Thx to all those who liked the JRIF, appreciate the comments ! :-)
To show the image quality and cropping ability of the 7D,here's a few crops at different % from images taken today with a 300 2.8 and 2x tele. All were hand held, exif is intact(in the jpeg's attached). (ISO 400, AI-Servo, Manual point mode). I can routinely crop an image to 50% with this camera and maintain enough detail to make a huge print. Settings applied to both the images are shown in screenshots. I usually sharpen in CS3, but for this test, I used DPP all the way.
This camera does demand the highest quality lenses you have, that's my opinion. :-). I've thrown everything I could imagine (humans, airplanes,mammals, birds, cars, BIF,fast action) at this camera so far, and it hasn't burped, yet.
Zenon Char wrote:
Nice captures. I thought about getting the 2X tele. Would love a 500L but I need to win a lottery first. I heard using the 2X will not autofocus but your lens is a 2.8. My 300 is an F4. I guess going over 5.6 is the issue?
A 300 2.8 IS is one of the best lenses that can take a 2x, I won't recommend it on your 300 f/4. Btw, the cropping ability of the 7D should give you pretty high quality images with a bare 300 too (one factor being your subject size, of course). Yes, 7D will only autofocus till 5.6.
Thanks kirry007 . That's what I thought. I am getting very nice results with my 1.4. Just wanted to see if I could squeeze a little more out it. I have a Kenko 1.4 for stacking but AF does not work.
I have spent a little time tonight looking at the artifacting problem again in DPP and C1. Here is what I am seeing with my 7D:
1. Problem occurs in both C1 and DPP.
2. Artifacting increases as ISO increases.
3. Artifacting is present to some degree in ISO 100 images and all images above this level. Although it is at it's lowest point at ISO 100.
4. Noise reduction seems to reduce the effect. But, the artifacts that you see after NR were there in the final form before NR.
5. Artifacts create rectangular patterns that are visible at 100% view. They give the image a rougher, more textured look than it should have ... even at normal viewing sizes.
6. Artifacts seem to be the most noticeable when they align either horizontally or vertically in the image. There seems to be a tendency to this pattern.
7. I wouldn't call it mazing as it is presented in DPP or C1. But, it may be the remnants of that patterning. I have no expertise or special insight into mazing.
8. If you bring conversions by DPP and C1 into Photoshop and place them on separate layers, then move the opacity from 0 to 100 you can study the differences between the two conversions. The same artifacts are in both images in the same spot. So the software is not making them up. But, there are differences in how noticeable a given pixel in the artifact might be.
9. Careful exposure seems to really pay off big. If you under expose and bring exposure up in PP, then it appears like you end up generating many more artifacts and they will be significantly more noticeable. This was the last area I was analyzing and I feel that there is more to this story. This may explain one thing I noticed when shooting: that the degree of artifacting was not constant and could vary from image to image. It is possible that I was adjusting exposure in PP and this was impacting the amount and degree of artifacting in the images. Also, the images I was using for evaluation initially were shot under tougher lighting conditions. Using a careful expose to the right strategy seems like (with a few more tests) a way to get the best results here. I think that this is the best approach for avoiding as much of the problem as possible at this point. Further testing may uncover a workable solution here.
10. After spending more time on this problem than I really wanted to, I am left with a general feeling that this artifacting problem is somehow tied into the ability to get better ISO performance.
I would consider it a new class of digital noise. One that I don't really understand or like. Perhaps Canon is using some new approaches in processing the data going into the RAW file and the mathematics of processing these files is still not optimized.
It seems that some of the 7D buyers are not seeing this artifacting. I don't know what to make of that (outside of the expose to the right comments above). The only explanations that I can think of is that our cameras are performing differently, our personal expectations are different or maybe people have not spent enough time working with images to start to see the problems.
I don't really know what I am going to do. I can't return my camera at this point in time. I guess my options are:
1. Buy another 7D to compare to mine and see if I get different results.
2. Sell my 7D and use my 5DII exclusively. Prior to the introduction of the 7D, I had been leaning toward just selling all of my crop stuff and going completely FF anyway.
3. Wait and hope Adobe pulls off a miracle with ACR 5.6 ... or that Phase One or Canon tweaks their software further. I feel that the current release of DPP and C1 are production releases. My own analysis seems to indicate that Adobe has some chance of creating a conversion that is better than DPP or C1. This could reduce the problem further.
