Stuck Pixels - Becoming less of an issue.

March 22, 2001

Another trend is being enforced, at least for in the early cameras. Pixels DO go bad over time. I recently got this Email:

Hello, I'm Alain from Angouleme in France. I've bought an Olympus E10 camera in December 2000 here in Angouleme. I've read your comments about this camera before and I was alerted about the stuck pixels problem.

No problem during 3 months. I didn't think about. But this week I see a picture: two bad pixels on 2240 x 1680 pics (one at x: 328 and y: 522 and another at x: 314 and y: 1054). In each case a square of four pixels exactly, black out and white in. All my pictures are like this now... Stamp with Photoshop is now inevitable for correct pictures... In France this camera is very expensive. I'm very disappointed with Olympus. My last camera was an Olympus 2500 L and work very well.

Alain Auzeral

February 3, 2001

If Phil Akey's Digital Photography Review Olympus SLR Forum is any type of indicator, the E-10s leaving the factory today are a lot less likely to have a pixel problem that those shipped in December and early January.

The past week was notable because so many new E-10 owners posted messages proclaiming their cameras to be free of pixel problems or any other defect for that matter. What a change from the postings of just six seeks ago that were punctuated by so many new owners receiving their second or third E-10 because of stuck pixels.

I wish I could say that I didn't know of any recent pixel problems, but just as posting after posting showed a very positive quality trend, I got an Email from a fellow in Monterrey, Mexico, who wasn't as lucky. His new camera did have a stuck pixel, and he was pondering the difficulties of exchanging it over just living with it. I had the impression that he purchased the camera from a vendor outside Mexico and that the exchange logistics would be pretty messy compared to the service we enjoy here in the United States.

From my somewhat distant vantage point, I have concluded that while the pixel issue may not be completely resolved yet, it does appear that Olympus has made notable gains in their quality process. This should be good news (and perhaps a confidence builder) for folks who are still waiting to see if Olympus can cure the E-10 start-up blues.

December 12, 2000

I am not alone. I have been fortunate enough to be in contact with the web masters of all four major digital camera web sites. All four have an E-10 in house. Two of them have received an E-10 with at least one hot or completely stuck pixel.

Jeff Keller at digital camera resource page just received his personal E-10 only to discover that he had two hot pixels. In his case, they start to misbehave around 1/40 of a second and by one second are fully on. You can see them in his posted samples.

The other web master received a production E-10 on loan from Olympus for his review! I would have expected Olympus to hand pick a camera for such high profile scrutiny..... makes me wonder. He is still working with the camera (and presumably Olympus) and has not yet commented publicly.

My advice, from one E-10 user to another is check your camera with a series of lens cap shots as soon as you get it, and then check it again every few hundred shots. I'd start with the lowest shutter speed you consider to be in your "normal" working range - could be 1/15 of a second, or 1/2 or 1/1 - you decide. Then I'd recommend a shot at 1, 2, 4 and 8 seconds. For me, a truly "stuck" pixel is unacceptable in a camera of this caliber. A hot pixel that shows up any place inside my normal range would be unacceptable. A hot pixel that shows up in exposures that are longer than I generally use are tolerable - this is a matter of tolerance and pain - everyone will have their own threshold. Finally, hot pixels at 10, 20 or 30 seconds don't concern me. I just don't think the CCD technology has matured to the point that it is reasonable to expect flawless operation on extremely long exposures.

I have some speculation at this point that there is a genuine yield problem with the 4MP CCD Olympus is using. I wouldn't be surprised if there is fundamental connection between the elevated noise across the CCD (which is I feel is generally over stated by most who choose to comment on it) and the hot pixels that seem to pop up frequently beyond 1 or 2 seconds. But this isn't my field, so it's just as I said, speculation.

Having said all this, I still believe that an E-10 with a good CCD is great value, and it's worth taking the risk to get one.

December 4, 2000

So far the new camera has not developed any pixel problems. I have come into contact with a few more people who experienced a pixel problem or were dissatisfied with the image quality of their camera. In most cases they have been able to exchange the camera. I'd also have to say that considering the number of cameras Olympus must be shipping, the numbers probably aren't statistically alarming.

