Ben Horne wrote:
I "fail to understand"?? Really? Is this the way you're seriously going to start this comment? That's pretty lame don't you think? I don't think you really understand my statement. I don't see any color definition, and I don't see any 3D feel to the image. It looks flat. I don't care what kind of lighting situation you are in... my comments are regarding the fact that the noise reduction is robbing the image of these characters that it SHOULD have. There should be more defination to the subject. The noise reduction is robbing the subject of this sort of detail, leaving it looking rather dull.
When I read the title of the post, I was expecting to see an amazing ISO 3200 shot, and I all I see is a ho-hum high ISO shot that looks like any other camera. If you want to see some more amazing high ISO shots, take a look at the D700 or the D3. At 3200, they seem to have more depth despite the heavy noise reduction....Show more →
No but your jacked up response is pretty lame IMO. But I'm waiting for your examples under the same conditions to show me how much better you can do it. Talk-the talk, now walk-the-walk.
I think it will happen sooner than you think, thank god, because now we have two formats. Not too long ago, we had 1.5 / 1.6 crop, and then the oddball FF cameras. Now that people seem to be making peace with the fact that we have the FF and 1.5/1.6 camps pretty much established and here to stay, the lightbulb may come on even in the DPR world.
Just because they wrote that spinach had tons of iron doesn't make it true, even if the whole world repeated it for 50 years.
stanj wrote:
I think it will happen sooner than you think, thank god, because now we have two formats. Not too long ago, we had 1.5 / 1.6 crop, and then the oddball FF cameras. Now that people seem to be making peace with the fact that we have the FF and 1.5/1.6 camps pretty much established and here to stay, the lightbulb may come on even in the DPR world.
Just because they wrote that spinach had tons of iron doesn't make it true, even if the whole world repeated it for 50 years.
Indeed. To which I will add that just because DPR, BobAtkins, DCP,etc... all make the same flawed comparsion doesn't make the comparison any less flawed. In fact, I would say it behooves those who do know better to spread the word rather than just go along with it.
Would you say that comparing 100% crops is the best way of assessing sensor noise?
Yes, I would, but ONLY if you remember brainiac's rule: always uprez files to the same number of megapixels before comparing 100% crops.
If you forget to observe brainiac's rule, then comparing 100% crops is positively the worst way of comparing the sensor noise from two cameras, because it is reliably and methodically biased in favour of the lower resolving camera.
Although comparing 100% crops depends to some extent on your monitor, the variation and calibration of monitors isn't generally worse than the variation and calibration of printers, prints, inks and papers, and not all print sizes and processes resolve enough to give you a general insight into the relative noise qualities of any two cameras. Also, comparing crops (after observing brainiac's rule) is much quicker and cheaper.
scott f wrote:
Not having followed a lot of this, so sorry if this is a really dumb question, if you are saying that you must resize the files to the same dimensions to evaluate noise properly, does it matter if you uprez or should you always downrez the higher MP file?
To save argument, I will say that you should do both, and compare at each res, but it really depends on iso and subject matter. At 12800 iso, there is very little detail at the 21 megapixel frequency, so downrezzing doesn't discard much. The lower the iso, the more the camera is able to retain real image detail. At 3200 iso I would uprez the lower Mpixel file to compare it to a 1Ds3/5D file. It can be a good idea to slightly sharpen one or both files, after uprezzing, to see how good, and how alike, you can make them. Ultimately it doesn't really matter which way you do it, but when you downrez you are inevitably throwing away some quantity of the fine detail which is the point of using such a camera in the first place, so although you would be fairly comparing noise, you would not be fairly comparing detail, texture, sharpness, and 3D effect.
Obviously pushing in PP is unacceptable, and these images don't really exist. The 5D isn't nearly as good at 6400 and 12800, but there's nothing wrong with pushing if that's what you have to do. Here's the 450D at iso 6400: http://cyberphotographer.com/450d/contax28f2.8/grumpypa6400_lowrez.jpg
Pixel Perfect wrote:
No but your jacked up response is pretty lame IMO. But I'm waiting for your examples under the same conditions to show me how much better you can do it. Talk-the talk, now walk-the-walk.
I sincerely wish you were right that the 5D2 will be better than the 5D at high ISOs.... but based on the photos I've taken with a pre-prod 5D2, my expectations are lowered. I'd settle for 8.xx more MPs while maintaining the 5D's high ISO noise performance.
Note that the 5D2 has a "21 megapixel CMOS sensor (very similar to the sensor in the EOS-1Ds Mark III)."
"DIGIC IV image processor, 14-bit ADC
The 5D Mark II becomes the second Canon DSLR (after the EOS 50D) to feature the new DIGIC 4 processor. This enables various new image processing features (such as lens peripheral illumination correction) as well as dealing with 21 megapixels of data at up to 3.9 frames per second (or 82 megapixels a second). It also enables video capture although we're told a second chip does the actual encoding."
I'm not sure if Digic 4 has a role in controlling high ISO noise, but I've accepted that 5D2 will have similar high ISO noise performance as the 1DS3, at considerably less than half the price, with HD video as bonus. If it turns out I'm mistaken, I'll be very happy to be corrected.
I tend to agree that the 5dmk2 is not much better at 100% than the 5d actually might even be a bit worse.
But due to the extra res you can do NR on it much better and it retains more detail than the 5d. I have compared 5dmk2 iso6400 shots (scaled down to 12mpx) with iso3200 5d shots and they were close with the 5dmk2 being a bit sharper. Now if you apply NR the diff gets even bigger. I would put it at 1.5 stops when you compare scaled down 5dmk2 shots to 5dmk1 shots.
Now in terms of DR, and shadow detail I can't say yet which one is better. (hopefully the 5dmk2 as it is 14bit).
Color accuracy is another thing to look at.
I really find most reviews lacking as they don't look at those things some measure DR but most don't look at color accuracy and shadow detail.
dnenciu wrote:
I tend to agree that the 5dmk2 is not much better at 100% than the 5d actually might even be a bit worse.
But due to the extra res you can do NR on it much better and it retains more detail than the 5d. I have compared 5dmk2 iso6400 shots (scaled down to 12mpx) with iso3200 5d shots and they were close with the 5dmk2 being a bit sharper. Now if you apply NR the diff gets even bigger. I would put it at 1.5 stops when you compare scaled down 5dmk2 shots to 5dmk1 shots.
Now in terms of DR, and shadow detail I can't say yet which one is better. (hopefully the 5dmk2 as it is 14bit).
Color accuracy is another thing to look at.
I really find most reviews lacking as they don't look at those things some measure DR but most don't look at color accuracy and shadow detail....Show more →
Well, I'd expect the 5D2 high ISO performance to be not much different from that of the 1Ds3, which is already well tested.
Here's a comparo by Dxo among the 1Ds3, 5D and 20D, with test and data sets that would satisfy even Braniac's advocated method:
Romy. Chuck Westfall stated that only in jpg is the 5D2 significantly better than the 1Ds III, 1.5-2 stops I think, but depending on RAW converters and all things being equal, RAW will only show marginal improvements.
Liquidstone wrote:
Here's a comparo by Dxo among the 1Ds3, 5D and 20D, with test and data sets that would satisfy even Braniac's advocated method: http://www.dxomark.com/
Thanks for that link. All I could find there was a whole load of graphs, which show that the D700 is a better camera than a 5D, and a 5D is a better camera than a 1Ds3. No readings were taken at isos 6400 and 12800 on the Canons. Having used all of these cameras in action, all I can say is that their method, whatever it is, credits cameras in inverse proportion to their picture quality. I advise anyone to test these cameras in real circumstances, and in the ways they will actually be used. While DXO has measured very slightly higher dynamic range for the D700 over the 1Ds3, my impression from shooting both cameras is that the D700 _seems_ to have less DR in real pictures. I have no idea why. Of course, most people worried about DR are landscape shooters, and I would guess that the D700's lower limit of 200 iso might neutralise any slight advantage these graphs claim that it has. Maybe the DXO analysis is yet another per pixel analysis. That would explain why their results are inversely proportional to actual picture quality.
Pixel Perfect wrote:
Romy. Chuck Westfall stated that only in jpg is the 5D2 significantly better than the 1Ds III, 1.5-2 stops I think, but depending on RAW converters and all things being equal, RAW will only show marginal improvements.
Chuck's statement on RAW files from both cameras seems to be a match to what I've seen so far.
The jpegs from the pre-prod 5d2 that I used (default settings) have very strong NR done, taking away a lot of detail in the process.
brainiac wrote:
Maybe the DXO analysis is yet another per pixel analysis. That would explain why their results are inversely proportional to actual picture quality.
