What does it take to make the 5D2 show banding / cross hatched pattern noise when pushing shadows and / or exposure in PP?
I am trying to determine this, NOT because I want to start another "troll vs fanboy" debate, but to get some meaningful insights in the "issue"... so that I can work around it in real world shooting.
So far I have made the following observations:
1) The banding shows regardless of ISO values. I can get banding to show in ISO100 files, but in ISO3200 as well.
2) The banding shows regardless of RAW converter (at least the ones I use). I get similair results in DPP and ACR/LR.
3) The quickest way (for me) to reveal any kind of banding is pushing the LR2.5 fill light slider to +100 (maximum). Even though I would never use such a drastic measure for real... only for testing purposes.
4) At first glance there seem to be a direct relation between the amount of underexposure (for either the whole image or certain parts of it) and how much banding will be visible in the shadows after pushing like descibed under #3 -in high contrast scenes that exceed the DR of the sensor excessively.
But there is also some inconsistency here. Sometimes a certain high contrast scene will show no banding when you push the underexposed parts, other times it will show banding. I can think of only three possible explanations why the results are so inconsistent:
- It could be that when there is banding present in the shadows of high contrast images, a certain DR threshold is exceeded. And the other way around, when there is no banding visible, that DR threshold is not exceeded. I this is the case, it will be interesting to know what that threshold might be. Maybe it is related to a certain amount of underexposure (either for the whole image or just certain parts of it).
- There could be some malfunctioning inside the camera that pops up from time to time. Maybe hardware related. Although I have never seen banding when pushing the shadows of (slighly) overexposed images. So, this seems unlikely to me.
- James Markus (FM member) in another thread on this matter: I wonder if there isn't some electronic noise going on...because the pattern is quite often a a decreasing cascade both horizontally, and vertically. It is like the signal gets amplified on one photo site, and then it leaks to the next in a slightly less amplified state...by the edge of the frame the it looks still whacked, but less so. Maybe it is the nature of processing the new gap-less photo-sites, and new micro-lenses.
Now on to some samples. The first set of high contrast samples shows no banding or cross hatched pattern whatsoever. The second set does.
It is the inconsistency that makes it hard to predict when a certain high contrast scene will generate banding in the shadows after pushing in PP. To me that is the REAL problem. Because every cam has its strengths and weaknesses. Which is no problem as long as you are aware what those are and if you can find a way to work around them. The appearant randomness of the banding makes this rather difficult.
Question: For real world use where you don't use the fill slider at +100. People I trust are seeing the banding without any pushing etc. Now it would be very silly to suggest that somehow you are always able to expose perfectly or never need to push shadows. No real photographer, that is someone who shoots in the real world would suggest that.
However I'd love some RAW files up which exhibit the problem without the need for extreme adjustments so I can make my own decisions. The most I ever push is to the end of the brightness slider in ACR/LR (fron +50 to +150). Any more than that and I'm 'pushing my luck' not just the image . Point is that the brightness slider is a midtone slider. Shadows are set permenantly set to no lower than +2 and I never use more than a +20 of fill flash and only that in very contrasty lighting.
I wouldn't ever discount anothers experience as I don't shoot the same way as others and wouldn't be so arrogant to suggest that if I don't see the problem due to my shooting method any one else who does consider it a problem is any less a photographer or person. What I would like to do is make my own decision without having to pay £1800 for the privilige.
So request, RAW files that show the problem without needing a 3-5 stop shadows push. I believe the people who have them, I just want to see it for myself using my own method of processing, to see whether it's a problem for me.
I wonder if it is related to the amount of details? In your first example there are lots of fine detail, and there are less of it in the second example. Either the pattern is drowned by the fine detail, or when the neighbouring pixels are recording very different value somehow the circuit is less prone to interference (assuming it is caused by electronic interference).
Another possiblity is the colour temperature. The second example seems to be shot in lower K lighting. If my memory serves, in my usage this seems to happen most often with low and warm indoor lighting (of course, not necessary due to it).
I think banding will become noticable in images when you've pushed shadowed areas that are fairly consistent in content (same color/contrast) with red it it. It seems the red channel is the culprit. By the way, I believe this problem is simply the way the Canon algorithm distributes the red channel info when handling very low amplitute (dark areas with red content). Whether Canon addresses it or not in the future, they merely would need to better randomize very low red color info (dithering, or adding "noise" to smooth out end result).
Beni wrote:
So request, RAW files that show the problem without needing a 3-5 stop shadows push. I believe the people who have them, I just want to see it for myself using my own method of processing, to see whether it's a problem for me.
Sorry, can't help you out on this one...
