msalvetti wrote:
Thanks for the careful reply. Should be a very interesting fall! Looking forward to a new camera for the upcoming youth and high school hockey seasons. Just need to pick between a used MkIII and a new 7D. It may come down to the AF.
Mark
Keep in mind that the 1D3 also has a much larger RAW buffer. That could make a difference for some folks.
thedigitalbean wrote:
I've downloaded the Imaging Resource RAW files and have taken a look at them using a Beta version of Camera Raw. A few notes:
1. The criss-cross pattern noise you see when you push the shadows really hard on the 5D2, its gone. Awesome.
2. The ability to resolve detail even at higher ISOs is impressive, very happy to see this.
3. Per pixel noise at ISO 1600 is visibly and considerably lower with the 7D than with the 50D. When applying the default color noise reduction in ACR, much of the color noise goes away however the 7D has much less color splotches than the 50D. Definitely good news for those who weren't impressed with the noise performance of the 50D.
4. Things only get better for the 7D against the 50D at higher ISOs. At ISO 3200 you get banding and really ugly splotches in your shadows (when you apply color NR) on the 50D. The 7D has no banding and though some spotching starts to show its controlled.
5. How does it compare to a 5D2? Only compared the images briefly, but sadly the laws of physics still apply. The 5D2 has less noise at virtually all ISOs, however the difference isn't as huge as it was comparing the 5D2 to the 50D.
Based on my observations so far, I'm quite happy I pre-ordered and pumped to actually use the camera when it eventually starts shipping.
Now I really need to get back to working on the stuff I get paid to work on......Show more →
All I wanted from 7D was better pier pixel noise than 50D, and clean ISO 100-800 and good ISO 1600. For my intended useage I'd rarely go over ISO 800, I'll leave that to 5D II. Although I still think they'd have been better doing this development on the 15.1MP sensor, I'm not going to complain if they've managed to improve the noise at the pixel level and given me an extra 3MP for birding.
The 7D will mean different things to many people, but on the photography side when I think of the 7D I think mostly of non-static subjects (e.g. sports) and also for objects which require some reach (e.g. wildlife). That's why AF will be one of the keys to the ultimate success of this camera, IMHO. It's not that the other factors aren't important, but form and function are sometimes separated by FF vs. crop.
For FF landscapes, I can't wait for the 5DmkIII (or whatever it will be named) to arrive, hopefully no more than 6-8 months out from now. Exciting times, indeed!
thedigitalbean wrote:
I have been scrutnizing the files to try to find an answer to this myself but its hard to say. Comparing the 5D2 @ ISO 6400 vs. the 7D @ ISO 3200, the 7D is definitely cleaner, so the advantage to the 5D2 is definitely < 1 stop.
The difference becomes minimal at lower ISOs however. My guess would be the 5D2 is cleaner by 1/3 to 2/3 stops (maximum) depending on ISO. Most of the difference I see is in the highlights (which I find very odd). What is making the comparisons difficult for me is that the difference in color and contrast (which is most likely because the 7D hasn't been fully calibrated yet in my ACR beta, and they probably used different lenses). Whats interesting is comparing both at ISO 12800, though the 7D has more noise in the highlights, it has virtually no pattern noise, where as the 5D2 has distinct pattern noise....Show more →
the deepest shadows get few photons but read noise matters and the 7D has improved pattern noise while the highlights suffer from shot noise only and a larger sensor simply collects a heck of a lot more photons per area of print.
thedigitalbean wrote:
I personally am now satisfied the high ISO image quality of the 7D will be ok. Dynamic range and color fidelity is still a question. However the biggest question for me is how well the AF will perform. I guess I'll have to wait until I have one in my hands before evaluating that.
Maximum DR has already been tested, see my post just a page or two back.
It should be almost as good, normalized, as for the 5D2 (and arguably better since the 7D might pattern band less).
Although the IR images may appears to give the 5D2 only a 2/3 stop advantage for SNR this seems almost impossible to believe considering that the gapless microlens on the 50D compared to 40D does not give enough relative advantage. More realistically would be at least 1 stop advantage to the 5D2 and quite possible 1 and 1/3 stops, at least away from the deeper shadows.
^ I agree, I did a test of the 7D vs 5D2 last night for shadow noise, the 5D2 was about 1 stop better based on JPGs, I used RAW but unfortunately I can't open the RAW files at the moment.
If the 1 stop advantage is correct in RAW, then it means that at the pixel level the 7D is noiser than the 40D (which is only 0.67 stops worse than the 5D2) and similar to the 50D.
Imo downsizing the 7D file is not going to equal things, because the issue with the newer Canon cameras (including the 5D2) is chroma blotching. That blotching is still there even if you downsize to 10 or even 5MP. And YES it does show up in prints...badly...
thedigitalbean wrote:
Keep in mind that the 1D3 also has a much larger RAW buffer. That could make a difference for some folks.
I personally have my doubts that the full resolution 7D images will be as clean as the 1D3 at higher ISOs, but I would not discount the 7D MRAW setting. If the 7D behaves like the 5D2, noise may improve somewhat at higher ISOs using MRAW, and at that point they're both 10Mp cameras.
Dawei Ye wrote:
^ I agree, I did a test of the 7D vs 5D2 last night for shadow noise, the 5D2 was about 1 stop better based on JPGs, I used RAW but unfortunately I can't open the RAW files at the moment.