Maybe before commenting on DxO's results, you'd like to read the following first:
At the expense of either stating the obvious or looking like a fool, is there ANY value in a test, whichever it is, which reveals issues/performance which one cannot detect from actual shots? After all, who is going to spend his/her time shootng only at brick walls and/or charts and rulers? Who only takes shots at ridiculously high ISOs and therefore chooses a camera based strictly on performance under these extreme conditions?
Conversely, if I am happy with the shots a camera enables me to take, why worry? And if actual shots show no difference between 2 pieces of equipment, so be it. I, for one, am not interested in theoretical advantages, but in getting the equipment part of my shots up to the desired level.
I have played for a few hours with a 40D, a 50D and a 5D. No "extreme" conditions, just "normal" (for me) shots. I printed them out without PP on A4, and looked for the best IQ of the lot. And this is what I intend to do with the 5D2. Either it reveals better IQ under "normal" conditions, and I may buy it, or not. And no amount of ISO 6400 capability, or video, or weatherproofing, or AF is going to change that. Sure high ISO and the rest is nice, but by far not as important as everyday IQ on every shot.
This article lists signal to noise ratios for the latest and greatest Nikon and Canon bodies, and also SNR's adjusted for the resolution differences. With resolution taken into account, there is very little in it between the D3 and 1Ds3, which is consistent with my experience. It's still hard to get an idea of the real differences purely through numbers and graphs though.
This article supports what I have been banging on about for years: don't compare 100% crops from cameras with different numbers of pixels. It's misleading.
philber wrote:
Who only takes shots at ridiculously high ISOs and therefore chooses a camera based strictly on performance under these extreme conditions?
You are one person, Richard, maybe not quite enough to be a majority
So maybe the rest of us can focus on conditions which more closely replicate the every day use of our cameras?
Besides, you may have noted that my advocating choice based on as close as possible to daily use is not exactly the same as focusing on 100% views?
brainiac wrote:
This article supports what I have been banging on about for years: don't compare 100% crops from cameras with different numbers of pixels. It's misleading.
Hmmmm..... have you seen that the graphs are presented in two ways?
1. Screen - pixel to pixel comparison
2. Print - frame to frame comparison (they arbitrarily assumed an 8 MP common size, or 8"x12" @ 300 ppi, and transformed all images to this size)
In the early pages of this thread, I've said I prefer pixel-to-pixel comparisons. This is because I can easily "normalize" the results if I wish to derive a frame-to-frame comparo. DxO presents both stages of data (pixel to pixel and normalized), so that makes it more convenient and normalization is less prone to variability of transformation methods.
I can't really understand why you are "banging on" one method when it's my understanding that both sets of data are giving the same info. The normalized results were actually derived from the pixel-to-pixel data, as well as the sensor size and resolution (the last two points are always known for each camera model).
After owning just about every Canon body released within the last few years, and shooting with each of them professionally and extensively, I can assure you that those DXO graphs, specifically in terms of S/N and dynamic range (the two parameters I'm always most interested in), are very, very accurate to my experience. Specifically, my 1D3 shows a little less noise than my 1Ds3 at typical output sizes (which DXO puts at 8x12 which is perfect for me), the 40D is about a stop less than either of those two bodies, and the 5D is competing very, very well for its age.
But once again I'll stress, for the benefit of those with less experience coming to these threads trying to pick a camera body: In terms of IQ these cameras are more alike than they are different, and as the DXO graphs show, we're really just splitting hairs for the fun of it. However, as alike as all of these new camera bodies are in IQ, they are vastly different in features and should be chosen more based on their features (or lack of) than anything. For example, if you need 21 megapixels, your only choice is Canon. But if you need to shoot at 12000 ISO, until the 5DMk2 comes out your only choice is one of the Nikons. The 1 Seres AF is vastly superior, particularly in low light, than the 5D series bodies. And so on and so on. Choose the camera with the features that meet your particular shooting needs. And fortunately choosing any of the top bodies out today will provide you remarkable image quality provided you have the skills (and the camera features) to get the shot. And don't miss the forest for the trees when making your decision, as too many that come here to research do...!
Brainiac, that's very impressive. Are there any special PP techniques we should emphasize to achieve that kind of result, or is it just the usual combination of RAW tools?
