Canon 5d2 high ISO problem

Hello everyone

Does anybody noticed obvious image split - like in the middle of frame while shooting at high with ISO setting in dark / night scene? I have noticed this at ISO 3200 but very pronounced at 25600. Has anyone got same problem or is this appearing only on my camera? I have latest firmvare installed. Sample image was shot with 24-105mm lens. I noticed that with other lenses so problem has to be somewhere else.









This "image spilt" also appears at daylight when using high ISO above 3200.
Any thoughts appreciated.
Thanks

never seen that. Do you have a filter on the lens?

I would also upgrade the firmware just in case it is corrupted.

If not I would send it for service...........

I had the same problem with my 5D Mark II.

I send it to service (Best focus adjust for Canon 24-70) and it was coming back with exactly the same problem as you have here.

I send it again to service and the problem was resolved. They haven't change the sensor. They have only calibrated it.

--
http://photo-design.at

I saw this line even in video (above 1600 ISO).

This is an example of my 5D2 problem:



--
http://photo-design.at

Wow--that is messed up. I own two 5DII and have occasionally shot at even a higher EI than your posts and never seen anything like that. I'm guessing your camera needs to go into the shop--be sure to include a printout of the images.

No, there wasn't any filter on lens and also this split image appeared with other lenses. I recently had camera body in service and that seems to be the result of repair. Thanks all for replies. I ll send camera for calibration. Also will try to update firmware if that should help it

I don't think the firmware update, or lack thereof, has anything to do with this. Ditto with filters or shadow cast on the lens--the line is too sharp and linear.

I've been shooting with the 5DII cameras from their introduction, always updating the firmware as it came out, and never saw anything like this camera display.

Good luck!

mariuss said:

I had the same problem with my 5D Mark II.

I send it to service (Best focus adjust for Canon 24-70) and it was coming back with exactly the same problem as you have here.

I send it again to service and the problem was resolved. They haven't change the sensor. They have only calibrated it.

Click to expand...

and

Juraj Lacko said:

No, there wasn't any filter on lens and also this split image appeared with other lenses. I recently had camera body in service and that seems to be the result of repair. Thanks all for replies. I ll send camera for calibration. Also will try to update firmware if that should help it

Click to expand...

It sounds to me like the Canon Service Center accidentally left these cameras in some kind of debug mode.

It was repaired ... the same story as mine.

Well Canon, you have to check your service quality. It doesn't bring a good light upon you. (I know Canon won't hear this, but I had to say it, though).

Sorry, but a firmware update will not bring anything, I presume.

Juraj Lacko said:

I recently had camera body in service and that seems to be the result of repair. Thanks all for replies. I ll send camera for calibration. Also will try to update firmware if that should help it

Click to expand...

--
http://photo-design.at

Probably the 5D2 sensor is combined from two different pieces. Severe underexposure reveals the stitching. Something similar happens to the Nikon D3 sensor with the differences that the parts are four an that is also observable when the green channel is saturated. This from my D3 sensor:





--
Panagiotis

My understanding is that Canon uses single "pieces", actually single exposure of single piece, where other manufacturers have had to make multiple exposures. This has to do with the lens used in the chip manufacturing process. I'll defer now to others who have more knowledge of this.
--

sample picture was taken @ H2 - 25600, this split image i have noticed alredy at iso 1600 and higher I will contact repiar centrum and ask them about it. This is thing I never noticed before repair and I was well shooting at Hi ISO regularly. I have a feeling I am going to change my gear brand soon.

Juraj Lacko said:

Hello everyone

Does anybody noticed obvious image split - like in the middle of frame while shooting at high with ISO setting in dark / night scene? I have noticed this at ISO 3200 but very pronounced at 25600. Has anyone got same problem or is this appearing only on my camera? I have latest firmvare installed. Sample image was shot with 24-105mm lens. I noticed that with other lenses so problem has to be somewhere else.

Click to expand...

