First let me say, I have no idea how many or what percentage of Canon 5D4’s this affects. It may be 2% it maybe be 20% or even more. If you shoot mostly in JPEG and do not aggressively edit, you may not even see it.
Here is how you do the test:
1. Use your hand or a piece of cardboard to over-expose in the left 1/3 part of the frame shooting RAW
2. Bring the file into Camera RAW
3. Bump up the exposure +1 or more (this makes it easier to see)
4. Raise shadows. It will look like the images below if you have it:
Side note, I did not see this in other cameras, including the Canon 5D3. It also does not happen in the middle, top, bottom or far right of the frame. Only on the left. If you do not over expose enough on the left, you may not see it as well. Here are some examples:
At +2 Ev
At +1 EV
0 EV (Very hard to see after it has been screen shotted and uploaded for web, much more visible in camera RAW)
Defect always runs parallel to sensor.
Defect doesn’t appear if there is not enough exposure in far left 1/3
Also no defect on right side of camera. 
I tried this, overexposed left side 5 steps & checked RAW in Lightroom and no, nothing is wrong and there is absolutely no sign of any "streaking"
Michael,
Haven’t taken the time to do controled test shots of a gradient step-wedge yet, but I tested on some -4 stop underexposed frames from bracketed HDR sets and I did not see the problem.
I suspect the "defect" you show may be more related to the shadow recovery function in Adobe Camera Raw than the 5D4 sensor. However, it may be more pronounced on the 5D4’s 30 mp sensor than a 15 to 20 mp sensor.
ACR’s shadow recovery is a very unique and strange beast. It uses a function called "Laplacian Filters" to analyze the image, which means the effect it applies depends on image content. The effect will vary based on the overall balance of shadows vs. midtones vs. highlights in the image. The effect will also vary depending on the size of a shadow area and it’s relation (proximity) to higher tones.
The more shadows there are in an image, the stronger the effect. Your test shots have very little highlights and a lot of shadow, so the effect would be very strong. And your left-side vs. right-side shots have a significant difference in the shadow/highlight ratio. That may account for the difference you see between left vs. right.
I’ll try to find time to do a controlled test later, but in the meantime I’d encourage you to do some additional testing. Shoot your step-wedge against a white background, making the frame mostly highlight with very little shadow. Shoot at different zoom levels to change the size and ratio of the shadow area. Also, consider testing with the "Fill Light" slider under the "2010" process of ACR. The older 2010 process does not use the Laplacian logic.
Haven’t had a chance to test yet. A theory though: the image is captured upside down and flipped left – right so this is actually the right hand side of the sensor. My 5d iv gets quite warm right around where the new control was added above the scroll wheel. I wonder if this heat is affecting the sensor?
Hey Michael,
I use Capture One for editing and tried to reproduce this within that program and could not. Have you tried reproducing the problem in Canon’s own DPP? Capture One? Another editing suite? The problem MIGHT lie with Adobe, and not with Canon (potentially).
An addendum to my prior comment… I used Adobe DNG Converter to convert the RAW file to DNG, imported to LR 5.7 and repeated the test, no problem there either.
Hi Michael,
I tried exactly as you said and my copy of Marh 4 have no streaking issue. Thaks for all the work you do.
Hi Mike have you tested this on a 5dmkIII or other cameras to determine whether this is isolated to the mkIV?
Hi Michael
i did the rest and i have the results similar like yours but mine looks even more then yours
how do i share the results with you
thanks
regards
satish
Can you do two things please.
1) Circle in red the artifact you claiming is a problem.
2) Be fore specific on how you want to cover a portion of the frame and how you want to expose/over expose it.
I have a 5D Mark IV that I just got last week and I’m more than happy to test it but I need a little clearer instructions. Thanks.
@Red Crow- I don’t think this is a ACR issue, I see the defect in JPEGS as well.
@Jonathan- Canon has replaced some of these already, so it appears to be a hardware issue. We are seeing them in about 50%
@Marv- Yes- Not seeing it on the 5Diii.
@Satish- Send us the images! [email protected]
@Steve- Simply cover half of your frame with your hand and take a picture of something really bright in the left side of the frame. Ill update the issue in these images soon.
Thank you everyone for your help!
I was skeptical about this at first but unfortunately this is true, and it seems to be really bad on my camera. It is only on the left side as you have stated.
I believe this must have something to do with the dual pixel technology. Each pixel below a micro lens sees half of the pupil, so if you mask a part of the pupil, you’ll prevent the light to reach each half pixel. If the demozaiking then does not take into account that the camera produces potentially 2 raw files which have to be processed in a specific manner, you can end up with such a problem.
In some occasions, this is not a bug but a feature, it is exactly this phenomenon which helps in reducing some specular highlights of flares from the lens.
This could also happen with other dual pixel cameras like canon’s 80D.
Not clear yet to me if it happens because ACR does not take into account the double nature of the pixel.
Maybe canon own DPP can demozaik the raw files while adobe’s ACR cannot since the algorithm therefore must be specific.
Cheers
I did this test thru the viewfinder and live view. I do not have the issue. I received my camera only 6 days ago. I wonder if it was with the first batch they sent out. Mine also came with the latest firmware.
