Canon 5DIV - Possible Strobe / Flash Defect - Can anyone confirm or deny this?

[MichaelTheMaven]

First let me say Im very impressed with the 5D4. Really incredible camera, even the video - I am very impressed, so this isn't a post to rip on the 5D IV at all. I think it is amazing.

I noticed something strange when shooting my Dynamic Range test, it looks like some sort of ghosting across the frame (see images below) and it occurs even when I block 1/2 of the lens with a poster board. It is a horizontal banding across the entire frame, even on the side that was completely blocked with a black poster board (right on the lens).

The strange thing is, I've posted this on Canon's Forum twice now, and both times it was taken down almost immediately. Which is really strange.

I was wondering if a couple of you guys could test the same conditions on your 5DIV's to see if you are getting the same results?

This is what I know:

1. The defect doesn't happen on a 5D3, A7rii, Nikon D500 or Fuji XT-2 under the same conditions.
2. It only happens on the very far left 1/3 of the frame when over-exposing a certain amount of it, about 1/4-1/3 of the frame, and it typically needs to start on the very far left side, and extend minimum about 1/4 into the frame. Overexposing in the center, the right, or smaller portions of the left side have no effect and seem normal.
3. The defect is strictly horizontal, even when the over exposure is oriented diagonally.
4. The defect is not visible until raising the shadows. You won't see it until then.

Im trying to figure out if this is something exclusive to the 5D4 I was using or if it is a model wide defect.

Can anyone else test this to confirm or deny?

1. Block the right half of the lens
2. Overexpose the left half with a strobe.
3. Open in RAW
4. Open up the shadows.

I thank you in advance!

Michael The Maven

 

[ajfotofilmagem]

The second picture shows what seems to be the phenomenon of "smear". Something very common in CCD sensors, but can also happen with CMOS sensors as part of the image is very overexposed by a large margin.

 

[MichaelTheMaven]

Thank you- bever knew about it.

Strange that I would see it on the 5Div and not the other cameras. Im wondering if anyone else can recreate the effect on theirs?

M

 

[StudentOfLight]

I wonder if it is due to the dual-pixel design. Could you perhaps try shooting the exact same test except with the camera it in portrait orientation to see if light still bleeds horizontally or if it bleeds in the direction of the long edge of frame?

 

[MichaelTheMaven]

I tried it in all positions, including top, bottom, right and middle which nothing happens.

It only happens on the far left side of the frame.

M

 

[wockawocka]

You can make anything appear to have a defect if you try hard enough.

You might be causing sensor bloom / blowing and that's how it manifests itself in the Canon sensor. You're essentially putting the camera through something it was never intended to do on any reasonable level and it's probably why you had the posts removed.

A bit like how if you shine a bright torch in your eyes you'll see spots for a bit afterwards even though the light isn't there anymore.

 

[StudentOfLight]

Just to clarify, take a look at the attached, if your test is conducted in portrait orientation, does the phenomenon happen as indicated by the arrows red arrows or is it in the direction of the long edge of frame?

What happens if you position the Stouffer Wedge on the right hand side of frame when shooting in landscape orientation. Does it bloom to the left of the overexposed block?

 

[MichaelTheMaven]

You can make anything appear to have a defect if you try hard enough.

You might be causing sensor bloom / blowing and that's how it manifests itself in the Canon sensor. You're essentially putting the camera through something it was never intended to do on any reasonable level and it's probably why you had the posts removed.

A bit like how if you shine a bright torch in your eyes you'll see spots for a bit afterwards even though the light isn't there anymore.

I test cameras this way all the time and no other camera is displaying the defect, so......theres that.

 

[MichaelTheMaven]

Just to clarify, take a look at the attached, if your test is conducted in portrait orientation, does the phenomenon happen as indicated by the arrows red arrows or is it in the direction of the long edge of frame?

What happens if you position the Stouffer Wedge on the right hand side of frame when shooting in landscape orientation. Does it bloom to the left of the overexposed block?

Thank you for taking the time to address this.

The defect only appears in one consistent direction.

In your top example it would be from top to bottom, and not at all in the other two.

It is specific to over exposing on the far left side of the frame if held horizontally.

