IR and DPP 4: whence cometh the unseemly tiny dots?

[scottburgess]

I was practicing IR photography with a Singh-Ray I-Ray filter on a Canon 300mm f/4 IS and 5D Mark III body. Exposures were typically around ten minutes owing to an incoming storm, apertures f/5.6 - 7.1. On processing images tonight, I see that the IR photos have a consistent pattern of white dots when viewed and processed in DPP 4. When the images are completely desaturated they are quite obvious.

Initially I was worried I had a sensor problem: I had cleaned filter and lens immediately before use in the field, and would have expected any remaining dust from the filter to shift a little as I had to take off and put on the filter several times. I took ordinary color photos of the same shots and cannot find a trace of dots in the color shots even when desaturated.

I tried opening one of the images in Bridge, and I could see the white dots initially in the preview... and then they suddenly disappeared! I proceeded to open the file in Photoshop 6 and couldn't find a trace of them. Closed without saving, went back to DPP 4 and there are all the dots waving at me, pointing and laughing. I repeated this process a couple times with the same and different images. Once Bridge completes an initial preview of an image, it remembers any processing it does so future peeks in that session show no dots.

A sample crop of a desaturated image included as processed in DPP 4; yes, there are lots of dots, and they appear sharply focused as opposed to diffuse (as one might expect for lens dust).

Question: is this a software problem with DPP4, a hardware problem like sensor dust or sensor error, a common issue with IR by filter, or some combination thereof? I presume the dots are artifacts in the images since both software programs initially detect them, but if that is the case then why does Bridge clean them up even in its preview window?

 

[Mikehit]

Have you tried a dark (non-IR) shot with the same exposure length. It may well be 'hot' pixels.
It may also be worth trying an IR shot with long-exposure noise reduction as well.

 

[Maximilian]

Hi Scott!

These are typical "hot" pixels of the sensor (white in IR, red or white in normal pics).
They typically appear more prominent with long exposure times.
I suppose you will also find some "dead" pixels (black) as well when you look closely because you will never have a sensor with zero pixel defect - or the camera would be way to expensive.

What you can do - and what I did when I had a very prominent red pixel - , if you don't want to handle these pixel every time in post, is to send the camera to Canon and let them make a sensor check.
The sensor is then recalibrated and the defect pixels are calculated out by internal software - typical for all kind of digital cameras, and it served me well with my "red dot"

I don't know how much that costs because for me this was still under warranty and so it was free of charge.

I hope that information helps.

 

[DominoDude]

Hi Scott!

These are typical "hot" pixels of the sensor (white in IR, red or white in normal pics).
They typically appear more prominent with long exposure times.
I suppose you will also find some "dead" pixels (black) as well when you look closely because you will never have a sensor with zero pixel defect - or the camera would be way to expensive.

What you can do - and what I did when I had a very prominent red pixel - , if you don't want to handle these pixel every time in post, is to send the camera to Canon and let them make a sensor check.
The sensor is then recalibrated and the defect pixels are calculated out by internal software - typical for all kind of digital cameras, and it served me well with my "red dot"

I don't know how much that costs because for me this was still under warranty and so it was free of charge.

I hope that information helps.

I would first try to map out the stuck and dead pixels by going into the menu on the camera and select "Manual sensor cleaning", then wait some 30 seconds and then turn off the camera.
I think the process is listed somewhere in the manual, but can't remember where.

 

[Maximilian]

I would first try to map out the stuck and dead pixels by going into the menu on the camera and select "Manual sensor cleaning", then wait some 30 seconds and then turn off the camera.
I think the process is listed somewhere in the manual, but can't remember where.

Hi DominoDude!

If you are refering to a "Manual sensor cleaning", meaning lock up mirror, usage of bellows for blowing dust away and/or using any kind of manual device, then I would strongly advise against doing so.
Manual sensor cleaning - even if neccessary because of dust - is something only someone experienced should do. Otherwise even more damage to the sensor could be the result.
And it will not help with hot or dead pixels!

But if you mean to try out the "Appending Dust Delete Data" function then I am with you to try that first as described in the manual.
But with my 5D3 it didn't work with hot pixels. But maybe I did it wrong.

