thanks for the generous comment...i cannot take the camera away from the table by far...it was not my camera and the 7D was not available for sale back then. try to take the subject lens or camera away from the table and i risk security chasing after me! i also did not PS the photo so that everyone sees a great image that the lens and camera cannot produce. I used ARC to get image then add my name to the photo in PS.
Its kind of silly to hear that a serious photographer needs to use FF to be serious. I use an APS-H 1D mk II as a sports photographer and just recently bought an APS-C 7D for sports photography. For studio work I use my 1Ds mk II. It depends on purpose/needs for the specific profession. I've seen very serious photographers take great studio photographs with the Nikon D70 and EOS 400D in China. I've seen many people with expensive toys like the 5D mk II FF camera in their hands asking me what are the controls beyond P mode.
DPP is the best raw processor for 7D at the moment. I used ARC that was not ready just just because i dont like the controls of DPP. if we all have to wait for ARC to be ready then no one will bother to write a review on the 7D.
Considering this person did more research than most of us, his input might be useful. A few things stand out:
1. The 50D, which is blasted for Pixel density and low light performance actually score best of all APS-C (7D is not plotted in the graphs)
2. The Mk 5D2, another pixel beast is top in IQ and is close to the D3 in Light performance.
3. The AIQ of the MK3 is strangely low, which actually does make it plausible an APS-C can in the future reach that IQ.
4. Read noise of the 50D and 5D2 are half that of previous gen Canon and Nikons. This should help in low light.
The only consistent thing is that bigger sensors are better, but more MP doesn't seem to hurt low light sensitivity, at least it doesn't show in any of his graphs.
And before the flames go up:
- The 7D isn't tested by this person, so it's impossible to say if it would plot better or worse in this graph than competitors.
- Proof is in the pudding, I have seen brilliantly sharp MKIII shots, so these graphs are just that: graphs.
But what it does support is that more MP=Bad is disputable. In fact I haven't found a thorough test that proves otherwise.
Roger's a good guy, but his AIQ is a somewhat arbitrary melange of different image quality characteristics, just as DxOMark is. And just as for DxO's data tables and graphs, it's much better to look at the individual data figures for read noise and sensor efficiency (gain) and normalize them for pixel count and sensor size in order to draw accurate conclusions.
The only consistent thing is that bigger sensors are better, but more MP doesn't seem to hurt low light sensitivity, at least it doesn't show in any of his graphs.
True enough, but the myth of poor performance of small pixels persists.
And before the flames go up:
- The 7D isn't tested by this person, so it's impossible to say if it would plot better or worse in this graph than competitors.
Measurements on a couple of 7D's indicate that it has a gain of about 2.1-2.2 e-/14-bit ADU at ISO 100, meaning its sensor efficiency per unit area is about on a par with the D3 and about 15% better than previous Canons -- assuming that they haven't messed with the ISO normalization (we'll have to wait for DxO for an accurate figure on that). ISO 100 read noise is about 5.1-5.2 raw levels at ISO 100, or about 11 electrons -- quite a bit lower than say the 50D, though some copies seem to exhibit pattern noise issues. ISO 1600 read noise is about 4.4-4.8 RAW levels, or about 2.2-2.3 electrons -- better than the 50D on a per pixel basis, about the same on a per area basis. And of course much less banding in the high ISO read noise.
ejmartin: ISO 1600 read noise is about 4.4-4.8 RAW levels, or about 2.2-2.3 electrons -- better than the 50D on a per pixel basis, about the same on a per area basis.
Do not mean to focus on that too much but would you be able to say if, as far as noise is concerned, 7D is better than 5D mkI? Why does 5D dynamic range of 5D is so poor at low ISO?
elfanucchi wrote:
quote]gfiksel wrote:
May I ask you, did you take the statement "Small pixels are not Hardy pixels" from a textbook on "photon physics"? If yes, can you give me the link, please?
You can check any number of current textbooks on Semiconductor engineering design - warning they are college and quatum math level.
Its basicly signal vs noise and leakage currents.
I have been out of college for over 30 years so dont think you want you want the few I have left.
Actually you should check current or take some optics and engineering courses.
Thats where folks learn this kinda stuff. Not Photo forums.
There would be way too many pages to scan and dont think many could understand the formulas.