4. Wait and give this problem a little time to be better understood. Maybe call CPS in December or January and see if they have any insights after the dust settles down. Send in camera for repair.
I will say that outside of this IQ problem, the 7D is my favorite Canon body that I have owned since I started shooting Canon in 1987.
UCSB wrote:
I have spent a little time tonight looking at the artifacting problem again in DPP and C1. Here is what I am seeing with my 7D:
1. Problem occurs in both C1 and DPP.
2. Artifacting increases as ISO increases.
3. Artifacting is present to some degree in ISO 100 images and all images above this level. Although it is at it's lowest point at ISO 100.
4. Noise reduction seems to reduce the effect. But, the artifacts that you see after NR were there in the final form before NR.
5. Artifacts create rectangular patterns that are visible at 100% view. They give the image a rougher, more textured look than it should have ... even at normal viewing sizes.
6. Artifacts seem to be the most noticeable when they align either horizontally or vertically in the image. There seems to be a tendency to this pattern.
7. I wouldn't call it mazing as it is presented in DPP or C1. But, it may be the remnants of that patterning. I have no expertise or special insight into mazing.
8. If you bring conversions by DPP and C1 into Photoshop and place them on separate layers, then move the opacity from 0 to 100 you can study the differences between the two conversions. The same artifacts are in both images in the same spot. So the software is not making them up. But, there are differences in how noticeable a given pixel in the artifact might be.
9. Careful exposure seems to really pay off big. If you under expose and bring exposure up in PP, then it appears like you end up generating many more artifacts and they will be significantly more noticeable. This was the last area I was analyzing and I feel that there is more to this story. This may explain one thing I noticed when shooting: that the degree of artifacting was not constant and could vary from image to image. It is possible that I was adjusting exposure in PP and this was impacting the amount and degree of artifacting in the images. Also, the images I was using for evaluation initially were shot under tougher lighting conditions. Using a careful expose to the right strategy seems like (with a few more tests) a way to get the best results here.
10. After spending more time on this problem than I really wanted to, I am left with a general feeling that this artifacting problem is somehow tied into the ability to get better ISO performance.
I would consider it a new class of digital noise. One that I don't really understand or like. Perhaps Canon is using some new approaches in processing the data going into the RAW file and the mathematics of processing these files is still not optimized.
It seems that some of the 7D buyers are not seeing this artifacting. I don't know what to make of that (outside of the expose to the right comments above). The only explanations that I can think of is that our cameras are performing differently, our personal expectations are different or maybe people have not spent enough time working with images to start to see the problems.
I don't really know what I am going to do. I can't return my camera at this point in time. I guess my options are:
1. Buy another 7D to compare to mine and see if I get different results.
2. Sell my 7D and use my 5DII exclusively. Prior to the introduction of the 7D, I had been leaning toward just selling all of my crop stuff and going completely FF anyway.
3. Wait and hope Adobe pulls off a miracle with ACR 5.6 ... or that Phase One or Canon tweaks their software further. I feel that the current release of DPP and C1 are production releases. My own analysis seems to indicate that Adobe has some chance of creating a conversion that is better than DPP or C1. This could reduce the problem further.
4. Wait and give this problem a little time to be better understood. Maybe call CPS in December or January and see if they have any insights after the dust settles down. Send in camera for repair.
I will say that outside of this IQ problem, the 7D is my favorite Canon body that I have owned since I started shooting Canon in 1987.
EDIT: i have some hope acr 5.6 might give pretty much the detail of 5.5 but with much less mazing
hmm my last post went who knows where
not gonna retype
but yeah it does seem to be there at higher iso too just the higher NR used there and more noise alters it a bit
DPP may get a bit more detail per noise at high iso but it also looks very unfilmlike and artificial to my eye
DPP and C1 should maybe filter the extreme white and black dots better like ACR does
it just might be possible to get C1v5 to give ISO 100 pics with very very close to the detail as ACR 5.5 and without nearly so obvious mazing and if you sharpen the two versions at that point the ACR one falls into maze-city while the C1 does not
actually ACR 5.6 beta is showing the most promise, now the conversion on DPR look softer than my ACR 5.5 conversions for sure, but it is probably just the settings they used and a little post-sharpening gets it close and yet no mazing
anyway rambling here since not wanting to re-type it but there is some promise and AR 5.6 beta also seems to get less moire colorations and artifacting than C1
C1v5 at higher iso can be set to look almost exactly like what DPP gives less the colors and almost seem to be the same
UCSB wrote:
I have spent a little time tonight looking at the artifacting problem again in DPP and C1. Here is what I am seeing with my 7D:
After spending more time on this problem than I really wanted to, I am left with a general feeling that this artifacting problem is somehow tied into the ability to get better ISO performance.