So far, I haven't seen any shadow banding / cross hatched pattern noise (after pushing) in files that were correctly exposed or slightly overexposed. I have seen it in files that were (partly) underexposed and needed an exposure/shadow correction in PP. But I haven't determined a threshold yet. Like I tried to explain in the OP, sometimes an underexposed file shows the pattern noise after pushing the exposure/shadows (even after an initial -0,33 underexposure), sometimes not. The reason(s) behind this inconsistency is what I am after.
Navyblue wrote:
I wonder if it is related to the amount of details? In your first example there are lots of fine detail, and there are less of it in the second example. Either the pattern is drowned by the fine detail, or when the neighbouring pixels are recording very different value somehow the circuit is less prone to interference (assuming it is caused by electronic interference).
Detail can make it more difficult to see banding / pattern noise. But even in the 100% crop of the ground (in the 1st set of images) there isn't any banding visible. I do think your thesis on electrical interference is interesting though. I wonder how that would work?
Another possiblity is the colour temperature. The second example seems to be shot in lower K lighting. If my memory serves, in my usage this seems to happen most often with low and warm indoor lighting (of course, not necessary due to it).
The 1st image was shot at 3950K +8. The second image was shot at 4550K +5. Both were shot in natural daylight. So, not much difference in color temperature. Overall I would have to say that I have seen the banding in images at all kings of color temperatures. But maybe it is more extreme in mixed/warm indoor lighting? Could be...
robert61 wrote:
I think banding will become noticable in images when you've pushed shadowed areas that are fairly consistent in content (same color/contrast) with red it it. It seems the red channel is the culprit.
Interesting... I have noticed that too. That would explain the difference in my provided samples. The forrest shot contains much less red that the wooden floor.
However, I shot the cat under the same shooting conditions, but overexposed by 1 stop. There is no banding anymore. So, I would include the factor underexposure in your possible explanation as well
What you've done is shoot information into those shadows, there just isn't any in the previous shots but instead of random noise you're getting banding. Exposing for the shadows and processing for the highlights is exactly what we did for film. If the information isn't there it just isn't there. The fact that you get banding is just a nastier type of artifact.
I could never shoot that was in the olden days until the 5D and ACR 2.4+ when you would actually prefer to overexpose slightly to get information into those shadows. That said the go too far and you get colour shifts, desaturation and lack of detail in those highlights once recovered. You're also playing near to the big white brick wall of highlights.
I'm hoping for others to perhaps provide RAW's showing banding in unpushed shadows or mid tones, places where there should be information not just a black hole.
Beni wrote:
What you've done is shoot information into those shadows, there just isn't any in the previous shots but instead of random noise you're getting banding. Exposing for the shadows and processing for the highlights is exactly what we did for film. If the information isn't there it just isn't there. The fact that you get banding is just a nastier type of artifact.
I could never shoot that was in the olden days until the 5D and ACR 2.4+ when you would actually prefer to overexpose slightly to get information into those shadows. That said the go too far and you get colour shifts, desaturation and lack of detail in those highlights once recovered. You're also playing near to the big white brick wall of highlights.
I'm hoping for others to perhaps provide RAW's showing banding in unpushed shadows or mid tones, places where there should be information not just a black hole....Show more →
I know what I did... but thx for explaining this to me I was merely trying to determine wether overexposure has an effect on the banding. It seems it does. From what I can tell the banding is strictly related to underexposure (and pushing).
But maybe there is somebody out there who can provide us with a RAW that shows banding in unpushed shadows/midtones. If that is the case, I bet the file is still underexposed to begin with. The question will still be: by how much?
I think I have some files kicking about on here from a studio test my friend was performing with my camera. From memory I seem to recall being impressed by the amount of detail in the dark side of the face (Some quite harshly lit portraits). I'l ask his permission to put them up
From those who have looked into this a little more, is there any possibility of this being solved by firmware alone? Or is this a hardware issue? I would happily give up ISO3200 performance for a super clean 50/100/200/400 if there was say, a low ISO optimised firmware?
Thanks for posting you you can get the banding to occur. Since I hardly every touch the fill light setting I not see this phenomena. However, I do remember using the fill light for back-lit shots back in the 20D days before Adobe bought the technology and not being happy with the weird color noise.
Also, I find that your examples look better if you leave the light the way you originally exposed them, Dark area create a negative space which adds drama.
jerrykur wrote:
Also, I find that your examples look better if you leave the light the way you originally exposed them, Dark area create a negative space which adds drama.
