The next instalment of my attempts to get my Fuji FinePix E900 repaired that started with the Venetian Blinds Fault.

Oh for crying in a bucket!

I got the camera back on Thursday (27/11/2008) and this time it wasn't making any nasty noises (apart from the default beeps and fake shutter noises which I quickly turned off). I had plenty I needed to do, and it was miserable outside so I wasn't going to get a chance to test it, but on the face of it things where now working - it was taking photos with no blatant problems.

Today (29/11/2008), I decided to redo the noise profiles since it has a new sensor which may be a little different to original.  Also, I have taken to doing noise profiles with a gamma of 2.2 (instead of linear) as this results in the brightness levels mapping better and hence more accurate noise removal from shadow areas (where most of the noise is).

After taking a load of shots of the noise target out of focus (prevents stuff in the image being falsely detected as noise), I noticed in the shots some bright (hot) pixels, and this was at minimum ISO (hot pixels increase dramatically as sensitivity and exposure time are increased).

Hot and cold running pixels

Hot (and cold / dark) pixels are nothing unusual. A certain amount occur on all sensors, and with time sensors grow additional ones. What normally happens is that they are mapped out (marked as bad so that the software in the camera guesses a suitable replacement colour for that pixel from the surrounding ones) in manufacturing so that on a new camera there are no noticeable ones.

My Canon dSLRs appear to also automatically re-map hot pixels when the sensor clean (mirror and shutter is locked open) is activated, so where they have developed new bad pixels, I just run a sensor clean and then it's fine. Fuji seems to think that their cameras need sending back for bad pixels to be re-mapped.

My original E900 had only 1 hot pixel and 3 cold pixels - not enough for me to loose sleep about.

At ISO80 (the lowest setting where fewest hot pixels should be visible), I count 6 hot and 9 cold pixels (my bad pixel mapping software which works from RAW counts 6 hot and 3 cold, but missing the cold ones may be due to the way it works and the RAW Fuji image being rotated). This was done by taking 10 slightly under exposed photos of a white surface out of focus (so anything that could be mistaken for a hot pixel gets blurred out). I then averaged them together (removes anything that is not consistent - like image noise), and applied some basic thresholds before counting. The 3 images are available for those that are interested:

• Average of 10 shots at ISO80
• Hot Pixels
• Cold Pixels

Back yet again....

I will be calling Fuji yet again on Monday morning to organise sending the camera back yet again (third time) for re-mapping the pixels. I think re-mapping the bad pixels should have been done as part of a sensor replacement since the new sensors are obviously going to have different defects to the old.

Why haven't they done this? Why don't they have a trick like Canon dSLRs for re-mapping the pixels within the camera?

Update

It's not over yet - read the next part.