November 29, 2000

From the small amount of research I've been able to do, I know that I didn't have the only E-10 with at least one stuck pixel. Based on Email I've received and a couple of follow up posts on Phil Askey's dpreview Olympus Forum I know there are users in both the US and UK that experienced a similar problem. Are the number statistically significant? I just don't know. Do statistics mater to the person who owns a flawed $2000 camera? Not a darned bit!

November 28, 2000

Once again, Norman Camera comes through - my replacement camera arrived today as expected.

I have a few observations to share and a bit more information about the camera I returned.

First of all, the new camera, with s/n 10239xx is using firmware 42-0119. My last camera, s/n 10104xx used firmware 42-0116.

This time I'll keep some base line images. I shot 4 black .tiff files 1/200, 2 sec, 4 sec, and 8 sec. No defects were noted at 1/200 or 2 seconds. At 4 seconds, I have one white dot and one gray dot (taking up 4 pixels of the bit map each). At 8 seconds both of these dots are white (and they are at the same coordinates in both files). Based on this. I'm assuming that I have two locations on the CCD that are inherently hot and may be future failures. I plan to run the same 4 exposure tests every 500 or so pictures and compare.

It has been pointed out that since the filtered CCD data is interpolated, it is likely that only one pixel of the CCD is misbehaving, but it effects 4 pixels of the output file.

When did the bad pixel actually show up on the first camera? I believe the answer is some place between 450 and 550 shots. It's just my luck that I deleted a bunch of junk test files about that time. I didn't actually discover the problem until I had taken over 1500 shots.

Finally, there is this little curiosity. The new camera has two small, round, white stickers on the box, one just to the left of the serial number window on the side, and the other in the upper right hand corner of the top flap. From my years in electronic manufacturing, these sure look like inspection or upgrade markers to me. The box of the first camera did not have any kind of sticker or marking.

So, was it a fluke or is it a trend? It's too early to tell. I do know this; I'm not the only person to have an E-10 with at least one bad pixel. My advice is to check your E-10 now and every few hundred shots until your confidence is high that you're not likely to have problems.

November 27, 2000

I checked my E-10 pretty carefully when I got it. One of the first tests was to be sure there were no stuck pixels. This is nothing more than a manual exposure of the lens cap at something like 1/200 @ f11, written out as a full size .tiff file. Everything looked good.

About 20 days and 1500+ images later I noticed a white dot where there should have been blue sky. At once I checked a couple other images and sure enough, there it was, again and again. A quick check of the coordinates confirmed that this wasn't just one pixel, but four adjacent pixels. Here's how it looks.

This was the image I was looking at. The frozen rose from the Lake Side Rose Garden series. Even reduced to this small size, you can actually detect the defect on a good monitor.

In a .tiff file, at about 800% magnification, the four pixels actually look like four adjacent white squares. In a .jpg file however, you get something that looks like the image above. The sharp contrast between the white pixels and the surrounding area sets up a combination of compression artifacts. It's sort of like getting a dark frame to show off the defect, and then setting it in a fuzzy image of a spiral galaxy. To see the exact defects, take a look at any of the images from the Rose Garden. Look 820 pixels from the left side, and 1255 pixels down from the top.

That's pretty much the bad news. Now the good news.

Over the years I've learned not to send in the warranty card right away on anything. Especially electronic things. Doing so generally eliminates any discretionary powers your merchant might have. I'm also one of those people who keeps all the original packing materials - so when I sell one of my toys to buy a new one, someone gets a pretty complete package.

All of this was true for the E-10. Now all I needed was a customer oriented merchant - and of course I had one - Norman Camera. I explained the situation over the phone and asked if they could just exchange the camera since I'd had it less than 30 days. The answer was simply "Of course." They very matter-of-factly checked stock while I was on the line, confirmed that they had a replacement, and arranged to send it out at the same time I sent the defective one back to them. The two cameras are crossing somewhere via UPS as I'm writing this. I expect to have my new one Tuesday, Nov 28.

Situations like this are EXACTLY why Norman Camera is my preferred supplier of photo technology products. In fact, this should be a case study in the way excellent customer service is carried out.

I'll update this section when I get the new camera. Until then, the lesson for other E-10 owners, is to be sure to check your CCD a few times during the first couple weeks you own the camera. I'm confident that my first E-10 was the exception, and not an indicator of things to come with the new 4MP CCD Olympus is using. But I'll be checking my new camera frequently, just to be sure.