Pictures taken with my 1D Mark IV show a blue horizontal line at the bottom from ISO 400 onwards.

[numenius]

Pictures taken with my 1D Mark IV show a blue horizontal line at the bottom from ISO 400 onwards. So far nobody has been able tot tell me the cause of this problem. The body is still quite new (15000 pics taken).

[rfdesigner]

Can you post a link to an example please...     Ideally in RAW format..  if we think it's sensor it's best not to muck about with the data.

Edit:   oh..  a picture's appeared..   internet must be slow today

[rfdesigner]

I think you have a dud column/row in the sensor..   seeing the RAW file (.cr2) would be very helpful.  I have a tool that can open it and view as original monchrome (before debayering) which would tell us a lot more.

[ebiggs1]

Try a different CF or SD card. Card may have gone south on you.

[TCampbell]

I agree that it looks like a column of photosites is not working.  But... I'm curious as to why this would only happen at ISO 400 or above. 

The sensor is covered with photosites -- they are monochrome.   On top of the photosites is the "bayer mask".  The mask is a checkboard pattern of red, green, and blue "lenses" which makes the photosite beneath only sensitive to that particular color.  They are group in clusters of 4 photosites in a 2x2 matrix.  Two of the four squares (in opposing corners) are green.  One corner is red.  One corner is blue.  If you were to lose a column which happens to be along the red/green column (as opposed to the green/blue) then you'd lose red sensitive and half the green sensitivity in that column.  The result would be that this column is flanked by two "working" columns which are full of green/blue photosites... hence the teal blue look.

I do not know if this will work for you, but oddly... you can actually sometimes "clear" stuck pixels (but usually this is an indvidiual stuck pixel... not a whole column of them) by putting the camera into manual sensor cleaning mode (and you may have to do this 2 or three times).  While the normal point of the self-cleaning mode is to flip up the reflex mirror and open the shutter so that you can inspect and clean the sensor (really the front-most filter because you can't actually touch the real sensor), it also neutralizes the charge on the sensor (or so I've read).  This charging and discharging of the sensor a few times can often get "stuck" pixels to start working again.  BUT AGAIN... I've only ever seen this used on individual stuck pixels... never a whole row.

[rfdesigner]

I really want to know if we have additional false signal, missing signal, or cross contaminated signal between two rows.

The ISO 400 and above could come from different bias for high ISO..  The ratio of well depth to readout noise on these sensors is sufficiently high that I would not be at all suprised to see additional in pixel gain added as an option, either by additional circuitry or by adjusting bias to improve gain.

If the damage is rock solid, it is possible that the OP could process around it..  but it would be a pain to have to do that, I don't know if the long shot dark subtraction algorithum would help.  The 5D has the High ISO multi-shot noise reduction algoritum.. if the 1DIV has that too it could help so long as the damage is just adding an offset to the affected pixels.

The idea of a highly charged spek of dust contaminating things is possible and worth doing sensor cleaning for, I'll remember that one.

[diverhank]

Have you tried turning on/off the high ISO filter and see if it makes a difference. The fact it only shows up at higher ISO shows that it's an ISO artifact, not a sensor problem. Anyway, just a guess on my part.

[numenius]

I tried to upload a RAW file but didn't succeed. Thanks for all the advice and options given. I'll try them out the coming weeks. 

[rfdesigner]

You can drop one on dropbox or any other hosting site (I use dropbox, it works, no other connection)  then post the link

i.e.

https://www.dropbox.com/s/bg1t8ue3av6r0pc/IMG_9685.CR2

[numenius]

https://picasaweb.google.com/103470919743498275798/20140201?authkey=Gv1sRgCMCe_LzY2cih_QE#5975415047...

I hope this RAW picture will help, so far no solutions offered did work.