5D Mark III and green RAw!!!

  • Thread starter Thread starter vasya4kin
  • Start date Start date
Status
Not open for further replies.
I am 99% sure I know the cause of this. When the AWB reading is somehow saturated by an extra bright light source at the time when the AWB is set, then the WB reading is way green. It's happened to me with many other cameras, but actually, with my 5D Mark III it hasn't happened yet. The same thing can happen if you try to set manual white balance from a very overexposed shot, with at least one of the color channels completely saturated.

A picture turning green can't possibly have anything to do with a memory card error, unless the card error somehow has an intelligence of its own and knows how to change the picture's color rather than randomly corrupting some bits.
 
Upvote 0
helpful said:
A picture turning green can't possibly have anything to do with a memory card error, unless the card error somehow has an intelligence of its own and knows how to change the picture's color rather than randomly corrupting some bits.

It might be a bit changed in a green channel value, affecting the whole image. I think that raws store the initial information and operations + values applied, but without changing the values of the pixels -take a look in xmp file.
 
Upvote 0
Freelancer said:
tron said:
But the 3500-3600 dollars/pounds/euro that Canon / dealer would get would be perfect!
I do not think they would accept e, hmm for example 3400 plus a piece of paper saying sorry can you please imagine this piece of paper as being the rest 100-200 missing (to gether with buyer's gratitude ;D )

buy a car or a TV, buy a printer, buy.. whatever.. everything has flaws.

you use a PC... i guess you never had a software bug?
my PC cost nearly as much as a 5D MK3.#

so above statement from you makes no sense in the real world.
Except that you compare software bugs with something that is not 100% verified that it is indeed a software bug. I wish it is but we do not know yet.
What is more certain is your desire to try to insult
 
Upvote 0
My initial thoughts on seeing this picture was "white balance issue".

I'm curious how it looked after you imported the RAW file into the computer and adjusted the white balance?
Did it then return to being an acceptable picture?

If so - then white balance setting is the cause - but what upset the WB setting is another issue altogether ???
Forgive my not asking - but it doesn't look too bad in B&W ;)
 

Attachments

  • B&W.jpg
    B&W.jpg
    31.8 KB · Views: 813
Upvote 0
Viggo said:
feil.jpg

The Matrix is starting to unravel, eh? :D
 
Upvote 0
neuroanatomist said:
marekjoz said:
I've read somewhere, that the most often reason for damaging computer memory modules was cosmic radiation.

I once received a box of Kodak BioMax MS (ultrasensitive autoradiography film), a sealed package of 100 sheets, and after developing the first one (following a 5 day exposure to a radioactive sample), I noticed a ~6mm diameter black spot on the film that didn't come from my sample, and could not be attributed to a light leak in the film cassette. I pulled a few more sheets from the pack and developed them immediately, they had the same spot - stacking the whole pack together after developing showed that the spots formed a complete column passing through the stack at a slight angle, and I suspect the pack was hit by a cosmic ray during shipping.

RE the green RAW wedding photo, I suggest you contact all the guests and a random sample of individuals nearby the locale but not present at the event. Check your watch, the watches of the guests, and the watches of the people not actually at the event. I suspect you'll find a small temporal difference between those there and those not there, indicating that the entire wedding party and all the guests, and you, were abducted by aliens. Your shot was taken just as the matter transmitter beam was collecting you all for transport, and while the aliens eliminated most other traces of their presence, they missed your 5DIII. I would not report this to Canon, but to the National Enquirer, instead. :P

Now see.... THIS is why I frequently skip ahead in a thread, and just look for a post with Neuro's avatar

Don't ever change your avatar bud
 
Upvote 0
vasya4kin said:
hello, yestarday i was on the wedding and shoot photo for grandma :)
but 1 photo was green(i shoot in raw)
picture is
4xunUA3N8HM.jpg
only raw convertor FsViewer.

Canon what it is????
Bride ask me, is it bag only one picture?
what i must to say????

iso auto
bw auto
Av
+2/3
135 2.0
multisegment
write on 2 card on sd and cf.

You could probably save this photo in a B&W.

Is it possible a faulty card corrupted the Image?
 
Upvote 0
This is most likely a firmware bug. To my understanding, red and blue channel information were missed while writing the data on the card. Since the image is similar on both the cards, I believe the loss of information took place even before writing.
 
Upvote 0
I've had this sort of thing happen a few times on my 5D2. I think I might have had it once on my 5D3, too. From what I recall mine were always really blue. It's not a white balance issue at all...it's an error in the camera. I would be shooting a series of shots and the middle one would randomly come out with nothing but blue, like not all of the color channels weren't recorded. I've heard other people report it as memory errors, but nothing really definitive. Conversion to B&W usually did produce acceptable results, too. I'll did through my library and see if I can find a sample to post.
 
Upvote 0
Chris Burch said:
I've had this sort of thing happen a few times on my 5D2. I think I might have had it once on my 5D3, too. From what I recall mine were always really blue. It's not a white balance issue at all...it's an error in the camera. I would be shooting a series of shots and the middle one would randomly come out with nothing but blue, like not all of the color channels weren't recorded. I've heard other people report it as memory errors, but nothing really definitive. Conversion to B&W usually did produce acceptable results, too. I'll did through my library and see if I can find a sample to post.

Same here, 200 normal photo's then one extremely green, like no other channel has recorded anything. The first few green shots I didnt give a second look, just gave up on them.
A few days ago I had a green photo that I wanted to use and opened him anyway in photoshop. Where I noticed that in my case it was just a WB + tint problem on that photo. Opening the older green photo's and I noticed they all had the WB + tint problem. 2000k WB with -150 Tint.
So in my case the photo's were not corrupt.
 
Upvote 0
helpful said:
A picture turning green can't possibly have anything to do with a memory card error, unless the card error somehow has an intelligence of its own and knows how to change the picture's color rather than randomly corrupting some bits.

Actually it could. It depends on how the RAW files are compressed. If some important bits get corrupted, it's possible an entire color channel could be lost, or for every value in a color channel to be effected when the data is uncompressed. I don't know the details of CR2, but for JPEG this is very possible since it uses discrete cosine transforms (which are similar in nature to discrete Fourier transforms) to store data in the frequency domain with separate luminescence and chrominance channels, and even uses Huffman encoding tables to save space! To display the data you have to do everything in reverse, including of course inverse discrete cosine transforms.

If the data was stored as a boring flat array of radiance values for each color, then yes, random errors wouldn't do this, however I suspect that's not the case with most image storage systems, including ones with lossless compression.

tron said:
Even if it is very rare it's UNACCEPTABLE!

Well it may have nothing to do with the Camera, and just be a random memory card error, in which case you should spend lots of money on error correcting memory cards if that's not acceptable to you. If it's the camera's "fault", it could be a flipped bit in the buffer, which is just fairly standard RAM (which corrupts data from time to time). If you wanted to spend a ton of money, I'm sure they could build cameras with error correcting RAM, it exists for servers but costs far more than normal RAM.

So yes we could likely make these things 1 in 1,000,000,000 events, but it would cost more than reasonable person is going to pay.

Of course they could also save parity information with the file to try and correct bit corruptions, but then our files would be a lot larger, and people would complain about that ;)
 
Upvote 0
Status
Not open for further replies.