you might go about it given the unique solutions you come up with
Honestly, I had to google that one! I have never come across one in my travels, but it does look very interesting! What is your application?
@@LesLaboratory I made for document authentication. It has a monochromator that illuminates just before the tube lens. At the ocular I have a slit followed by a beam splitter. The camera has its lens removed and the imaging sensor exposed. The grating had a lens in front of it and the grating reflects back through the lens and into the beam splitter and directly images the diffracted slit onto the sensor. I have a scanning table to fully translate the specimen to get the full document.
Wow, fantastic results!
- Is there any chance you could get high enough peak pulse power from a tightly-focused UV laser diode to ablate the photoresist?
- You can get UV metal-marking machines (galvo-driven mirrors) for under $5K, I wonder if you could throttle one of those down enough to work without damaging the sensor?
- Per another thread below, I suspect blasting with UVC for a long while, followed by a hot solvent (high-proof IPA? Acetone? TCE?) might do the trick. The UV may or may not bleach the CFA dyes, but enough UV might break the polymer bonds, and the resulting shorter chains could be removed with a solvent.
Fantastic work though, maybe you should make a little side-business of stripping camera sensors! :-)
It would be interesting to strip half the bayer layer away and then take images to showcase the difference it creates in the same image.
Amazing video! I am also trying to remove the Bayer layer of a Pi camera v2. I tried with NMP alone (80 C for 4 hours) and it does not work
Yeah, it turns out they have changed the type of photoresist used in the version 3 cameras upwards, and NMP won't touch it.
If one removed the CFA, and then added a green filter in front of the whole sensor, will the image then be entirely tinted green, or Black and White?
The images would be black and white, since each subpixel woudl receive the same amount of light. You could always color it on software.
what mega big steppers you mounted for mooving such tiny things
Cool stuff! Are the sensors more sensitive even without the microlenses?
Cheers! The response is flatter, and in principle they should be more sensitive, since the filter in a red sub-pixel for example removes green and blue light, but I have not quantitatively measured this.
@@LesLaboratory Would be cool to see what a 50% removed sensor would "see", in order to get a side to side comparison. Thanks for your response!
@@stefanmayer444 I probably still have a half stripped one from testing. It is interesting to see one half in color and one half in B&W. I wonder if there is an actual use for a sensor that does that?
@@LesLaboratory Cool, that would be very interesting to see! I don't know any applications for a half half sensor, other than demonstrational purposes. But what about a striped pattern? An application could maybe be digital night vision. One line would be the sensitive but B&W, and the other would be in colour for the indentification of colors. Traditional night vision tubes are monochrome, but are fairly sensitive. Digital night vision isn't as sensitive yet, but has the advantage of differentiating colors. With your method, it would be possible to strip the sensor only partially in a controlled manner. For such a niche product it might be more economical to use your method, but I'd use a different sensor for that application. Do you know if other sensors behave the same way as the PI cam? Sensitive sensors are expensive to experiment with ;-)
So if you were to put a quartz lens on that M12 mount it should measure well into the UV spectrum?
Yep!
Wow this is a great application to remove bayer filter. I was thinking of using the bluray tracking from a drive or similar setup to use the nitrogen laser output to remove the layers. Still wondering if you used a quartz lens or a high purity plastic lens on the focusing etc..... Some thoughts.... Would be intriguing to use the imaging of x-ray or soft x-ray detection and monitoring etc.
Thanks! I do not think Blu Ray laser diodes would be anywhere near powerful enough, you need brief, high intensity pulses.
The lens is a 'glass' of some description that passes UV. Plastic won't transmit UV very well (if at all).
Sure the sensors will probably detect soft x-rays, but there are better ways of doing that too...
@@LesLaboratory I guess i am not very clear in what I was trying to convey, what I meant was to the use the structured support of the blu ray laser setup to scan and remove the coating using another laser( Uv laser) instead of the existing blu ray semiconductor etc....sorry for the confusion.
