Firmware update for 5D3 Problem? Black halos around stars in RAW

[pedro]

Hi,
Still enthusiastic about my 5D3 but...Did some night sky photography last night and detected in all the frames black shadows around certain star clusters, no matter what ISO setting it was.

Guess I've seen this in early reports about the 5D3 in March or April. So as I bought it exactly for this kind of photography...What can I do? Did I get a lemon? Serial number: 04230012696. bought the cam on August 27 from a retail dealer...

http://www.flickr.com/photos/guatitamasluz/7956396466/#in/photostream


Thanks for your help. Cheers, Pedro.

Here's the dpr thread about it
http://forums.dpreview.com/forums/readflat.asp?forum=1032&message=40779387&changemode=1

[GuyF]

The black circles seem to appear only around the stars that look to be clipped by the exposure. That may have something to do with it - possibly coupled to the amount of sharpening used too. Have you tried less sharpening? Can you give any other exif data?

Then again, maybe you've discovered a new astronomical phenomenon! 

[risc32]

I can't say but what about the fact that your points of light are ovals? I thought i'd see batwings but they all look like uniform ovals. seems weird to me, but i don't have much exp with astro work, and i usually don't do much pixel peeping. esp when it come to my astro stuff, as it's just a mess to begin with. I tend to throw standards out the window, and just hope i can get something interesting, leaving the technically awesome stuff for nasa. But there are some guys on here that are fairly heavy into this, hopefully your get some info from them. Did you use long exposure NR?

[pedro]

@GuyF, risc32 thanks for your concern.

The phenomenum appears in RAW.
Sharpening mode setting is the in camera setting 3.
yes, I applied long expsoure NR, after the picture was taken.

There was a firmware update (1.1.3) out by June but as far as I could read, it did not address the black halo phenomenum.

Here's a full exif sample at mRAW (or is it related to mRAW?)
http://www.flickr.com/photos/guatitamasluz/7956960556/#in/photostream

The black halos aren't "visible" at print size, but they appear as strange clusters...

Hope Canon gets this fixed by the next firmware update.
Anyone out there having the same problems while doing night sky?
 

[Mt Spokane Photography]

There is another thread from a person seeing white X's on a black background.  You might compare notes with the poster, however, he did not yet post a image.  It sounds like a similar issue.

[GuyF]

I took a look at your link and can only suggest the points look a bit like hot pixels (assuming not all hot pixels would go red). I'm only speculating but maybe a combination of higher ISO and long exposure has caused them to get a bit hot.

Were you taking a number of long exposure shots in quick succession? This could produce hot pixels. Perhaps our resident guru Neuro might know.

[wockawocka]

One thing to remember with any long exposure is the sensor will get hot.

Hot sensor = not good for IQ.

If this 25 second exposure was one of several you can expect to see distortions like this. Often.
The 1Ds3 did similar if you did several in a row but the distortions were red.

Try it again but give a few minutes between each. There's a very good reason why the worlds leading telescopes are supercooled with liquid nitrogen....

At ISO6400 you are pushing the sensor 6 stops harder than it's native ISO, thus 6 times the amplification, which is driven by electricity. In an enclosed space. You should really be saying 'look at how much clarity there is' rather than 'I need a firmware fix'.

Also, turn all sharpening off in camera and selectively do it instead as ANY sharpening will enhance the distortion.

There's a fine line on the 5D3 where the heat will build quicker than it will dissipate. I wouldn't go above ISO1600. Just because you have it on your camera doesn't mean to say you should be using it.

[pedro]

I took a look at your link and can only suggest the points look a bit like hot pixels (assuming not all hot pixels would go red). I'm only speculating but maybe a combination of higher ISO and long exposure has caused them to get a bit hot.

Were you taking a number of long exposure shots in quick succession? This could produce hot pixels. Perhaps our resident guru Neuro might know.

You are right. I  took about 45 pictures within 15 minutes, mostly way up to ISOs as high as insane 51k and 102k. Absolutely unusable, though. Well, Neuro, may have another take on that too 

@wockawocka: thanks for your input as well! I will take less pictures. I followed examples seen on flickr concerning nightphotography. But yes, I will take it slower next time. As I aimed at "testing" the ISOs for the first time outside, there was a good thread in nightphotography on here, 1Dx vs 5d3 in ISO...Your input will be considered next time I am outside. The only thing is, that it started to look like this right after the first picture I took...as seen in post. Therefore I cannot be a 100% sure if that is a heat issue only...or at least not while I was starting to take pictures...otherwise I just know too less about any camera tech to judge on this...Thanks anyway, your input as all the others is much appreciated! Cheers.

[wockawocka]

What you're ambient temperature too.

A sensors at 15 degrees will take longer to heat up than one at 30 degrees etc.

Just keep experimenting!

[GuyF]

45 long exposure shots in 15 minutes? Yeah, that oughta do it! The sensor has to cool down a bit between these types of shot.

Just because it can also do video for long periods doesn't mean the sensor is used in the same way.

[nubu]

I am more or less certain that these halos are sharpening artefacts around saturated stars!  Switch off any sharpening and they should go away!  It has noting to do with the length of your exposures. The chip does NOT heat up during long exposures. Its heats up from frequent readouts as during live view and video! Yes the dark current is higher when the temperature is higher and my only suggestion would be to wait a few seconds after a longer period of live view but not after long exposures. Actually the chip cools during a long exposure. one can easily test this by taking long dark exposures (cap on). The signal S that you can measure is a bias level B and a time dependent dark current d times t (e.g. seconds) leading to S = B + d x t. Measure the B with a short exposure and then do multiple exposures of diffent length t. You will see that d is constant or even goes down when the camera cools in the cold outside air... (you have also to switch off long exposure noise reduction!). Better correct for the dark current by subtracting a(or many) dark picture(s) of the same length...  (free DeepSkyStacker does this job nicely)

[bkorcel]

I suspect that it may be an issue with light bleeding to the adjacent pixels.  It becomes apparent with long exposures where stars become saturated.  It could also be an impact from the antialiasing filter used in front of the chip.

Honestly the image looks great.  Don't pixel peep so much and enjoy the image in its full spendor.

If you are really interested in cropping astrophotos you should consider the 60Da which removes the antialiasing filter.

[pedro]

bkorcel, nubu,GuyF,wockawocka: Thank you for your inputs. I will take them into practice. It helps me a lot as I thought I had a defective device. But, as in most cases: It's always the dude behind the cam What a privilege to live in this century, with helpful mates around the web! You're great folks.

[Bosman]

I have just discovered this recently. I am concerned. Hasn't this been a problem on other Canon cameras in the past where there was a fix?
Check out the bright spot on her lip and veil. White orbs encircled with black. Yikes!

By the way this was a jpeg file at 1600 iso and i don't believe overheating had a thing to do with it since it was a huge place and i had ot get the bride in position on the balcony far far away.

[wockawocka]

Looks like blown highlights to me. The light on the shiny point of her lip and the sequin of her veil catching it in exactly the wrong position.
It looks quite poor for ISO1600 too, (is this part of a larger image or have you pulled the exposure)?