The Pixel Sweeper
The Pixel Sweeper
Finding, preparing, testing, and using brushes to clean the sensor on your digital SLR.
My original article on the Pixel Sweeper, written in March, 2005, got me a lot of feedback. I learned a quite a few things about dust, brushes, surface effects, cameras, and even static electricity in the process. I have also heard from hundreds of people who have used the technique, and only two or three reported any problems; these were due to liquid contaminant making its way onto the sensor, either from propellant or lubricants in the camera's mirror mechanism. I've also done a quite a bit of experimentation myself, and discovered a sensor-brushing trick that's both simpler and safer than the one described in this article. Therefore, this update -- actually, rewrite -- of the article is long overdue.
There are at least two well-attested sensor-cleaning methods that are known to work on dust that simply refuses to go away by blowing at it. One is wet cleaning the sensor with a non-abrasive, lint-free tissue wrapped around a spatula of some kind, and moistened with methanol or lens cleaning fluid. There are several variants of it, including specially prepared, individually packaged, and commercially supplied Sensor Swabs, and the do-it-yourself technique commonly known as "Copperhill method". The other is the "Sensor Brush" -- a commercial product by a Canadian company called VisibleDust.
Wet cleaning is probably one of the only ways to safely remove gunk that's actually stuck to the sensor. However, it does have its risks: while a flexible spatula and a square of optical paper cannot put a scratch on the filter in front of your sensor, a mineral particle caught in it can. While you're certainly not supposed to push hard on the spatula, you're still exerting pressure. It's not likely that you'll scratch your sensor this way, but it is possible -- and therefore, in my opinion, less "invasive" sensor-cleaning methods should be preferred whenever possible.
The other, newer technique, glowingly endorsed by some high-profile websites like Luminous Landscape and Rob Galbraith, is the Sensor Brush method. VisibleDust will be happy to sell you a kit for about US$100, give or take a few dollars, which contains a couple of brushes. They claim that these brushes have the interesting property that when you blow air through them, they acquire a static charge, and become very effective at picking up dust. Lots of people are using them and are getting excellent results. In fact, I believe that this method is at least as effective and inherently safer than wet cleaning, for removing dirt that isn't actually stuck to the sensor. The only thing that some people aren't entirely happy with is the price... and some features of VisibleDust's marketing, which is rather heavy on hard-to-verify claims and Fear, Uncertainty, Doubt.
In this article, we will be looking at one particular sensor cleaning method: brushing it with a suitable brush. When done right, it is highly effective, very low-risk, and inexpensive. However, as always when getting near your sensor, there is an element of risk involved, so if you're not absolutely certain you can read, understand, and follow these instructions, don't do it.
WARNING: If you intend to clean your sensor by following these instructions, be sure to read through
of the article. Using just any old brush on your sensor without first making sure it's clean will give you a major headache that will require lots of swabbing to cure (I would suggest wet cleaning with distilled water, since sizing may not dissolve in methanol.)
Dust is a fact of life in photography, whether digital or analog. Dust spots negatives, gets into slide scans, makes its way onto and into lenses and viewfinders, and ends up on sensors. Dust is everywhere. Unless you do all your photography in a microprocessor fabrication plant, it's impossible to get rid of it. The best you can do is manage it. To be able to do this effectively, it helps to understand a few things about what dust is and how it behaves.
Dust comes in a tremendous variety of particle sizes, from clusters of a handful of molecules to fluffy bunnies that float lazily in the sunlight. We're most concerned with medium-sized dust: particles big enough to be seen if they land on the sensor, but small enough that surface adhesion will keep them there when you blow air on them.
Unfortunately, this size dust is still very, very small -- far too small to be seen with the naked eye, and small enough to happily make its way through joints and cracks in the lens, camera, and shutter curtains, especially as focusing and zooming pumps air in and out of the box. Yes, even through weather seals.
Dust levels within connected areas will equalize over time. That means that since your camera isn't sealed and air is going in and out of it, it's only a matter of time -- and not a long time at that -- before the dust concentration inside is the same as the dust concentration outside. Only the largest particles will get stopped by the seals and joints in the camera, not the ones that make those little smudges on the sensor.
Dust floats. Dust is so small that you can almost ignore gravity when considering how it behaves. It floats on the air just like plankton floats in water. Yes, in a still room it will eventually settle down, but this happens very slowly. The visible dust that gathers on furniture (both kinds, the light kind that collects on dark furniture and the dark kind that collects on light furniture) is much larger than the kind of dust we're interested in, and settles much more quickly. Even so, it'll take the better part of a week for that kind of dust to produce a layer on a surface you've just wiped.
Dust clings. When a dust particle touches a surface, it will stick to it. This is known as surface adhesion. If electrostatic charges are involved, dust will actually get sucked to a surface from some distance. However, I believe that electrostatic effects are of very minor importance when considering camera sensors (and cleaning them). Unlike many people believe, the imaging sensor is not "charged" most of the time, and when it is charged, the charge is tiny; too weak to attrack dust further than a couple of millimeters. Similarly, I believe that the VisibleDust theory that dust sticks to their brush through electrostatic attraction is false -- I'm pretty sure it's a matter of simple surface effects.
The biggest source of dust is... you. We people are constantly shedding dust -- dust we've collected outside, dead cells, dead hairs, dead mites, live mites, fibers from clothes we're wearing, and so on.
What does this mean in practice?
Don't sweat about changing lenses. Most of the dust inside your camera would have made it in even if you never changed a lens. And it really doesn't matter which way up you hold your camera: the time the lens is off the camera is much too short for dust to have time to float down with gravity.
Don't get too anal about eradicating dust from your cleaning area. Since dust levels tend to equalize, even if you do your cleaning in a perfectly dust-free environment, it won't take long for dust to make its way right back in. Cleanliness is important, but it's enough that your cleaning area is significantly less dusty than your regular shooting environment. In particular, try to pick an area as free of mineral grit as possible: for example, don't clean in a room with a vent that opens directly to a busy city street.
Don't get too close to the camera when cleaning it and if there's an air current, keep downwind of the camera.
Dust must be seduced, not forced. To get dust off a surface, your best bet is to provide it with another surface it likes better. Dust isn't particularly fond of very smooth, shiny surfaces. Fortunately, your sensor is extremely smooth and shiny. Therefore, most dust will be happy to abandon it if you present it with a surface that's less smooth. Such as a bunch of very thin, soft fibers on a brush.
Find a way to get the dust off your brush. Everything in nature tends to equilibrium; so too your cleaning brush. If you do nothing, it will eventually accumulate so much dust that for every particle lifted on, one drops off.
While a perfectly dust-free environment is not attainable by anything you can do at home (and indeed wouldn't do you much good even if you could get it), you can do some things to reduce the dust content in the air significantly. You should do this no matter what your cleaning method is, since it will make a significant difference to the ease of the operation, the results, the time the results last, and the risks.
Pick a room that's free of textiles and drafts. Textiles shed dust and air currents distribute it. Reduce both and dust content will go down too. Vacuum the room with a HEPA-filter equipped vacuum cleaner, and you will make a significant dent in the kind of dust we're dealing with here.
Wear a lab coat and hat or something like them. That is, smooth synthetic or long-fiber cotton clothes, rather than wool or anyhthing at all fluffy. A long-sleeved shirt and shower cap (or even baseball cap in a pinch) will get the job done too.
Do your cleaning in the bathroom, and scrub the air with a (cold) shower. The shower will effectively circulate the air and scrub dust out it. However, as the water dries, dust carried along with the water will get released right back into the air again, so if you use this method, you only have a fairly short window to work in. Obviously you should eliminate all sources of dust from the bathroom first, as far as possible. Such as, don't run the clothes dryer.