I would consider it a new class of digital noise. One that I don't really understand or like. Perhaps Canon is using some new approaches in processing the data going into the RAW file and the mathematics of processing these files is still not optimized.
I will say that outside of this IQ problem, the 7D is my favorite Canon body that I have owned since I started shooting Canon. ...Show more →
+100. I could not agree more with the above.
Like you, I'm somewhat at a loss as to what to do next.
I will be keeping my 7D, despite this fairly large disappointment with respect to IQ.
10. After spending more time on this problem than I really wanted to, I am left with a general feeling that this artifacting problem is somehow tied into the ability to get better ISO performance.
I would consider it a new class of digital noise. One that I don't really understand or like.
This is my observation too.
I'm sticking for now with my original supposition that Canon intentionally introduced the green channel imbalance to combat noise.
The downside of this technique is the mazing effect - or, for those converters that can deal with it, the new class of noise and overall image harshness.
I will say that outside of this IQ problem, the 7D is my favorite Canon body that I have owned since I started shooting Canon in 1987.
Agree completely.
To me, the 7D is a fantastic camera ... with an IQ issue.
Granted, this an issue for IQ purists only but still.
It really seems the issue has a lot to do with gain fluctuations along columns, perhaps combined with a different spectral response for the two greens. I compared skibum's original 7D color chart with one I had lying around from calibrating my 1D3. Here are squares 19 and 21 (white and middling gray), 7D on the left and 1D3 on the right (the relative sizes are the jump from 10 to 18MP; both charts filled the frame on the respective cameras):
The bottom row shows the Fourier transform of the square above. The way to read it is that uniform speckling is random noise; the brightness of the horizontal line is the amount of vertical banding, while the brightness of the vertical line is the amount of horizontal banding; the extent to which the lines are bright near the edges is the extent to which the banding varies strongly from column to column or row to row.
My interpretation is that the 7D has much stronger vertical banding (note the bright horizontal line goes all the way out to the edge of the Fourier plot, while the 1D3 tapers off well beforehand -- it has some banding but it is lower amplitude, and lower frequency which would not affect interpolation which is a fairly local operation), and indeed it can be seen above the shot noise that is present even in midtones and highlights, and in the direct image without resorting to the Fourier space results. Note that this is not a color channel imbalance issue, since what is displayed is for a single green channel.
Interesting to note on the C1 7D crop, in the uniform red patches the misinterpolation artifacts are quite predominantly in the vertical direction. This would agree with the line noise observation in my previous post.
Interesting to note on the C1 7D crop, in the uniform red patches the misinterpolation artifacts are quite predominantly in the vertical direction. This would agree with the line noise observation in my previous post.
Might ACR 5.6B be reading the mask area and trying its best with the limited info there to try to balance it column to column?
I did some extreme sharpening up of an ACR 5.5 conversion of the DPR scene and found that even in detailed areas the mazing is still causing havoc (with DPP a bit too) and little fine lines of Pluto's fur and the fur balls near him are actually purely mazing artifacts, little perfectly horizontal or vertical lines of 'fur' and not real detail at all, comparing to a sharpened up DPR 5.6B sample you can see where the actual fur goes. So a standard conversion bungles up even highly detailed areas at times too not just smoother ones.
I'm not quite sure, but it might be possible to get reasonable detail out of C1v5 while minimizing artifacts and perhaps with 5.6B too though, although it is hard to really tell from a posted JPG in the 5.6B case. It's tricky to try to even up micro detail and noise to make sure it is not just averaging bits of detail away. I'm getting pretty sure you can do better than 12MP detail while removing mazing though.