If the 1 stop advantage is correct in RAW, then it means that at the pixel level the 7D is noiser than the 40D (which is only 0.67 stops worse than the 5D2) and similar to the 50D.
Imo downsizing the 7D file is not going to equal things, because the issue with the newer Canon cameras (including the 5D2) is chroma blotching. That blotching is still there even if you downsize to 10 or even 5MP. And YES it does show up in prints...badly......Show more →
At the risk of repeating of myself, I'll say that comparing the RAW files, the 5D2 is definitely NOT 1 stop better than the 7D, I put the 5D2's advantage at between 1/3 to 2/3 stop. Furthermore, the 7D has visibly lower noise than the 50D at all ISOs 800 and higher.
Dawei Ye wrote:
^ I agree, I did a test of the 7D vs 5D2 last night for shadow noise, the 5D2 was about 1 stop better based on JPGs, I used RAW but unfortunately I can't open the RAW files at the moment.
If the 1 stop advantage is correct in RAW, then it means that at the pixel level the 7D is noiser than the 40D (which is only 0.67 stops worse than the 5D2) and similar to the 50D.
Imo downsizing the 7D file is not going to equal things, because the issue with the newer Canon cameras (including the 5D2) is chroma blotching. That blotching is still there even if you downsize to 10 or even 5MP. And YES it does show up in prints...badly......Show more →
hmm in RAW tests, most testers have found the 5D2 to be a solid 1.5 stops better than the 40D/50D for SNR (which makes sense considering the EDIT: oh you are talking pixel level
I found even at the image level (after downsizing) the difference was less about 1 stop tops. I have the full resolution downloads here in the test I did. I equalised everything between the tests (and I mean everything )
At the pixel level, it's about 0.67, whereas at the image level it's about 0.8 or so I'd say. I agree with DPReview in that downsizing doesn't make much difference to the noise characteristics
Dawei Ye wrote:
I found even at the image level (after downsizing) the difference was less about 1 stop tops. I have the full resolution downloads here in the test I did. I equalised everything between the tests (and I mean everything )
At the pixel level, it's about 0.67, whereas at the image level it's about 0.8 or so I'd say. I agree with DPReview in that downsizing doesn't make much difference to the noise characteristics
DxO and I believe Gabor too got more like 1.5 stops over the 40D for SNR at 18% gray, which really does fit, 2.6x surface area and reasonably similar sensor technology. I'm gonna try it myself and I expect that is waht I will see too. Note that they did at the raw RAW level so converter issues didn't come into play.
although for DR is can be much less difference at low ISO between the 2.
also note that the 40D exposes darker for a given ISO than a 20D or a 50D
i kind of disagree with DPR, they also resized JPGs that had already had some NR applied
Comparing noise per pixel is a red herring and helps generate mass delusions, unless you uprez to the a common number of pixels. This is an 18 Mpixel APS-C camera. It has a new sensor and the lower noise floor first seen in the 500D, and that is possibly improved too. Since it has 1080p, it illustrates that the pattern noise problem of the 5D2 was not a necessary by-product of doing video, and therefore was a designed threshold. You can see that the criss-cross pattern is effectively absent in low light 1080p video from the 7D, so it may compete well with the 5D2 for general shadow quality. Per pixel discussion is a waste of perfectly good bandwidth.
thedigitalbean wrote:
I just compared the 7D RAW files to the D300 NEFs (they don't have D300s files yet) and to my pleasant surprise, the 7D is on par at lower ISOs and cleaner at higher ISOs on a per pixel level.
Thanks for the good work so far.
Can you compare 7D RAW files to the D90 / D5000 NEFs?
The latter sensor is an improved version of what's on the D300 and will most likely appear on the D300s as well.
I know D90 RAW files look funny (the edges of objects break apart) at high ISO due to in-camera RAW manipulation.
I'm ambivalent, but leaning towards an early purchase. I still want to learn more about AF performance in the field in "real" settings, though. That's the game-breaker or deal-breaker for me.
The pixel peeping is nice, and provides value, but what really counts is how the camera provides real-world results in the hands of professionals or semi-professionals (its intended markets).
john660 wrote:
I'm interested to know who's planning on buying this 7D right away...or are most thinking they'll wait to see how it actually performs like I am?
I am getting one asap as I have done in the past with the other D series camera's. Only issue I ever had was with the 40D err 99 but that happens. I also bought one of the first groups of the 24-105 but it had the flare issues and it was a new problem and Canon replaced it. But it is always risky being Canon's tester
john660 wrote:
I'm interested to know who's planning on buying this 7D right away...or are most thinking they'll wait to see how it actually performs like I am?
I will be purchasing a 7D soon as my bank account permits. It will be a great camera for shooting in the water, as well as a backup camera. I have no doubt it will suit my needs in this capacity and 18mp is more than enough.
Can you compare 7D RAW files to the D90 / D5000 NEFs?
The latter sensor is an improved version of what's on the D300 and will most likely appear on the D300s as well.
I know D90 RAW files look funny (the edges of objects break apart) at high ISO due to in-camera RAW manipulation.
I did a quick comparison to D5000 NEFs, the D5000 has less noise per pixel at ISO 800 and above, however I see some strange artifacts in the D5000 images at higher ISOs (especially ISO 6400). Can't be sure but they sure look like noise reduction artifacts.
)