cameron12x wrote:
These "tests" definitely provide value, but what would make this all more understandable (and perhaps more germane and usable) to folks is to shoot "real-world" images side-by-side, comparing the output from each camera.
yeah, I sold all my other APS-C bodies long ago though so all I have is the test chart that I can replicate.
Nick Nishizaka wrote:
I agree with the others in saying that this has not affected my shots at all and I never view my images at 200%. Not to say that the problem isn't there...but what I don't see won't hurt me!
The OP used 200% as an example (I'm not sure why), but Skibum's test samples are actually shown at 100%.
abqnmusa wrote:
You can see this "mazing" in any camera viewed at 200% plus. I saw it in the full frame 5D, 20D, 10D, 40D, 7D, D3, D300. But not at 100%
Looking at at image beyond 100% is just created artifacts that are not in the data. There is nothing beyond 100% in the image.
that is ridiculous that would be like saying printing at different sizes of viewing on different monitors makes up RAW data
(note I didn't use interpolation, just 1 pixel to 4 pixels straight blow up)
and it just makes it clearer that the harshness is due to mazing and not just general noise or something else and prevent jpg compression from altering the patterns
and as for you can see it any camera, well I never noticed it in any images I processed from all those other cameras and I noticed it in the very first 7D image I ever procesed (the IR test shot) and even when i applied crazy amounts of processing I still could only make out the barest hints of it with the other cameras, if even.
some 7D do seem to resist it noticeably better than others and only marginally worse than the other canons so it might not matter with those copies
and yes i know at the end of the day, for many photos it's lighting that matters more than what camera or lens or anything
for my usage, since I do already have a 5D2, I would be doing lots of heavy crops since the 7D would mostly be my "reach" body and i'm not so sure that my old 50D didn't do better at that (although it did have worse AF, was slower, etc.) than my 7D copies (although some copies like pauls might do better)
and do we really want canon to go this way with future bodies where the exact raw converter is now suddenly a critical issue and unless you buy 4 bodies and cherry pick the best one, you need to mess around with trading off fine detail vs. artifacts when a little better QC might mean just tossing files around however you like as with the previous releases and not have some people ending up with copies with 80-100% worse noise at some ISOs than other copies (which is a huge, huge variation compared to any previous canon release)?
NCAndy wrote:
Is the mazing visible in jpegs straight from the camera? Or is this a RAW conversion issue?
much less in jpgs since canon's algorithm tends to break it into dotted lines instead of solid lines to begin with and also doesnt try to extract as much detail for lower contrast areas and jpgs tend to sort of smear details away compared to whatever you do with RAW files
paulfeng wrote:
I don't know that it's fair to dismiss some of these discussions as "complaint" threads. I very much want my 7D to perform as best it can. US$1700 is a lot of money for me, and I think the camera should work up to its potential.
In comparison, when I spend $4 on a mail order iPod FM car transmitter, and it turns out to be nearly useless, I chalk it up as a lesson to myself and I don't start a thread on an iPod enthusiast website (nor do I bother returning it for a refund).
Yes, complaint was probably the wrong word. I agree that a new camera should work well in both mechanical and imaging aspects. I have my fingers crossed.
Your'e just seeing the software rendition of the data once you go beyond 100%.
Do a conversion to highest res JPEG and look at it 100% and print it.
Dan
This is not true, at least not for the 200% crops I have shown. The 200% crops I have shown are not software upsampled interpolations (which I agree may display features that are not literally in an original image).
The 200% crops I have shown are lossless png screen grabs of DPP displaying the image at 200%. In that situation, DPP shows each image pixel as a block of four identical screen pixels, simply to make each image pixel easier to see. It does not create interpolated detail. I have seen the mazing at 100% on the screen.
Note that I am not necessarily saying that in real use, that mazing is or is not a problem for final image quality. That is another question. (I believe that it can be in some circumstances, for some cameras.) But it is incorrect to say that viewing at 200% is creating artifacts that are not there at 100%.