Alternatively, you could just look at actual fair tests done between concurrent cameras with radically different sensor pitches to see if your armchair quantum physics misunderstanding has any truth in it. If you do that you will see that a 1Ds3 (6.4µ) does iso 12800 just as well as a D700 (8.5µ): http://cyberphotographer.com/d700v1ds3/d700v1ds3_12800iso_b.jpg
And you will see that a Fuji FZ50 (1.97µ) does iso 13,500 just as well as a Canon 450D (5.7µ)
Could you please explain that? I always thought that sensor size mattered (even if we may agree that the number of pixels does not have to be detrimental for S/N).
mfurman wrote:
Could you please explain that? I always thought that sensor size mattered (even if we may agree that the number of pixels does not have to be detrimental for S/N).
I can try. What you are looking at is an image recorded by EQUAL AREAS of sensor, at equal magnification, and therefore it shows you how good an FZ50 _would_ be if it had an APS-C sensor. The outcome is that a 400D with a 1.97µ pitch using the Panasonic sensor design would be a better low light camera than the 400D is, even though the 400D has a wonderful 5.7µ pitch. It illutsrates that those people who think lower Mpixel sensors offer some kind of iso/noise advantage over denser sensors are just... wrong. The Panasonic sensor technology would give the 400D an 84 Mpixel sensor AND be better in low light at 13500 iso. Soon, hopefully, people will begin to understand that fewer pixels does NOT mean better low light performance. It's the myth that never wants to die, but we must all kill it if we want better cameras.
Please read this if you want to understand this better and avoid being misled by the large number of outspoken commenters who do not understand this issue but propagate and defend this falsehood like a religious doctrine: http://cyberphotographer.com/megapixelmyth
This is why I look forward to a 50 Mpixel DSLR. The sensor technology already exists and works well, and the benefits will be real and visible in large photographs. When used in good light, such a camera will provide great freedom to crop (i.e. carry fewer lenses).
brainiac:
The Fuji sensor technology would give the 400D an 84 Mpixel sensor AND be better in low light at 13500 iso.
Thank you. I admit that, I did not read the thread, you have linked to, carefully enough.
If we agree about the principles of the points, you are making, why we do not have any sensors with more than 24 Mpixels (I do not include Medium format) and some (Sony 900 Alpha) do not have great high ISO and dynamic range performance).
mfurman wrote:
If we agree about the principles of the points, you are making, why we do not have any sensors with more than 24 Mpixels...
Lens limitations, speed of data throughput in the camera, and cost.
...and some (Sony 900 Alpha) do not have great high ISO and dynamic range performance).
The Alpha uses the same sensor as the D3x, which is probably the very best DSLR ever at low isos, and performs pretty well at high isos once ouput has been appropriately processed. Here are crops from D3x and 5D2 at iso 12800: http://cyberphotographer.com/d3xv5d2/d3xv5d2_12800_crop.jpg
These are 100% crops of both files at 24 Mpixel, without noise reduction. With noise reduction at normal print sizes the differences are fairly minor.
mfurman wrote:
Do not mean to focus on that too much but would you be able to say if, as far as noise is concerned, 7D is better than 5D mkI? Why does 5D dynamic range of 5D is so poor at low ISO?
Thank you
Michael
As far as electronic read noise, the 7D is substantially better than the original 5D at both low and high ISO. Note however that the 5D has about two and a half times the sensor area and so for a given Tv/Av will collect that many times more photons -- photon shot noise at a given scale in the image gives about a 4/3 stop advantage to the 5D (perhaps a bit less, because the 7D seems to have a somewhat more efficient sensor).
Low ISO DR is limited by the electronic read noise added after signal amplification, in the ISO amplifier and ADC. Cleaner amplifier/ADC or other methods of reducing this noise have the potential of gaining of order two stops in low ISO DR (according to the engineering definition of DR). This is for instance a large part of the reason why the D3x scores so well in low ISO DR in DxO's testing.
mfurman wrote:
The question is why it is not better if we follow "higher number of Mpixels" is better theory.
Does Richard (brainiac) want to comment as well?
Don't misquote me. I am saying that the number of pixels squeezed onto the sensor has little effect on low light performance. I would expect the D700/D3/D3x/α900/5D2/1Ds3 all to have very similar iso 12800 performance after appropriate processing because their sensors are the same size and roughly the same age. My crops above show that 1Ds3 and D700, D3x and 5D2 do all have about the same performance at 12800. Sure the correlation isn't perfect, but the differences are pretty small.
Thank you. I guess, the problem is how the better electronics of 7D will overcome sensor advantage of 5D mkI. I saw some tests in which 7D comes very, very close to the original 5D, as far as noise is concerned.