Interesting to note on the C1 7D crop, in the uniform red patches the misinterpolation artifacts are quite predominantly in the vertical direction. This would agree with the line noise observation in my previous post.
Your noise spectra correlate well with what I see. More popular said, there are some very small traces of vertical banding in normal exposures.
I think perhaps what is being called "mazing" is not unique to the 7D
I examined old images from the 20D, 40D, XTI, and 5D. I could see a similar "mazing" in some images. Many of those were DPP processed, and some ACR. 5D imges were processed with ACR.
I shot some landscapes today with the 7D & 17-40 @ ISO 100
DPP conversion with no sharpening or NR. NR with neat image in CS4 and shart sharpen in CS4. The images look very clean in the sky. Essentially the same as a 5D landscape shot.
I re-processed some birding shots also. They were shot at various ISO settings up to 400. I used DPP but did not sharpen or remove noise in DPP. Instead I opened them in Photoshop CS4 and used Neat Image for NR and smart sharpen for sharpening. The resulting images have cleaner skies then the DPP only images. The images also have a less of a harsh tendency. Even with the bright lighting so common here in New Mexico. It seems Neat Image & Photoshop CS4 do a better job then DPP. I would process all in CS4 if it had more then beta support for 7D.
So to summarize, I think some of this is due to processing. I think much of this may get resolved in software updates. ACR v5.6 or later, C1, DPP.
All in all, I sure do like the 7D overall. I do not even use the 5D anymore.
abqnmusa wrote:
I shot some landscapes today with the 7D & 17-40 @ ISO 100
DPP conversion with no sharpening or NR. NR with neat image in CS4 and shart sharpen in CS4. The images look very clean in the sky. Essentially the same as a 5D landscape shot.
I re-processed some birding shots also. They were shot at various ISO settings up to 400. I used DPP but did not sharpen or remove noise in DPP. Instead I opened them in Photoshop CS4 and used Neat Image for NR and smart sharpen for sharpening. The resulting images have cleaner skies then the DPP only images. The images also have a less of a harsh tendency. Even with the bright lighting so common here in New Mexico. It seems Neat Image & Photoshop CS4 do a better job then DPP. I would process all in CS4 if it had more then beta support for 7D.
So to summarize, I think some of this is due to processing. I think much of this may get resolved in software updates. ACR v5.6 or later, C1, DPP.
All in all, I sure do like the 7D overall. I do not even use the 5D anymore. ...Show more →
Thanks for sharing your experiences...
RE: "I think much of this may get resolved in software updates."
I believe (and hope) you are right, but I wonder and question "at what cost in real image resolution will we have to pay?"
Will we need to sacrifice and realize only 10mp effective resolution? 12mp? 14mp?
LCollector wrote:
My observation:
I find it very interesting that most of you scientist(minds) never post any images taken with the 7D yet bash the camera based on some stock images and flowcharts and numeric tables, none of you even own the camera yet (well, some of you do), but I don't see any good or bad "real life" pictures to support your claims. Is this a storm in a teacup? Leave the camera alone! , Go out and please take some pics....
with ya man...over 3300 posts and very little to no real pics...you know the reason you are into photography is to take pics..I think maybe Ive been doing it wrong like LCollector says bring on the pictures already...
I dont care if its Babies,kittens,puppies ect
BTW I got flamed earlier for the same concerns
I compared RAW & mRAW
It may be the mRAW images are cleaner. I could not see any noise in them. But did not reach any conclusions. I do not see noise in all images.
Only DPP can process mRAW at this point in time.
call me crazy, but I would be OK with a firmware option that re-mapped the 18mp into RAW of 12 MP, mRAW of 10 MP, and sRAW of 8 MP
Even just using mRAW of 10 MP. You still get output of 40D or 1D Mark III
What I wanted most was a camera that could focus coming from 5D.
Cameron 12X said:
Thanks for sharing your experiences...
RE: "I think much of this may get resolved in software updates."
I believe (and hope) you are right, but I wonder and question "at what cost in real image resolution will we have to pay?"
Will we need to sacrifice and realize only 10mp effective resolution? 12mp? 14mp?
. Settings applied to both the images are shown in screenshots. I usually sharpen in CS3, but for this test, I used DPP all the way.
, Go out and please take some pics....
like LCollector says bring on the pictures already...