Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Banding issues caused by Auto Focus pixels

  • Please log in to reply
6 replies to this topic

#1 sharkmelley

sharkmelley

    Vanguard

  • -----
  • topic starter
  • Posts: 2000
  • Joined: 19 Feb 2013

Posted 19 March 2016 - 08:24 PM

I've started a new thread to discuss this issue because it is not just a Canon problem.

 

This has been mentioned in relation to the Canon 750D and Canon 760D in this Cloudy Nights thread:

http://www.cloudynig...ing-help/page-3

 

Another thread on the Canon 750D banding issue caused by the Auto Focus pixels:

http://photomacrogra...pic.php?t=30169

 

Apparently Canon's response is to not heavily post process.  Most deep sky images require what would be described as heavy post-processing (e.g. light pollution subtraction).  If it is true that calibration frames (i.e. flats and bias) do not fix this issue then one has to question the suitability of the 750D (and 760D) for deep sky astro-imaging.

 

The Sony A7r2 has similar banding caused by the Auto Focus pixels that can even be seen without post-processing:

http://www.dpreview..../thread/3931191

 

The Sony A6000 has the same issue without post-processing:

http://www.dpreview....s/post/56591734

 

I hope these sensor issues are not the start of a new trend.  The cameras I mention above are just ones I happened to find doing internet searches. How many other cameras are similarly affected? 

 

If you have examples from other cameras then please post them. Is it really the case that calibration frames cannot deal with this issue?

 

Mark


Edited by sharkmelley, 19 March 2016 - 08:38 PM.

  • Charlespl likes this

#2 dciobota

dciobota

    Surveyor 1

  • *****
  • Posts: 1646
  • Joined: 07 Aug 2007
  • Loc: Omaha, Nebrrrraska

Posted 19 March 2016 - 08:41 PM

Hi Mark I just posted in the other thread.  I have a Canon M3 which does not seem to exhibit af banding issues.  I have not used it for AP much (just one short session) but I did stretch the final image pretty heavily.  Also I used no filter and the camera is unmodded.

 

Hope this helps.



#3 NMBob

NMBob

    Apollo

  • *****
  • Posts: 1446
  • Joined: 21 Nov 2015
  • Loc: New Mexico

Posted 21 March 2016 - 10:57 AM

What would be the simplest way to see if your camera has the problem? I just got a T6i, so I'm not subbing or stacking or anything like that yet. That should probably be in this thread if it's going to be one of those colossal kind of threads.

 

This picture is 8s @ iso1600. I sucked the raw image into Pixelmator (Mac photo editor), ran the Auto Enhance function (which has no idea what to do with a picture like this) which brought out the snow, upped the simple contrast a bit, then dragged the RGB Curves up and to the left. Shrank and sharpened a bit for the post. If I look down the picture from the top of the screen I can see some black bands mostly on the right side of the picture. A couple white bands in the upper left area. There's also a diagonal dark line in the area to the right of M42. Don't know how that would get there from this sensor problem. I think these lines were all visible after just the first Auto Enhance step. The contrast and curves adjustments just made them a little easier to see. Bad sensor? Didn't Canon say there was a problem with serial numbers 01xxxx... and 02xxxx... when these first came out? Mine is 05xxxx....

 

bandingQ.jpg

 



#4 t_image

t_image

    Vanguard

  • -----
  • Posts: 2391
  • Joined: 22 Jul 2015

Posted 21 March 2016 - 01:26 PM

My 2 cents:

mini-rant: I wish they would leave alone autofocus stuff on prosumer products.  I knew the AF pixels would cause a problem somehow!!!!!

How hard is it to turn the focus dial of your lens?

 

As far as lines....Maybe a different issue, but I found I could create scan line aberration exploiting the electronic front curtain with a lens aperture manually stopped down and the framerate dialed high-that was visible in the EVF in live view using video.

 

For the life of me I can't repeat this with my unmodded a7s. Maybe it was my modded a7s that I could do it with (not with me at the moment).

Could the lack of an anti-aliasing filter have something to do with it? (at least make the lines easier to see?)

Interestingly the a7r2 comes factory with an AA filter.

 

It makes sense the explanation on the conditions that go beyond the tolerances of the mechanisms set to hide the phase detect pixels-in the a7r2 link.

 

If the On-Chip Phase Detect pixels are to blame: one would expect not to find this phenomena occurring on the original a7s and a7r which lack such feature, yes?



#5 sharkmelley

sharkmelley

    Vanguard

  • -----
  • topic starter
  • Posts: 2000
  • Joined: 19 Feb 2013

Posted 21 March 2016 - 03:39 PM

Most of the examples show this effect happening in the shadow areas. My guess is that for astrophotography if you keep the back-of-camera histogram well away from the left hand side then it may not be an issue.  Part of the reason for this thread is to gather information and possible solutions because I have a horrible feeling this may become more widespread. 

 

I'm pretty sure the problem doesn't exist on my Sony A7S.

 

Mark



#6 entilza

entilza

    Soyuz

  • *****
  • Posts: 3548
  • Joined: 06 Oct 2014
  • Loc: Oakville, ON, Canada

Posted 21 March 2016 - 04:32 PM

This is definitely a cause for concern moving forward.



#7 dciobota

dciobota

    Surveyor 1

  • *****
  • Posts: 1646
  • Joined: 07 Aug 2007
  • Loc: Omaha, Nebrrrraska

Posted 21 March 2016 - 04:51 PM

One thing I find strange is the actual presence as "dead" pixels of the AF sensors (sensels?) in the final raw/jpeg.  I mean, why is the reader reading AF pixels in the first place?  Or why is the software not ignoring those "pixels" when writing out the final image?

 

 What I am suspecting (and of course I've been wrong before) is either those are not af "pixels" (I remember some sensors way back when had thin supporting wires across them that caused this kind of effect) or somehow the "dead" af pixels are having some shadow effect on neighboring sensor pixels under certain conditions.

 

 Just guessing. But I do think Canon can mitigate this with just a firmware update.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics