brainiac wrote:
...and BTW, it's simply a scientific fact that the 7D needs sharper lenses than the 5D2 to get good per pixel sharpness. The interval of the 7D sensor is 4.3µ whereas the interval of the 5D2 sensor is 6.4µ. I hope you can understand that it takes a sharper lens to saturate detail in 4.3µ pixels than 6.4µ pixels. In fact the margin is quite big: a lens which just satisfies the 5D2 would have to get about half as sharp again in linear terms in order to satisfy the 7D. That's a lot of extra sharpness, and in any fair lens test we would regard the lens adequate for the 7D as absolutely destroying the lens adequate for the 5D2.
Of course, that's one of the major advantages of full frame: good performance from lesser lenses....Show more →
mfurman wrote:
I have only one question to ask, after reading numerous discusions, points of view and reviews of Canon EOS 7D:
Is EOS 7D a 18 Mpixel camera in the same sense as 5D is a 12.8 Mpixel camera and EOS 40D is a 10 Mpixel camera?
I guess you are asking, can you get the full 18MP out of the camera? Yes, of course. I'm doing a set of skin tone / portraiture tests today and the detail is amazing at ISO 100 (using the 17-55).
See my primitive analysis in my next post...
EDIT: just realized that I updated the first body to 1.1.0 firmware, but the 2nd still has 1.0.9. I will update the firmware and see if it seems to make a difference.
For anyone interested, have at it and tell us what you find. FWIW, this is an OOF image of a color checker that I displayed on my monitor. 1/10s exposure (to eliminate monitor flicker issues) at ISO 100.
I had committed to making this available from my 7D before I had looked for any mazing, G1-G2 issues, etc., so for anyone gathering statistics (ha-ha), this represents an unbiased sample (as long as you believe that I would have followed through w/out question once I committed, without regard to any other factors).
Actually, later this week I will post another, as I am getting a replacement for this body. I just got this one two days ago and discovered that there is a spot of something on the sensor that won't move under serious rocket-blowing; it ain't just dust. Other than that, though, the sensor appears to be pretty spotless from Canon.
IQ questions aside, this camera is fabulous.
Oh, w/regard to the shutter release mentioned by the previous post-er: I agree with the description, and add the note that to my subjective impression, it takes a fair bit of force to get the shutter to release, more than I would prefer. The BG-E7 grip has much more of a hair trigger. I wish I could average the two. I don't know how this compares to 1-series bodies....Show more →
Ok, so it's possible that in my self-induced sleep deprived state, I am making a silly mistake, but I don't believe this is the case. Each OOF color checker CR2 file was opened in DPP, and I just cranked the contrast and sharpness up to the max. Here are some results:
(Yes, these are viewed at 200% to make seeing the effect easier. It certainly is visible at 100% on the screen as well.)
(As noted above, I realized after taking the shot w/the second body that it had the 1.0.9 firmware, while the first body was running 1.1.0 firmware. So I did redo the second body shot w/1.1.0 firmware, but the results were essentially the same as seen here.)
I've processed the file in Raw Therapee, Cap One 5 and Bibble 5 and - with contrast and sharpening on full - it's not pretty, but no mazing in either the RT or the Bibble 5 conversion. There's something in the Cap One 5 conversion, but not really mazing either.
And at "normal" contrast/sharpening - no problems with any of 'em.
paulfeng wrote:
Ok, so it's possible that in my self-induced sleep deprived state, I am making a silly mistake, but I don't believe this is the case. Each OOF color checker CR2 file was opened in DPP, and I just cranked the contrast and sharpness up to the max. Here are some results:
(Yes, these are viewed at 200% to make seeing the effect easier. It certainly is visible at 100% on the screen as well.)
(As noted above, I realized after taking the shot w/the second body that it had the 1.0.9 firmware, while the first body was running 1.1.0 firmware. So I did redo the second body shot w/1.1.0 firmware, but the results were essentially the same as seen here.)...Show more →
Interesting. I took the difference G2-G1 of the first copy, and of the second copy, and applied a median filter to each of the results to beat down the column gain noise (which seemed to be about the same on both copies). Then I took the difference of these to see how much change there was in Green imbalance between body 1 and body 2:
Whiter means copy 1 had more imbalance, blacker means copy 2 had more imbalance. It seems pretty systematically that copy 2 is better, but not dramatically so -- about 1% improvement in (G2-G1)/(G2+G1). I'm surprised it makes that much difference in the conversion output, but there may be a nonlinear threshold beyond which the algorithm can't cope.
