Canon 5D MKII Black Dots / Black Spots Discussion

[Noel_Carboni]

Being that I have been doing a lot of pixel bending of my own as part of plugin development lately, I have some ideas about what could be happening:

The observation:

Quite clearly the black spots are appearing right next to bright white overexposed places - e.g., the filament of light bulbs, and specifically only on one side.

The possibilities:

In these spots the imager chip is likely reaching a "full well" condition, in which the light measuring capacity has been reached or exceeded. One sees this with astrophotography work when imaging bright stars in long exposures. With early imagers when this happened the overflow could spill into adjacent photosites, and make a bright overexposed point light source into a long white smear. Even the Hubble Space telescope did/does this.

Modern chips have special circuitry to alleviate / prevent the spillover due to full well overflow, and the circuitry in the case of the 5DII might possibly have been made too aggressive - i.e., once it's triggered it might actually be going too far in its attempt to prevent spillover into adjacent sites.

Or possibly the D to A converter is ringing after having shifted some very high levels through. With the mad dash to increase the number of bits of precision, is it possible someone overlooked what happens when the D to A is confronted with severe overexposure?

Also, bright white parts going to black have the smell of digital data overflow. I see this all too often if I optimize accurate math into fast math, and a routine delivers a value of, say, 256 instead of 255 (max white). Converted to 8 bits, 256 loses bit 8 and becomes 0 == black. I think the former conditions - full well overflow overcorrection and ringing in the D to A - seem more likely here, but I thought I'd mention the latter as well. Again, the digital processing stream is suspect because it's all just been redesigned.

It's possible the silicon will need to be redesigned to alleviate this, which means there's not likely to be a fix any time soon. Certainly software could be devised to help reduce it after the fact. I might even be able to come up with an action or plugin to fix it. Let me see what I can come up with.

-Noel

[Noel_Carboni]

Probably means that without some work the 5DII will be a bust for astrophotograhy though. I'm sure some folks will be disappointed by this.

-Noel

[Jerry Skrocki]

Another reason I will be looking elsewhere for a new camera.

The 1Ds Mark III does not exhibit the "black dot" phenomenon which is why I think it has something to do with the circuitry redesign to allow for increased ISO sensitivity.

"However, this second-generation (5D MarkII) sensor offers increased sensitivity and improved noise reduction circuitry that enables standard ISO’s ranging from 100 to 6400 and expanded ISO ranges from 50 to 25,600."

The other difference between the image processing on the 1Ds Mark III and 5D Mark II is the Digic Processor. The 1Ds uses Digic III and the 5D2 uses Digic IV.

Jerry

[Terry Zorich]

Noel,

I am impressed and even a little awed at your knowledge and theories about the causes of the phenomenon. Thanks for taking the time to think about this and post it - it was fun to try and wrap my brain around it!

[Terry Zorich]

Drew,

Not a personal affront; just a difference in opinion:
I think it's comical that you believe the "black dot" issue is of little concern compared to the video mode shortcomings. I see it the opposite way.

The 5DM2 is a still camera which happens to have a video capture mode. Capturing HD video is not the primary purpose of the camera, and the implementation has some shortcomings specifically BECAUSE the 5DM2 is a still camera and not a video camera (that there is an optical viewfinder, for example).

Video cameras with CMOS sensors can produce excellent images, but they are not good for fast lateral motion or panning. The whole thing is a compromise. Yet - as evidenced by a number of online shorts I've seen - it is still possible to capture some breathtaking footage with the 5DM2. And I don't recall Canon making ANY claims about how easy it would be for the average consumer to pick up a 5DM2 and shoot video like he would with a Handycam.

Have you reviewed the video feature in the Nikon D90, by chance? How does that camera's video capability compare to the 5DM2's?

[Johan_Elzenga]

Originally Posted by Jerry Skrocki

The 1Ds Mark III does not exhibit the "black dot" phenomenon which is why I think it has something to do with the circuitry redesign to allow for increased ISO sensitivity.

Not all 5D Mark II cameras have this issue. I just tested one, and whatever I try and whatever ISO I use, I can't reproduce this problem. That makes it less likely that it's a design issue and more likely that something went wrong in the production of a certain number of cameras.

[Jerry Skrocki]

Originally Posted by Johan_Elzenga

Not all 5D Mark II cameras have this issue. I just tested one, and whatever I try and whatever ISO I use, I can't reproduce this problem. That makes it less likely that it's a design issue and more likely that something went wrong in the production of a certain number of cameras.

It seems like another quality control issue from Canon, similar to the auto focus issue that affected the Canon 1D Mark III.

That puts consumers in a precarious situation. Do you take the gamble and hope you get a good copy, or wait for Canon to come out with a fix?

Jerry