Can somebody tell me exactly what I’m looking at? I would like to test my 5D IV. I don’t understand what I’m looking for, or what the black shadow is on the right?
Contact us USA toll-free 1-888-993-6399 USA for canon tech support, canon printer help, canon drivers, canon printer setup, canon printer software and all related problems of canon.
After initial testing, it looks like my camera doesn’t show this phenomenon. I just received it a couple of days ago, so it might be from a newer batch.
Confirmed… Take a quick shot of the moon on left 1/3 side, set to +3 Exposure and raise the shadow… Got streaks!!!…
No defect, when the moon is in the middle of the camera.
Performed the test according to your description with my 5DMKIV and could NOT reproduce it!
Preformed the test and could not reproduce it.
Just if anyone having that kind of a issue could reproduce it with Canon Digital Photo Professional?
Hi, tested it with my 5d4 and lightroom cc and NO I did not see the issue.
I think for comparison of cameras the actual firmware version should also be mentioned since CANON might have fixed it in firmware.
So I dont see the problem, Firmware version on my camera is 1.0.2.
An other thing: if the problem only shows in the left side of the frame I would say this can NOT be a problem introduced in post. It must be generated in the camera.
Firmware 1.0.2, post in lightroom cc, Issue not seen
Greetings from Germany
Hello I also have the firmware 1.0.2 and have the image in LR tested. I could not find any shadows. Greetings from Germany
Michael The Maven: Have you got any further information on how Canon is proceeding with this issue? because ? i wanna buy one
I’ve done the test twice: 5DMkIV early sn 02302100049x (one of the first 500), Firmware 1.0.2. With the settings: 70mm/f4.0, ISO 100, 1/10sec. I can’t reproduce the effect in Lightroom. I share my result here:
http://project77.de/root/canon/IMG_2377.CR2
Olaf Your picture has too much data in the shadows – not clipping enough. That is why it does not show. The highlight is also not bright enough. If those would be, I guess that would show, like in every 5D4 tested so far. Mine included.
Similar banding issues seen in this video https://youtu.be/RmfMLeB3lJw. It looks like a shadow recovery issue. Hopefully it’s just a software problem that can be fixed.
I have that problem too! Greetings from Germany.
Is there more discussion on the web about this issue? I can’t find much updates. I wonder if the ones who find their cameras not defective, are simply not replicating the problem correctly. I also saw a lack of sharpness in the center of the image during long exposures. There is a video online showing this. And problems of hot pixels, sometime wrong exposure or white balance. Lots of issues to be clarified, and not much testing happening. If you have good links, post them! Thanks!
Received a 5d Mark IV on 1/6/17. I have the problem. Imused my kitchen and the big window in it. At ISO 100, proper exposeure no problem pushing with no streaks. Exposing for the window hightlights makes it -3 stops. You can see the streaking a 60 shadow push althoughnot very distinct green and purple. Raising the exposure slider to even +.5 makes the green and purple streaking visible across the frame. I would bet this affects all cameras to some degree some worse than others. This along with the loud USM focusing noise in 4k video is going to make me return the camera. Inexcusable for a $3500 camera.
Hey Michael , just found out mu 5Dmk4 haves this sensor defect, i was wondering if from the moment you posted this you have any news? firmware o something? Thanks!
I have these purple and yellow streaks too. Also, If you take the raw files from the DPreview studio shots (I tried it with "793A6390.CR2") that are underexposed you can see the streaks in their images also.
i just pulled my camera out of the box and took it for a test run and was horrified to see the banding problem. It showed up most on out-of-focus areas, and is very noticeable. My Mark III has none of this. I’m getting a hold of Canon this week. This sucks big time after all of who have spent the high price to upgrade. Not happy. Love the way the camera works, but this issue is going to have to be resolved by Canon very quickly.
I can also confirm that I have the same problem.
Interestingly, when you look at a lot of samples and examples on the internet, you can see that in 90% of cases the strip is 1/3 the height from below.
And I wonder if this is the result of software, an algorithm, or Canon has produced a series of sensors with the same defect.
Generally, + 5EV + and 100% shading, with a lot of underexposed picture it’s a bit overdone. I do not think that people using cameras in this category, when taking pictures do they make the picture was quite underexposed.
The examples we see are 35-50% of the frame of the underexposed image.
Let’s get back to 5D mkIII and pull Ev +5 and shadows + 50% in the same case. The difference can be seen immediately.
That is why I am looking forward to some solution, explained by Canon – that MK IV is so under such conditions and post process.
Hoping it is a software fault, not a mechanical one.
For all, have a nice day.
I did this test on my grey market 5div and fortunately did not see this on my copy. Whew!!
I have this issue in my camera that was purchased from Adorama in April 2017. I first noticed it when I was taking fireworks pics on the 4th of July and the fireworks were causing the streaking. It also seems that my copy has quite a bit higher amounts of noise at 100 ISO than my 5D Mark II did.
I have the same problem – Seeing greenish, yellowish and redish stripes in Same light situations. I purchased my 5Dmark4 in November 2017.
I discovered this issue while doing some astro photograpfy. My images show a green bands or stripes in the upper third and lower third.. Its been a couple of years since the last post, has anyone found a solution?