Additionally, the defect only runs parallel to the the sensor orientation, no diagonals or ups and downs across the short side.

 

[IglooEater]

Looks like a case of blooming. I'm not used to seeing that on CMOS, but I had heard it's possible. The interesting thing is that CCD blooming fades off, while this stays pretty constant across the sensor. It would seem to be a digital bloom, if you will.

I'm curious- how much did you push the shadows in your examples?

 

[StudentOfLight]

Looks like a case of blooming. I'm not used to seeing that on CMOS, but I had heard it's possible. The interesting thing is that CCD blooming fades off, while this stays pretty constant across the sensor. It would seem to be a digital bloom, if you will.

I'm curious- how much did you push the shadows in your examples?

It must be flare(internal reflection), and I guess it's due to the the dual pixel orientation. When light hits the sensor it can either be absorbed (converted into electrons via Photo-electric effect) or reflected. Perhaps due to the microlens in front of the dual pixel... the dual pixels would reflect light at slightly different angles to each other. This might be causing an interference pattern with alternating constructive and destructive interference.

If there is constructive interference the light might not lose intensity with reflection and could propagate horizontally (in the direction of dual pixel orientation). I wonder does this phenomenon occur at all aperture settings or only at wider apertures or at one specific aperture setting?

It's curious that it happens only at the bottom of frame though. What is the position of the light source relative to the step wedge?

 

[LSXPhotog]

Well this isn't good, but my camera is experiencing the exact same phenomena.

 

[StudentOfLight]

Well this isn't good, but my camera is experiencing the exact same phenomena.

Does it affect any images in real world shooting? From what has been described so far I think you'd need to overexpose as well as raise shadows. I haven't seen complaints of blooming/flair issues in real world shooting so far. If you have, please share.

 

[LSXPhotog]

Happens anytime you need to raise shadows on my camera.

https://goo.gl/photos/QwCTidLFCEBgLrsE7

 

[Mikehit]

Happens anytime you need to raise shadows on my camera.

https://goo.gl/photos/QwCTidLFCEBgLrsE7

Your other post explains that you underexposed 4 stops, pushed 5 stops and lifted shadows 100%. That can hardly be described as 'anytime you need to raise shadows'.

 

[LSXPhotog]

Happens anytime you need to raise shadows on my camera.

https://goo.gl/photos/QwCTidLFCEBgLrsE7

Your other post explains that you underexposed 4 stops, pushed 5 stops and lifted shadows 100%. That can hardly be described as 'anytime you need to raise shadows'.

Once again, that was to exaggerate what is happening so I can more easily show the issue. If I posted it pushed 3 stops or 4 you would barely see it and people would act like it wasn't an issue. Some people have different standards off images quality.

 

[Mikehit]

Happens anytime you need to raise shadows on my camera.

https://goo.gl/photos/QwCTidLFCEBgLrsE7

Your other post explains that you underexposed 4 stops, pushed 5 stops and lifted shadows 100%. That can hardly be described as 'anytime you need to raise shadows'.

Once again, that was to exaggerate what is happening so I can more easily show the issue. If I posted it pushed 3 stops or 4 you would barely see it and people would act like it wasn't an issue. Some people have different standards off images quality.

Or it is an issue created by auto-suggestion? A very powerful force in sensory physiology.
Knowing the banding appears by pushing 5 and the shadows lifting 100%, you believe you can see it when you push 4 stops and it is not really an issue at all if you hadn't done that extreme test.

 

[jebrady03]

No problem here, but I didn't use a current version of Adobe's products, either.

I tried to create the issue in Capture One and could not.

I converted that same RAW file in Adobe DNG Converter and opened the DNG in LR 5.7 and could not create the issue that way either.

So that leads me to one of two conclusions....

1) My camera is unaffected

2) It's an issue with the current version of Adobe's products

Michael, I looked for a download of one of your RAW images and couldn't find it to try it out with my 2 methods. If you'd provide one of your RAW files, I'd be happy to check my 2 methods. Also, I'd recommend trying in Canon's DPP software (which I also have and would be willing to try for you if you don't want to go through the trouble) to see if you can recreate the problem there too. It's possible Adobe is to blame.