After examining both files at 100% there is much more detail in the mountain foliage in the 5D2 image than on the 7D. I would like to repeat the HDRs using the 7D RAW files and compare them again.
I don't like to pixel peep but it is clear to me at least that the 5D2 image has more detail than the 7D image, but that is not the point of this post.
I think the 7D is capable of producing good landscape photos, but if the OP is going to print super large I think the 5D2 is a better choice.
mfurman wrote:
By the way, what I am really waiting for now is a DxOMark test. I got really tired of reading all the squabbles about low ISO noise (or mazing artifact).
Another opinion of a very good photographer, I trust, is the one of liquidstone (Romy) who recently said (after testing production camera):
"Question: are you planning to buy the 7D?
Maybe, sometime in the future. I'm waiting for the price to stabilize a bit, and I wish that some RAW converters and/or firmware will come out to mitigate the noise in the shadows at ISO 200-800. The noise at mid-ISOs for the 7D unit I tested is a bit more than what's acceptable to me (I actually passed up on the 50D for the same exact concern).
If the mid-ISO noise is what it is (due perhaps to the limitation imposed by the tiny pixels), I might still buy the 7D if the pricing softens a bit. I'll use it as an ISO 100 - 400 birding body, and just mount my 5D2 when light goes down.
" ...Show more →
one thing to keep in mind is that DxO test PRE-demosaiced data and mazing noise doesn't show up until you go and demosaic, so unless DxO specifically writes a note about it or finds something in their color transform test then DxO won't tell you anything about it at all, if anything it might show slightly better performance pre-demosaic.
it does now seem that some copies fall under a threshold where the damage is not much and you can ignore it and the camera is pretty fine while others (quite many) over the threshold so the damage is, as even Romy says (assuming his complaints about noise are due to this, which is an assumption), a bit much.
I don't like to pixel peep but it is clear to me at least that the 5D2 image has more detail than the 7D image, but that is not the point of this post.
I think the 7D is capable of producing good landscape photos, but if the OP is going to print super large I think the 5D2 is a better choice.
hmm 18MP vs 21MP should not be that different
what aperture,shutter speed/mirror-lock/ISO/tripod/converter etc.?
(some have suggested that some converters might avg two channels thus lowering resolution on parts of the image to hide mazing)
it looks like you only have like 12MP resolution there
oh one thing is it is clear you did no CA correction and that will penalize the 7D more than the 5D2 so you should really apply CA correction
Both images where shot equally, one minute apart from each other, except that I had the 7D at 10mm(16mm) and the 5D2 at 17mm. I used on the 7D the 10-22 Canon lens and on the 5D2 I used the 16-35L v1 lens. The where captured in a 3 shot bracket, in AV at f/16. They where processed in Photomatix using the same parameters and they have the same develop setting in Lightroom on the resulting HDR Tonemapped image. I used no Noise Removal nor did I sharpen the images.
For the 7D HDR I used the JPG out of the camera, I also have the RAW files to redo the HDR when support for them is better available.
I didn't process the images all the way I would if I where going to print it, just the bare minimum. I didn't correct CA. The files where exported out of Lightroom with the long edge at 800 pixels to upload to flickr. The 100% crops where screen captures from Lightroom on my Mac. The originals have no crop in post processing.
If anyone wants to see the original JPG output I could find a way to upload them for anyone to see. I'm no professional and my comparison could very well be flawed, if anyone has any suggestions I could retest.
I'll have to take and show a series of shots - at f/16, you will have visible degradation on the 7D's shots when viewed at 100% due to diffraction. The pixel density of the 7D's sensor is high enough to show it at that aperture. It's not devastating, but it is visible compared to lower apertures.
hvilorio wrote:
Both images where shot equally, one minute apart from each other, except that I had the 7D at 10mm(16mm) and the 5D2 at 17mm. I used on the 7D the 10-22 Canon lens and on the 5D2 I used the 16-35L v1 lens. The where captured in a 3 shot bracket, in AV at f/16. They where processed in Photomatix using the same parameters and they have the same develop setting in Lightroom on the resulting HDR Tonemapped image. I used no Noise Removal nor did I sharpen the images.
For the 7D HDR I used the JPG out of the camera, I also have the RAW files to redo the HDR when support for them is better available.