Use a dust hood. You can make it quite simply by taping some plastic sheeting to a clean table, and making holes for your hands to work in it. If both the sheet and the table are clean, there will be significantly less dust inside than outside, long enough for you to get the job done anyway. For the ultimate home-made cleanroom, combine this with inflating it with air from a HEPA-equipped vacuum cleaner that you only use for this purpose. Don't try it with your regular vacuum cleaner; HEPA or not. If you clean cameras a lot, you might actually want to consider buying a vacuum cleaner just to provide you with clean air, but most of us would probably consider this overkill.
Preparing your sensor cleaning brush
A sensor cleaning brush has a few particular requirements. Luckily they're not too difficult to meet:
It should be the right size and shape. In other words, flat, square, and small enough to fit easily into the chamber, but large enough that you can cover the sensor in two parallel sweeps.
It should be soft and uniform. Synthetic brushes (nylon, taklon etc.) are the best bet, since the fibers are highly uniform, very thin, and taper. They're also easy to clean, since they absorb very few materials. The thin, tapered fibers give dust lots of surface to stick to, while still being slippery enough that it's reasonably easy to dislodge the dust once it's off the sensor and on the brush.
It should contain no contaminants. Most artists' brushes are sized -- treated with a starchlike substance that holds the brush in shape for transport. It is essential that you get the size off before using the brush on your sensor; otherwise you will leave streaks that only wet cleaning will remove. Cosmetics brushes are usually not sized, so if you can find one that's the right shape, size, and material, great.
Selecting the brush
There are two sources of brushes you'll want to consider when looking for a brush that meets the above criteria: make-up (cosmetics) brushes, and artist's brushes. Alternatively, you could buy one that's sold specifically for the purpose; Copperhill sells brushes at a pretty reasonable price, for example. Many are sold on eBay as well. No matter where you get it, though, if you're not perfectly sure about the suitability of the brush, be sure to subject it to the "filter test" described below.
A soft nylon cosmetics brush. I understand this is called a "mask" brush.
In my opinion, cosmetics brushes are preferable, because they're always soft (ladies won't drag something that feels like straw across their cheek) and they're never sized. However, in my brief survey of cosmetics counters around town, I found that cosmetics brushes meeting the above criteria don't grow on trees. There were a few that were about the right size and shape, but they were made with hairs pulled from the behinds of various exotic animals, and a few that were the right material but the wrong shape and size. Eventually I found one that seems to fit the bill: a "mask brush" from a place called Make Up Bar. It passed the Filter Test (see below) with flying colors on the first try, without washing.
A couple of synthetic artist's brushes -- yep, the black one is synthetic too, although it's cleverly faked to look like sable or something like it. Being a Canonian, I figured the black one with the red ring had got to be better, but I was wrong -- it had so much size on it I had to wash it five times before it passed the "filter test" below. The other one had barely any size on it and cleaned up with a single pretty light wash. If shopping for artist's brushes, look for a brush that's soft at the store, or at least one where the bristles "snap loose" with a gentle push.
On the other hand, there's a far greater variety of artist's brushes in the right material, shape, and size available. Go to any halfway-decent art-supply store, and look for synthetic brushes. You'll find easily a half-dozen varieties that look approximately right. The trouble is, in all the art stores I visited, the bristles are always sized together for transport. For an artist this doesn't matter, since the size dissolves when you dip it in water -- but for this purpose, it's extremely important to get all the size out, not just most of it. This is certainly possible, but an unsized brush is clearly preferable to one with size in it.
So, if you can't find a make-up brush that fits the bill, go ahead and buy a suitable artist's brush. However, do try to find one that has a minimum of size on it -- it should easily "snap" open and become fluffy and soft right away. I made the mistake of trying one that was pretty damn effectively sized, and it took me five careful washes to get all of it out.
Out of curiosity, I tried getting a super-macro of one of my brushes. Here it is. Notice the fine, uniform-thickness, pointy fibers. Just like on the official version...
Cleaning the brush
If you bought a brush that you suspect has or had size in it, or was handled in the store by people with oily fingers, the first thing to do is clean it.
- Put a drop or two of dishwashing detergent (for handwashing dishes, not the machine kind) onto your palm, add some water, and roll the brush in it.
- Work it with your fingers, first against your palm and then under running water.