 

[midluk]

At least for extremely hot pixels (i.e. visible during normal day-time photography) you activate manual cleaning (with the camera front cap on and viewfinder covered), wait for the mentioned 30 seconds and then turn the camera off. The camera will find and mask the hot pixels (had to do this two or three times and it really does work). Doesn't work for pixels that only show as hot on very long exposures, though.

Some raw converters can reliably detect hot pixels and ignore them for the image and/or can use a list of dead pixels to ignore (rawtherapee does both).

 

[DominoDude]

I would first try to map out the stuck and dead pixels by going into the menu on the camera and select "Manual sensor cleaning", then wait some 30 seconds and then turn off the camera.
I think the process is listed somewhere in the manual, but can't remember where.

Hi DominoDude!

If you are refering to a "Manual sensor cleaning", meaning lock up mirror, usage of bellows for blowing dust away and/or using any kind of manual device, then I would strongly advise against doing so.
Manual sensor cleaning - even if neccessary because of dust - is something only someone experienced should do. Otherwise even more damage to the sensor could be the result.
And it will not help with hot or dead pixels!

But if you mean to try out the "Appending Dust Delete Data" function then I am with you to try that first as described in the manual.
But with my 5D3 it didn't work with hot pixels. But maybe I did it wrong.

Hey, Maximilian!
Nope, I meant it like midluk also explains. There will be no poking inside and no exposure of delicate parts.

 

[Maximilian]

I would first try to map out the stuck and dead pixels by going into the menu on the camera and select "Manual sensor cleaning", then wait some 30 seconds and then turn off the camera.
I think the process is listed somewhere in the manual, but can't remember where.

Hi DominoDude!

If you are refering to a "Manual sensor cleaning", meaning lock up mirror, usage of bellows for blowing dust away and/or using any kind of manual device, then I would strongly advise against doing so.
Manual sensor cleaning - even if neccessary because of dust - is something only someone experienced should do. Otherwise even more damage to the sensor could be the result.
And it will not help with hot or dead pixels!

But if you mean to try out the "Appending Dust Delete Data" function then I am with you to try that first as described in the manual.
But with my 5D3 it didn't work with hot pixels. But maybe I did it wrong.

Hey, Maximilian!
Nope, I meant it like midluk also explains. There will be no poking inside and no exposure of delicate parts.

As the two of you mentioned the same wording for this procedure, I am willing to believe you, but I couldn't find something like that in the manual.

My manual (5D3, en, p. 299) states for the topic "Manual sensor cleaning",
(displayed in the in-camera menu as "set-up, point 3"-> "sensor cleaning" -> either "automatic", "now", "manually"):

"Dust that could not be removed by the automatic sensor cleaning can be removed manually with a blower, etc.

Sorry, I didn't get that procedure as you described out of the manual, though I read it and even though I thought to read it well enough.

Could you please give me the chance to learn and tell me on which page to look?
(best would be the manual for the 5D3 as it is the same body as the OP used)

What you describe seems to have to do something with the "Clean now" function, but also here I couldn't find something you've described.

Thanks in advance for your reply.

 

[midluk]

Afaik Canon does not officially document the procedure.

But you can find it when googling for example "canon fix hot pixel".
https://www.youtube.com/watch?v=WJBuGhMnvFo

 

[DominoDude]

The procedure was so simple that when I first read about it and used it (years ago) I didn't note down where I found the information. Now I assumed it was in the manual. But I've been known to be wrong before. The procedure works though.

Did search my own bookmarks and the CPN site but the best info covering the subject I found here -> https://community.usa.canon.com/t5/EOS/Red-and-blue-dots-on-EOS-5D-MK2/td-p/33501

 

[Maximilian]

Thanks, midluk and DominoDude for providing this information.

Interesting to see that this is indeed mapped to the "Manual sensor cleaning" function.

 

[gwooding]

If you find that doing the pixel remapping using the manual sensor cleaning technique doesn't remove enough of the "hot" pixels, try taking 10+ min of video before initializing the sensor cleaning.

This helps to warm up the sensor ensuring more of the "hot" pixels are mapped out (I managed to remove pixels using this technique that only showed up with 3+ min exposures).