Fantastic!
Can I use this as PPFD meter for growing plants??
Hmm, that would be interesting in reflecting telescopes. When using first surface mirrors, absorption shouldn't be much of an issue right?
Sure, dependent on the overcoating on the mirrors, Aluminium should be relatively flat across the spectrum.
Actually block of four represents 4 pixels. The colour data is extrapolated into neighbouring pixels.
This is really wild!
So when are you going to start to sell these cameras? I'd buy this or the pro camera.
Shortly... I never expected that that many people would want one. I will be doing OV5647 sensors to begin with. Depending on yields, I may consider debayering the Pro version.
Hello Les, why not using a monochrome sensor? If the specific version of the monochrome sensor is not available, would you be keen to provide this as a service?
There aren't any monochrome pi-cams. Yes, I might well do. I am trying to get the yield up and am in the process of upgrading the whole machine into something more reliable and repeatable.
@@LesLaboratory Hi Les. As far as I am aware you can purchase monohromatic sensor with global shutter from arducam.
@@gtalckmin have you tried an arducam, and does it work?
@@LesLaboratory I have not yet done so. There is no reason why it would not work as basically they just create the breakoutboard/drivers for Sony (and other) manufacturers. The sony sensor with higher dynamic range would have a better linear response to radiance (which you could test in your integrating sphere).
@@gtalckmin let know how it goes. When I tried it I found the driver support to be poor, and the resolution was not spectacular. I have tried a nice black and white is cam, but again, drivers and resolution. In contrast the support for the picam is vast. All that said, it would be cool to analyse and compile a list of astronomy/spectroscopy capable cams for the pi!
Blimey! Les good work mate really interesting indeed! ;-)
Thanks! :-)
Have you taken any sample images with the filter removed and a normal lens to show that the total pixel resolution is like 4x greater? (32megapixels b&w image)?
I have taken pictures with it, but what you are really asking for, is to pull the raw pixel data off with all the sub-pixels visible. I have not tried this, but the picam should support it...
@@LesLaboratory ive got a python script to look at the normally 'red,green,blue,green' pixels if you take the image from --raw command on pi
Have you ever tried this with a Dslr camera sensor, there's quite a few amateur astronomers tried to debayer their cameras (mostly canon) with varying success. You could probably make a few quid if it works
I have not. I will have to try and pick up a few broken ones off eBay and give it a go though!
@@LesLaboratory did you ever give it a try?
This method also removes the micro lens layer so there will probably be no sensitivity gain in the visible range at all or even reduced sensitivity. Therefore this method might not be suitable for astro cameras.
@@martinbruckner2109 not really to do with sensitivity, its to allow narrowband imaging using different filters to mitigate light pollution. dedicated mono astro cameras are hideously expensive in comparison. i've stripped and re-cased a canon 1100D ready to add peltier cooling to the sensor but at the moment it's still basically a colour camera with the various pitfalls of using colour one-shot imaging
Hi, I am a PhD student at ETH Zurich. I need a camera with as small pixel size as possible to image an optical lattice. The Pi camera v2 would be perfect for this except for the bayer filter. I tried removing it with various chemicals without success. I was wondering if I could pay you to debayer a few of these sensors
Amazing stuff indeed!
Thanks!
It probably is 16MW, I used to play with a KrF laser when not getting a PhD, it had a 1ns pulse length at 10Hz at which point 16MW during the pulse is less than 1W on average.
Makes sense, that would mean output of 10's of millijoules.
Would have been nice if the paper had given figures in MW/cm3, and given some more data on stripping speed, etc.
The N2 Laser puts out just a few 10's of kW as best, and takes a good half hour to strip the filter.
@@LesLaboratory yeah, it is coming back to me, the pulses were up to 1J, so 1GW for a ns. It was later ablation of YBCO to deposit in think films so needed to make rather more of a plumber than your application.