- Rinse very thoroughly under a tap, pointing the brush up towards the stream of water so the water flows freely through it.
- Repeat until the brush no longer feels slimy. It's supposed to feel slippery -- nylon is slippery -- but not slimy. There is a difference that's quite easy to tell.
- If you live in an area where the water is "hard" (has a high calcium content), rinse the brush in distilled water.
- Allow to dry overnight before doing the "filter test."
The Filter Test
VisibleDust explains how they tested various brushes for size residue: by brushing a glass slide 1000 times and then examining it through a microscope for residue. The Filter Test does more or less the same thing. If you have a microscope, I highly recommend you repeat VisibleDust's method, but if you don't, I'm quite certain that my simpler test gives you a result that's reliable enough.
This is how a clean filter should look -- a colored, pretty dim reflection, and no markings. Start with this. If it still looks like this after you did the brushing, your brush is safe.
Pick a multicoated filter. Surely you have one? If not, this is a great excuse to buy one that you've been coveting. Don't worry, you won't damage it. At most you'll get some water-soluble gunk on it that'll wash off.
Thoroughly clean it with your favorite method. I use dishwasher fluid, work it with my fingers, then rinse it under running water, using the water stream to "wipe" the filter clean; then I dry with a fresh linen towel (one that's been washed without fabric softener).
Examine by looking at the reflection of a lamp. Multicoats are nice because any gunk on them will shine like a beacon -- it'll affect the reflectivity of the coating. If your filter's clean, you should see a very dim, greenish, orangeish, or violetish reflection of your lamp, with perhaps a few marks from an imperfect cleaning job.
Take your brush, and brush the filter vigorously back and forth about 250 times. This makes for a total of 500 sweeps. Hell, if you're nervous, make it 500 back-and-forth sweeps, so you'll have the same 1000 as VisibleDust is doing.
Examine the filter again. If there was size or oil on the brush, you will see something that looks like a pattern of very very fine scratches, clearly visible as more reflective against the nearly-non-reflective background of the multicoated filter. They're not scratches: they'll come off if you wash the filter again. If there are no such marks, you can be certain that the brush isn't leaving anything nasty behind. Your sweeping was much more vigorous than you'd do for your sensor, so this test is more demanding than at least a couple of hundred times of cleaning your sensor -- a quite a lot.
If your brush failed the test, wash it again. Repeat until it doesn't fail. My first brush failed four times, but came out clean on the fifth. My second one passed with flying colors after the first wash. My first brush was quite solidly sized at the store; I made the mistake of thinking it would come off easily. You don't need to make that mistake -- just pick one with no or very little size on it.
If you see something like this, wash your brush again (or discard it). This is gunk deposited onto the filter by the brush. It was a bit tricky to photograph, but is even clearer to see with the naked eye.
Storing the brush
Once your brush is selected, cleaned, and tested, you'll want to keep it clean. The best way to store it is in a sealed container. Failing that, a clean zip-lock plastic bag will do the job fine. The main thing is keeping it away from dusty air as far as possible.
Good to go!
All right, time to clean the sensor -- assuming you weren't scared off by the above paragraph. Here's how to proceed.
First off, use a freshly charged battery for all your cleaning fun. You do not want the camera to shut down in the middle of it and snap the shutter and mirror down on your brush.
Before you begin, take a reference shot: point your camera at something white, stop down to f/22, throw it out of focus, and shoot a frame, and never mind camera shake. Then examine it in Photoshop (Auto Levels optional.) I strongly recommend that you hold off cleaning if there are only a few boogers on it: if you get really anal about it, you'll never have a quite a clean sensor, you'll be terribly stressed every time you change lenses, and you'll be cleaning the camera way too often. If there are more boogers than you care for but still not too many, I strongly recommend the approved Canon method: just blow air on the sensor with a blower bulb (not canned air). After all, there's always a risk of something going wrong -- for example, if something startles you while you're in the middle of cleaning it, you could make a violent movement and drop the camera, or something along those lines. So, treat the sensor like your lenses -- don't clean it if it ain't dirty.