NCAndy wrote:
Yes, complaint was probably the wrong word. I agree that a new camera should work well in both mechanical and imaging aspects. I have my fingers crossed.
Well, I have not been able to duplicate any of the severe 'mazing' results like the examples posted here by (member) 'skibum5'. Granted, I'm using DPP only which (presumably) seems to show less 'mazing' but even when I push the sharpening all the way over the top in DPP, all I get is a very randomized over-sharpening pattern. I suppose this is good news for me but I'd like to pose the following question: Unless I missed reading it in the replies, is the most severe 'mazing' limited to the use of 3rd party converters? Is anyone reporting 'mazing' in DPP under normal processing...and in DPP severe processing? Also, if you're not seeing 'mazing' in DPP and use the export to Photoshop function does the 'mazing' suddenly appear in Photoshop (when it wasn't visible in DPP)?
This is how I see the issue. Forget the maze pattern (it is pronounced in LR/ACR that does not officially support the 7D, but much less in other production converters) it is unlikely that you will see a maze in your images. What you might see is noise, rectangles. These artifacts can show up in your images. The artifacting will be more likely to occur as ISO increases. It may be more likely to occur if you under expose your image and then bring up exposure in PP.
If you want to see what can happen, I would suggest shooting some portraits at ISO 400 or 800. You might find the texture of the skin ruined in some of the images with the artifacting described above.
From my perspective, the bad thing about this problem is that you just don't know which images will be affected. So you start shooting a series of portraits or a family event and you just don't know which images will have enough of this noise/artifacting in them to make them of marginal quality.
DPP is OK at reducing the problem. C1 can work around it to some extent. The new beta of Bibble seems to be able to minimize the problem also. I am hoping that the production release of ACR/LR will minimize with the problem when it is release.
I'm still trying to understand how to avoid/minimize this problem. Things that seem to help ... shoot to the right, RAW converter selection and use can reduce, lower ISO can help. My current 100% fix is just shoot critical work with my 5DII.
I've had my 7D for about 3 weeks, and I think I've finally figured out the AF so it works great for me. So now I've succumbed to trying to reproduce the mazing. This is what I get in DPP and ACR:
msalvetti wrote:
I've had my 7D for about 3 weeks...So now I've succumbed to trying to reproduce the mazing. This is what I get in DPP and ACR:...
Mark
Your DPP results match mine exactly...what's interesting about the ACR results is how much more pronounced the 'mazing' is in the pink area than anywhere else.
msalvetti wrote:
I've had my 7D for about 3 weeks, and I think I've finally figured out the AF so it works great for me. So now I've succumbed to trying to reproduce the mazing.
I see some DNA strands in the ACR processing.
The biggest question that I have is whether excessive sharpening is required for processing the "soft" 7D raw images that some have reported.
And if excessive sharpening is required to bring out sufficient detail, then is it also counterproductive because it creates the "harsh" images (a product of the maze artifacting) which some people have also described?
M Vers wrote:
Sure does. Just out of curiosity what happens when you bump contrast/saturation levels in the DPP window to match ACR? Any difference?
Those four squares seemed to be the worst in the chart, so I showed those.
I'm not a DPP (or RAW) shooter; how do I bump the contrast in DPP above 4? I was trying to adjust the contrast in ACR to match DPP by eye, because ACR seems to have a much wider adjustment range. I think the contrast is the biggest difference, not so much saturation.
The short answer is that even with much less contrast in ACR, the pattern is obvious. This image is with the same sharpening (150/0.5) but contrast all the way down.
By the way, I left radius at 0.5 because increasing the radius reduced the mazing.
msalvetti wrote:
I think the contrast is the biggest difference, not so much saturation.
You can't bump up the contrast levels past 4 on DPP (my fault for not seeing the slider at 4). What I'm seeing is a difference in saturation, the ACR version being more saturated. TBH I don't think we'll see much of a difference in noise, if any, even if the files are equalized...I was just trying to remove all doubt
(Is there an emoticon for fingers crossed?)