Very cool stuff by the way. I played with webcams and spectroscopes attached to a pi 1 for a museum exhibit a in long time ago and have been meaning to try again with a picam ever since.
You are doing it a lot more thoroughly than I would have.
I think adding a high flow centrifugal fan is not bad idea to get rid of the remaining film while scrapping
Absolutley. I am in the process of building a part 2 for this, with a fan and an upgraded laser system for fast stripping! It takes far too long with a niotrogen laser!
Now that you have the laser engraver, have you thought about doing this project again using the engraver?
The Laser engraver is unlikely to do the job, given its wavelength. To disassemble organics like this you really need UV to break chemical bonds, without damaging the silicon. I have tried with a CO2 Laser, but all it does it heat up the Silicon, it won't do anything at all to the Bayer layer. The photoresists used in these in very tough stuff.
Why the camera is not showing infrared like in ir filter removed webcamera
AWESOME!!!
I'm surprised the sensor hasn't been destroyed by the laser! Any ideas why?
I destroyed sensors initially whist I was working out the damage threshold for the silicon. The pixels latch up white when they are damaged. There are a number of factors to take into account when doing this to get it right, such as peak power, spot size and beam angle.
This is super interesting! Thank you for the video. Have you ever considered offering this as a service? It looks like you could do this without even having do dismantle the camera. I am a fellow creator with a channel for filmmakers… if you where interested to commercialise this and/or try some things with a little higher end gear (I will NOT give you my Alexa for this 😂), let me know
Hi, I would like to do some tests on corona discharge, using a pi camera without bayer filter. Unfortunately I don't have all the equipment you have. How can I do? Can i buy a pi cam from you without bayer filter? I thank you (Daniele)
Yes, I might well do. I am trying to get the yield up and am in the process of upgrading the whole machine into something more reliable and repeatable.
Sell some modified sensors, link your sensors on every DIY-Spectrometer tutorial and cover your losses in PI Cameras ;-)
I was wondering, is it possible to make a quartz lens for better spectral response?
Yes, but is is getting hold of them at a reasonable price that is the issue. I did try a single lens element that transmits a good portion of UV to good effect though.
@@LesLaboratory What about making from scratch?
This so rocks!!
Thanks Jon!
I know I'm late to the party, but are there any plans to do another small production run on this? I am very interested. I sent you an email with details but have not yet heard back.
Ah, yes, I have replied :-)
That's the tiniest and highest tech CNC milling machine I've ever seen....
It would be a thing of beauty to modify it to machine tiny metal parts...
I wonder if a blue 450nm laser would do
Unfortunately you need high peak powers (many 10's of kilowatts) to ablate the Bayer Layer. CW Lasers just heat the sensor to destruction.
You do this commercially ? Or sell the machine ? I was after a affordable old dslr mono camera for my astrofotography trip
There is interest so I will be doing a short run when more Picams arrive. I would be real interested to see ho they perform in astophotography applications...
@@LesLaboratory For astrophotography i would prefer old full frame Nikons, Canons. A bigger sensor with 12 or 16mp. There is a ton of these used cameras on aliexpress. Maxmax already sells new modified Sony A6000 but they jump from $600 to $2000. For $4000 you could buy a new cilíndricas full frame astro cam like ZWO. There is nothing on the cheap camera niche. It is a shame there is no new replacement cmos for these cameras. Sometimes you find some for $14 but it is a limited supply
An older big sensor converted to mono with a cheap chinese filter wheel could do marvels
Would the removal be beneficial for light gathering? I mean the Laser stripped the microlenses away as well...🤔
I came here looking for cheap UV cameras.
There are monochrome modules available for
I was interested in stripping a genuine picamera, because I wanted a sensible spectral response out of it.
Third party cameras (at least the ones I tried) have poor driver support on the Pi too, and that was another factor.
The other reason, was to see if it could even be done, as there are applications beyond what is demonstrated here.
Have you used the FSM-AR0144, and is it any good? I am always keen to try new cameras!
What power did you use?