My 20D sensor pretty much as it was when I got it. Sorry about the funky lighting effect...
1. Get your tools ready
You will need the following tools for your sensor cleaning operation. Prepare them before you start.
A clean surface. A fresh sheet of photo paper will do the job fine.
A brush. We covered that already, didn't we?
A hand pump or blower bulb. You'll need it to clean the chamber and do the first round of sensor cleaning.
A clean stainless steel table knife, not necessarily sharp. Wait for it...
Lens tissues and lens cleaning fluid or a clean microfiber cloth. For cleaning the knife.
2. Clean the air (and your tools)
- The first thing you want to do is stack the odds in your favor against getting more dust into the camera than you remove from it. (Re-)read the Managing Dust section above, and take whatever steps you find most suitable for your circumstances to reduce the amount of dust in the air of the room in which you'll be doing the cleaning.
- Lay out your tools and your camera on the clean surface.
- Wipe the table knife carefully with your clean microfiber cloth or (preferably) a lens tissue moistened with lens cleaning fluid.
3. Clean the chamber
- Pop off the lens from your camera, but don't go into sensor-cleaning mode (yet).
- Take your hand pump or bulb blower, and carefully puff your way around the chamber. Make sure not to touch anything (in particular, the mirror or the focusing screen).
- Look through the viewfinder. See any dust? If yes, puff again. Dust still there? Repeat a few times.
- If your viewfinder is reasonably clean, skip the next steps and go to "clean the sensor."
- Turn your back on your camera, hold your cleaned table knife horizontally in your left hand, and strike the brush's head against the flat sharply several times, so that the fibers snap back and forth against it. Not like you were cutting the brush; the other way around. This will very effectively dislodge any dust caught in the brush -- more effectively, I believe, than canned air, with no risk of contaminants.
- Sweep the focusing screen once with the brush, making sure to hit the area where there was dust.
- Repeat the above two steps until no more dust is visible in the viewfinder, you're convinced that the dust is under the focusing screen, or you get bored.
Note: If you want to brush all of the chamber clean, use a different brush than the one you intend to use on the sensor. There are lubricants in the moving parts in the mirror box, which could make themselves onto the sensor via the brush. Be very careful if you have to brush the mirror; it is far more sensitive to damage than the sensor (or anything else you'll be encountering).
4. Blow the sensor clean
- Make sure that you're using a fresh battery, and put the camera into sensor cleaning mode.
- Blow the sensor clean with your hand pump or bulb blower. Hold it at an angle and blow towards the corners.
- Take a test shot (see Test below). If the sensor is clean, stop here.
5. Brush the sensor clean
- Clean your brush as above, by striking it against your cleaned table knife. Again, keep away from the camera to disperse the dust somewhere else than right back in it.
- Sweep the sensor from side to side once.
- Clean the brush against the knife again.
- Sweep the sensor from side to side in the other direction.
- Repeat the above steps until you have fully covered the sensor going both directions.
- Pop on the lens and take another test shot as above.
- Examine the picture. Is it clean enough to satisfy your requirements? If yes, you're done. If not, repeat "Clean the sensor" above. If you see some specks that just won't budge, they're probably stuck: you'll have to wet clean the sensor to get rid of them.
This is clean enough to satisfy my requirements. Someone more anal than me would probably have gone one or two passes more.
Tip: works great on slides!
As anyone who's ever scanned film on a scanner without digital ICE knows, dust is even more of a pain with slides and negs than with digital SLR's. It's plumb impossible to keep those films clean, no matter how carefully you store them. The Pixel Sweeper works great on them, too: charge it up as usual, gently sweep the neg or slide on both sides, and voilà, pristine -- well, almost pristine, anyway.
After. Still a few specks left, but pretty damn clean. No marks or scratches either. A slide is much softer and easier to damage than the glass on top of the sensor, and dirt shows up much more clearly too, since it's actually on the image plane rather than a few mm on top of it, so this gives a pretty good idea of how safe and effective the method is.