I recently heard that Canon's FF sensors are actually two stitched 18x24mm sensors. You situation tends to suggest that this may be true. It may require a repair, but before you give up your camera for 3 weeks, try doing a fake manual sensor clean - start the cleaning so the mirror flips up and the shutter opens, and then turn the camera off after a minute. You never know, the camera might calibrate this aspect as part of the cleaning cycle (we do know for a fact that it maps bad pixels when you do this).

--
John

Thanks John

I have tried that manual sensor cleaning as You adviced but sadly there is no any improovent in split image. I will have to send it to repair. However i did some testing and at 0EV compensation i have got split image. when i usep EV compensation +2 i noticed the line dissapeared and image was smooth as it would be shot at base ISO 100. My test were done at ISO 25600 at daylight. will try that in drk scene again.

John Sheehy said:

I recently heard that Canon's FF sensors are actually two stitched 18x24mm sensors.

Click to expand...

Do you recall where you heard this? Is it actually stitched sensors, or is it two separate etchings on the same sensor?
--

panos_m said:

Probably the 5D2 sensor is combined from two different pieces. Severe underexposure reveals the stitching. Something similar happens to the Nikon D3 sensor with the differences that the parts are four an that is also observable when the green channel is saturated. This from my D3 sensor:





--
Panagiotis

Click to expand...

I don't believe that is stitching in the silicon (though that will be there, there is no reason why it would cause differences like that, stitching will not cross a pixel boundary). What you are seeing is stitching in the moulds for the plastic microlens layer, which are made using the same photolithography techniques as the chips. That will cause small differences in alignment of microlenses with pixels, which will cause the effects above.
--
Bob

Victor Engel said:

John Sheehy said:

I recently heard that Canon's FF sensors are actually two stitched 18x24mm sensors.

Click to expand...

Do you recall where you heard this? Is it actually stitched sensors, or is it two separate etchings on the same sensor?

Click to expand...

It is highly unlikely that all layers of the Canon FF sensors are made in one pass without stitching. The geometry of the stepper they have available with a large enough reticule is simply too large. I believe that the theory that the Canon sensors are unstitched comes from a chipworks article:

http://www.chipworks.com/en/technical-competitive-analysis/resources/technology-blog/2008/03/dslr-sensor-economics/
where they show this image



And declare that they can see no such marks on Canon sensors.

Now, their reasoning is slightly flawed. When you look at the surface of a sensor, what you see is the microlens layer, then the CFA then the silicon below those two. It's highly unlikely that you'd see artifacts as small as stitching artifacts through the microlens and CFA. On the other hand, you will see stitching artifacts in the microlens layer, since that is the closest layer to the eye.

So, my belief is that what chipworks discovered is that Nikon stich the moulds for their microlens layers and Canon doesn't. I think one cannot make any deduction about the silicon, and since Canon has repeatedly said that APS-H is the largest sensor that can be made in one pass, it's probable that they are using more than one for the FF sensors, as does everyone else.
--
Bob

John Sheehy said:

I recently heard that Canon's FF sensors are actually two stitched 18x24mm sensors.

Click to expand...

You may have heard that, John, but I can't believe you gave it any credence!

The sensor is a single chip cit from a hige wafer containing dozens of the damn things. The "stitching" people refer to is "reticle stitching", a standard process where the final chip size is larger than the projection area of the photo-lithographic equipment. On the original 5D there were two reticle stitches across the sensor, on the 5DII there is one - just where the OP sees the distinction. As Bobn2 notes, reticle stitching is required in every photo-lith stage and, since the microlenses (and Bayer filters) are fab'ed using photo-lith, this might be just a microlens issue.

Either way, it should be calibrated out of the final images from the camera using factory non-uniformity correction (NUC), and this camera will require to be returned for servicing to fix the problem.
--
Its RKM

Victor Engel said:

John Sheehy said:

I recently heard that Canon's FF sensors are actually two stitched 18x24mm sensors.

Click to expand...

Do you recall where you heard this? Is it actually stitched sensors, or is it two separate etchings on the same sensor?

Click to expand...

Sorry, I didn't realize that the term "stitched" meant something very specific. I used the term loosely.

--
John