I didn't process the images all the way I would if I where going to print it, just the bare minimum. I didn't correct CA. The files where exported out of Lightroom with the long edge at 800 pixels to upload to flickr. The 100% crops where screen captures from Lightroom on my Mac. The originals have no crop in post processing.
If anyone wants to see the original JPG output I could find a way to upload them for anyone to see. I'm no professional and my comparison could very well be flawed, if anyone has any suggestions I could retest....Show more →
well at f/16 and using different lenses, etc. you really can't make a comparison in terms of pure sensor quality
although i guess, since some people like 'real' world stuff only, this is very real world since if you need f/16 you need f/16 and that will blur the 7D more than a FF and if the 10-22 is what you have, it is what you have (but of course that is not fair either since that gives the APS_C landscape more DOF)
but in terms of whether the 7D really produces softer images in general you really can't determine anything from that test
thedigitalbean wrote:
The so-called "mazing" artifact is unrelated to pixel density and indications are that its a QC issue given the differences between bodies.
You are right – the "mazing" artifact is not caused by the pixel density itself.
The "mazing" results from a technique that Canon is using on the 7D sensor where the color filter is more transmissive to light - but at the same time more color-blind.
This color-blindness is correctable in software at the expense of small color artifacts.
QC issues are only exacerbating these artifacts, which in its worse form look like mazing in pictures.
Without the QC issues, the color artifacts would be smaller.
The color-blindness technique is used to combat noise.
If the 7D had a smaller number of megapixels, the color blindness technique maybe wouldn’t have been necessary and the color artifacts would have been avoided.
Btw, this technique so far has been used only on P&S cameras.
There is actually nothing wrong with having 18mp on an 1.6x sensor – provided that these are high quality, crisp, noise-free, and artifact-free megapixels. Doesn't seem that this is the case with the 7D.
If the new 1D4 is any indication, Canon has the technology to make a ~12mp 1.6x sensor with fantastically high, pro-level image quality (the 1D4 has the pixel density of a 10mp 1.6x sensor).
For some reason they chose 18mp with so-so image quality.
To me, 12mp with fantastically high image quality beats 18mp with so-so image quality any day.
But I guess Canon felt that so-so image quality is OK for a $1699 camera .
mfurman wrote:
This "degradation" is more than what could be caused by diffraction, I think. How about the noise level difference? It is only ISO 200.
Quite possible. Or perhaps the in-camera JPG's aren't quite up to par (sharpening?). The noise is certainly visible - was in-camera NR activated? I know that if NR is totally disabled in DPP, the noise on the 7D is quite visible.
I'll do some toying around with my own 7D when I get some time. I've got the 7 and the 5D2 - I already did some high-ISO testing and found that the 5D2 does a better job with noise and detail retention at 3200 & 6400. But that's a different situation than here.
mfurman wrote:
The still unanswered question is if it is really a 18 Mpixel sensor, from the resolution point of view.
What part of the DPR review is not clear:?
******************
The Canon EOS 7D makes the most out of its 18 MP sensor and delivers the highest resolution that we have so far seen on an APS-C DSLR. There are some signs of moire at very high frequencies but it outresolves all other cameras in this comparison in terms of both absolute and extinction resolution.
******************
Quick diffraction test, 7D, ISO 400, 85/1.2 lens, tripod mounted, MLU, 2 second shutter timer. RAW converted in DPP, Standard picture style, Tungsten WB, Low NR, sharpening set at "3" (I forgot to turn it down, but the difference is a little more noticable here anyway). These are all 100% crops. I tested at f/5.6, f/8, f/11, and f/16:
7D vs. 5D2. I'll compare at f/5.6 and f/16. Same lens, same settings, same scene.
I did 3 things wrong with this test -
first - In my haste, I inadvertently had +1/3 stop exposure compensation on the 7D so it is slightly brighter. Second - I used a tabletop tripod for the 5D2 and didn't level my image quite as well as I could have.
Third - I did not resize either camera so these are 100% 800 X 533 pixel crops from each.
I do not think that any of these will detract from the results to a significant level that it could be deemed invalid.
Not a huge difference - yes, the 5D2 has more detail, but it isn't "night-and-day". I think a little careful sharpening on the 7D's image would make it really close.
Not sure what this test proves or disproves. Diffraction is a property of the lens, not the sensor. Printing or viewing, everything else being equal it's always advantageous to 'digitize' a diffraction-blurred image with a higher resolution sensor since it results in a higher overall lens+sensor resolution.
.