Hi Jerry, like I said... for testing purposes only
BTW I think you can get the banding effect also if you use levels/curves to open up the shadows or when adjusting exposure. I don't think it is a fill light exclusive. But the fill light tool exposes the shadow noise the quickest (yes, I am lazy )
Is this type of banding noticable in large web sized images or prints? I see it in the crops, but not in the image posted, at least not at that small size.
I often push my 5D classic files; mostly using the fill light and brightness slider in LR and although I do notice increased shadow noise, I've not noticed banding.
Here are three 5d classic shots in which I increased the brightness in LR by 80-100....(I can't post crops right now, but you can definitely see increased noise in the shadows, but no banding)....
Do you expect that banding would show with the 5dII with these types of shots?
Daan, thanks for starting this thread. As I am thinking about buying a 5D2, I find this issue very interesting.
Although I have never seen banding when pushing the shadows of (slighly) overexposed images
So it looks like you can overexpose slightly when in doubt? In another thread someone mentioned that one can recover a lot of detail from (overexposed) 5D2 highlights.
Is this type of banding noticable in large web sized images or prints? I see it in the crops, but not in the image posted, at least not at that small size.
These files only have been processed with the fill light slider. I guess adding some additional contrast might make the banding more visible. But, pushing the fill light slider to +100 is pretty drastic. Normally I won't go above +20 fill light adjustments.
Anyway, in case of the pushed cat shot, you would be able to see the banding with larger web sized images or prints. If you print the non-pushed version, you won't see any banding (not even at very large print sizes).
I often push my 5D classic files; mostly using the fill light and brightness slider in LR and although I do notice increased shadow noise, I've not noticed banding.
Here are three 5d classic shots in which I increased the brightness in LR by 80-100....(I can't post crops right now, but you can definitely see increased noise in the shadows, but no banding)....
Do you expect that banding would show with the 5dII with these types of shots?
The banding seems directly related to underexposure. But how much underexposure will result in banding is unclear. And not all underexposure leads to banding. My forrest shot has underexposed parts, but doesn't show any banding after pushing the shadows.
So, the answer to your question can't be given with much certainty, unless we discover what exactly causes the banding and at what threshold it will become visible.
So, there is a chance that if you would take these kinds of pics with the 5D2 there will be banding (after pushing shadows/exposure)... but there is also a chance that it won't show banding. That's what makes it so disturbing...
lexvo wrote:
So it looks like you can overexpose slightly when in doubt? In another thread someone mentioned that one can recover a lot of detail from (overexposed) 5D2 highlights.
Well, by overexposing you allow more recording of detail in the shadows... at the expense of blown out highlights. It seems to have a positive effect on the shadow noise (at least in case of my "test" shot). If the overexposure is very slight, you will be able to bring back a lot by PP. But if you go overboard with this, the penalties will be severe... lost unretrievable detail, hue shifts, loss of saturation, etc.
Daan B wrote:
So, there is a chance that if you would take these kinds of pics with the 5D2 there will be banding (after pushing shadows/exposure)... but there is also a chance that it won't show banding. That's what makes it so disturbing...
Daan B wrote:
Well, by overexposing you allow more recording of detail in the shadows... at the expense of blown out highlights. It seems to have a positive effect on the shadow noise (at least in case of my "test" shot). If the overexposure is very slight, you will be able to bring back a lot by PP. But if you go overboard with this, the penalties will be severe... lost unretrievable detail, hue shifts, loss of saturation, etc.
Most of my shots in which I increase fill light and brightness are landscape shots, where I have the choice (usually) to expose to the right more, which it looks like I would be better off doing if I upgrade to the 5dII from the 5d.
But this would bother me if I shot weddings or other dimly lit environments where one often can't expose to the right, simply because there isn't enough light to allow for an adequate SS.
I just bought a 5d2 and I have noticed the problem in a few shots without pushing them to extremes, thought I think it would take a rather large print to be evident.
I get the impression that the problem arises when you are right on the edge of dropping your shadows. It seems almost like the different channels get clipped at different levels, which could also account for the fact that it seems to show up in under certain lighting or in particularly colored areas of a photo. Perhaps this is an issue with the design of the Bayer filter?
On a side note, I just started playing with Capture One. My (very preliminary) impression is that it does slgihtly better with handling this issue than LR/ACR.
I think the banding problem may also be related to sensor temperature. Sensor temperature is effected by both ambient temperature and sensor use (e.g. live view, video mode, lots of shots).
Interesting thread. I am too interested in 5D2, so I've been gathering some info about this camera. I found a thread about banding in 5D2 but not in the shadow area...I am sorry if this already posted...here's the link
. Point is that the brightness slider is a midtone slider. Shadows are set permenantly set to no lower than +2 and I never use more than a +20 of fill flash and only that in very contrasty lighting.