This appears to only happen when really trying to adjust contrast in a poorly shot picture.
what is the percentage of pictures that are metered this bad?
btw I have taken some awsome pictures at iso of 6400 that I would have never been able to use on my 20D..
I have shot a good deal of pictures with my 5dm2 and have yet to make that big of an adjustment to my curves... the only way I can make this banding is to make severe adjustments and frankly if I took that many bad pictures I would just go back to a point and shoot camera as this would not be a good hobby for me...
just my personal opionion and if this is not a problem on other cameras I can understand the frustration...
The point is this really does not affect me.
I hope canon can help out the people that see this as a real problem though or else canon may loose a few customers and some reputation over this.
jcrawford wrote:
what is the percentage of pictures that are metered this bad?
I'll try to explain this again.
Many types of shot contain a range of Exposure Values which exceed the dynamic range of the camera, even when camera/raw contrast is set to minimum. If the photographer is unable to take multiple bracketed exposures because subject is moving, or no tripod is available, then it is not possible to devote hours of your life to composing 'high-dynamic-range' blends on the computer. As a result the photographer is forced to do the best she can with the existing dynamic range of the camera.
When blown out, highlights can look ugly, but because many cameras have an ability to store good tonal image data deep into the shadows, it is possible with such cameras (e.g. 1Ds3) to make a fairly dark exposure which retains detail in strong highlights, and then use a curve in processing to raise the exposure of the shadows towards the middle where they belong according to the naked eye. The effect of this is to dramatically extend the effective dynamic range of the image, and it works very well on a 1Ds3, because the deep shadows have good tone and colour accuracy and very low and random noise which does not catch the eye or suggest structure. Noise reduction software can be used to clean up the shadows when this technique is used on a mk3. The result is fantastically tonal results which capture quite wide dynamic range.
Note that the technique I am describing here IS NOT DIFFERENT TO "EXPOSE TO THE RIGHT", because I am suggesting exposing such that the brightest areas of the scene extend to the far right of the histogram and not beyond. This technique IS 'expose to the right', NOT UNDEREXPOSURE. When small bright highlights are very much brighter than the rest of the image, and remain within the camera's exposure range, the majority of the picture can look very dark. That is why it is necessary to push the shadows - because the exposure is correct, but the shadows are much darker than you would like them, as dictated by the highlights. Using a curve which rises rapidly in the shadows and flattens in the highlight region the shadows can be dramatically lightened in processing while detail is preserved in the highlights. Here's an example of a shot of this type: http://cyberphotographer.com/5d2/turnersreachb.jpg
This technique is difficult to use on my 5D2 because it has much more noiticeable and structured noise in the shadows at low iso and low contrast. Effectively my 5D2 has far less dynamic range than its sibling the 1Ds3, even though Canon stated that the 5D2 would have superior image quality.
Please understand that the advice "don't underexpose your pictures so much" is not helpful. The criss-cross patterns are stopping skilled post-processors from achieving the same dynamic range from the 5D2 that they can get from the mk3's.
The criss-cross gauze-like patterns arise in many situations and they are not exclusively the result of HTP or ALO.
If people are going to compare the results of their bodies, can I make a suggestion so that we can have a uniform basis for comparison.
Perhaps people should shoot a frame with no lens, the body cap on at ISO 100 for some predetermined shutter speed, say 1/50s or 1s or whatever folks agree upon, standard picture style. Then take the file, process in DPP boosting the exposure by 2 stops. Take the resulting image into Photoshop and auto adjust levels. You should get noise soup. The idea is not the amount of noise but the to see if there is structure in the noise. If everyone does the same thing, then we can get a more consistent idea of how many bodies (and serial number ranges) are affected and to what degree. Folks with 1Ds3s can do the same. Any takers or am I just out to lunch?
I think we have established the structure of the noise already. We just want to know how many stops it can be pushed before noise become too pronounced, whether there is a sample variation, and whether ISO200 or perhaps even ISO400 is more pushable than ISO100.
thedigitalbean wrote:
If people are going to compare the results of their bodies, can I make a suggestion so that we can have a uniform basis for comparison.
Perhaps people should shoot a frame with no lens, the body cap on at ISO 100 for some predetermined shutter speed, say 1/50s or 1s or whatever folks agree upon, standard picture style. Then take the file, process in DPP boosting the exposure by 2 stops. Take the resulting image into Photoshop and auto adjust levels. You should get noise soup. The idea is not the amount of noise but the to see if there is structure in the noise. If everyone does the same thing, then we can get a more consistent idea of how many bodies (and serial number ranges) are affected and to what degree. Folks with 1Ds3s can do the same. Any takers or am I just out to lunch?...Show more →
makes sense
since currently it is hard to know whether some people are just pushing PP and shooting HDR with lower contrast settings or whether some bodies really are much worse than others for pattern noise
All processed in CS4 with all updates applied to date, all settings set to 0 including sharpening and noise reduction. Once converted, they were saved at "12" (100% quality) as jpeg's.
I'll post links of the two large photos because they're actual size, not downsized.
OK I just took a couple of shots. Will have them up in 10 minutes or so. But must admit they are MUCH cleaner than what you got there. I am using LR 2.3.