There is. Both DPP and LR use a minimum noise threshold per exposure level, specified per camera model. Detail under this level of contrast CAN be completely disregarded by the interpolation algorithm if there is no geometrical support for the detail being "real" and not "noise".
ejmartin wrote:
It seems pretty systematically that copy 2 is better, but not dramatically so -- about 1% improvement in (G2-G1)/(G2+G1). I'm surprised it makes that much difference in the conversion output, but there may be a nonlinear threshold beyond which the algorithm can't cope.
So this then implies that any G1/G2 imbalances have an accentuated (more than expected) effect on the maze pattern noise that many people are seeing? If so, than what may have appeared on the surface internally at Canon as a minor QC issue might have more real-world shooting implications than first thought. Especially when rushing a camera to market in advance of the Holiday shopping season.
Is the G1/G2 imbalance something that is best fixed by having Canon "calibrate" a given body, rather than having the various converters algorithmically deal with existing G1/G2 variances from body-to-body?
And once the these imbalances are "moderated," it is then, and only then that we realize the "full 18mp potential" for this sensor?
(Skibum--we're looking forward with great anticipation what you hear from Canon with respect to your 7D copy).
I was quite surprised at how much difference there is... one of those situations that makes you immediately think, 'did I do something wrong?' Similar to what Andy Westlake says over at DPR in the conclusion to his review of the new 100mm macro.
BTW, I only have DPP handy for now, so if anyone else wants to post conversions with other raw converters, please go at it. (Even though I can download some for free, I have some time constraints that make it unlikely that I will do so and futz about with them soon.)
I have continued to continued to evaluate my camera in the hopes of understanding exactly under what conditions I can rely on it. Paul, three observations that I have made are:
1. Noise (pattern) does not seem to correlate well to G1/G2 imbalance (so it may be something else).
2. Exposure is key to controling these effects. If you underexpose the image (.5 to 1 stop seems to cause problems at ISO 400, for example --- not always), you increase the chances of patterning. Given how you are pushing the sliders, small changes in exposure may be leading to your results.
3. JPG compression when you convert the image is interacting with the final results. You can change the result by converting to TIFF and going straight to Photoshop.
I think getting the second body was a good experiment. After the dust of this product introduction settles, I will pickup another one myself.
mfurman wrote:
Is EOS 7D a 18 Mpixel camera in the same sense as 5D is a 12.8 Mpixel camera and EOS 40D is a 10 Mpixel camera?
Marty DiBergi: Why don't you just make ten louder and make ten be the top number and make that a little louder?
Nigel Tufnel: [pause] These go to eleven.
As an aside, I've just read the 7D review in the UK's PhotoPlus magazine.
It is awarded 97%.
I've never seen a higher score in that magazine.
The reviewer raves about its image quality across the ISO range (the "harshest" thing they have to say there is that above 6400 ISO you lose some detail! ); the AF is as good as anything out there and better than most (a UK bird 'tog who contributes to the review says she is "at last" able to keep up with Peregrine falcons at speed); metering and exposure are "pretty faultless"...
And it goes on in the same tack throughout the review.
It also makes a point of acknowledging that it's very clearly a camera best suited to sport, action and wildlife photography - it's not a "low cost alternative to the 5D Mk II, it's a different kind of camera... better suited to subjects that need long telephoto lenses ".
In other words, while it will shoot landscapes, it is not a landscape camera.
keithreeder:
it's not a "low cost alternative to the 5D Mk II, it's a different kind of camera... better suited to subjects that need long telephoto lenses ".
Does it mean that it is working well with wide angle lenses?
brainiac wrote:
Marty DiBergi: Why don't you just make ten louder and make ten be the top number and make that a little louder?
Nigel Tufnel: [pause] These go to eleven.
I think that you understand what I was asking, after all :-)
The difference between 18 Mpixel and 21 Mpixel is not much. Can we compare these two cameras on a "pixel level"?
?? So much for evolution....
); the AF is as good as anything out there and better than most (a UK bird 'tog who contributes to the review says she is "at last" able to keep up with Peregrine falcons at speed); metering and exposure are "pretty faultless"...