ulrikft wrote:
Those that think that the d3/d700 is better than the 1ds mark III (when resized), check out brainiacs analysis on the topic, that goes for the 1ds mark III being worse than the 5d mkII too :P
Brainiac's analysis is fine. However it says nothing about relative performance of the 1DsMkIII and the new 5DMkII. Both of these cameras have exactly the same pixel count so we can indeed compare them directly. As noted in many sources, the 5DMkII sensor is an update/improvement (particularly with respect to noise) of the 1DsMkIII sensor.
douglasf13 wrote:
That is a JPEG! It has been gone over many times that the jpeg engine in this camera is not very good, and I firmly believe that. Don't post JPEGs for this discussion, and if you're a jpeg only shooter, the A900 isn't for you. You're swift boating, here.
So there is yet another potential use the the a900 is not useful for. The market for this camera keeps shrinking! What's left, just studio work at 100 ISO? I'm not "swift boating" here at all. Just reporting what I see. So far, I've seen no evidence to date that performance with RAW with this camera is going to be significantly different with respect to noise. The RAW converter is not magically going to eliminate 2 or more stops worth of noise without applying a lot of noise reduction -- and that is going to remove detail. The noise is in the RAW file -- the RAW converter (like the jpeg out of the camera) can just apply software to reduce it but at the expense of detail and sharpness. But I'll be waiting to see the magic.
In the meantime here is comparative crops from 800 ISO samples from the a900 and the 1DsMkIII, from the recent a900 review at imaging-resource.com. These are the crops with noise reduction turned off. At 800 ISO, the a900 is clearly noisier, and shows less detail/less sharpness than the 1DsMkIII. Since the 5DMkII improves upon the noise performance of the 1DsMkIII, the difference between the a900 and the 5DMkII will be even greater. With NR turned on, the detail in the samples on the site is further obliterated in the a900. http://boncratious.com/images/a900-1DsMkIII-800.jpg
I was under the impression that all the IR files had some NR?
Have they updated?
(and the a900 has better noise performance than my 30d, a better build, most likely better af, better resolution, better dynamic range, FAR better viewfinder and zeiss af-lenses... I wonder what my 30d is good for if the zeiss is _only_ good for iso100 studio work.. a door stopper? see? going to the extremes is a bad argumentative style, try being more reflected :P )
Lotusm50. First of all, those aren't NR turned OFF, because there is no such thing for ISO 800. NR kicks in at ISO 1600. Secondly, you're showing us JPEGS, which isn't what we've been talking about. If you shoot jpeg, then YES, the 1Ds III is clearly better. Just like in the V1 of the A700 firmware, the A900's jpegs smudge up and show very little detail. JPEGs are the true weakness of this camera. The 1ds III isn't two stops less noisy, it just has finer grain, which the A900 has with RAW conversion as well. BTW, this isn't taking into account the fact that the A900 shots were underexposed, according to Iliah Borg. Give it a rest. To show the difference in detail between jpeg and RAW for the A900, here is a RAW conversion (early beta RAW program) of the A900 vs.1Ds III at ISO 100. Compare this A900 RAW at ISO to the jpegs at IR, which is right here. A900 ISO 100 JPEG
http://www.grimadde.com/forums/1Ds3_vs_A900.jpg
FWIW, Here is what the full review at Chausser d'Images said.
- image quality on par and sometimes exceeding 1dsIII;
- noise on par with 1ds III;
- very good noise handling until iso 1600, d3/700 better at higher iso's;
- best optical viewfinder of all dslr's.
I will add that if the 5D ii blows away the 1DsIII in noise and detail, then it will be a better camera than the A900, IQ wise.
ulrikft, all of the RAWs at IR do have NR on, but that kicks in at ISO 1600.
douglasf13 wrote:
Lotusm50. First of all, those aren't NR turned OFF, because there is no such thing for ISO 800. NR kicks in at ISO 1600.
Yes. . That's right. NR kicked-in and started obliterating the detail at 1600 (that's the NR sample I was looking at. when I made that particular comment). But the point still stands, these are samples with NR turned off for both.
I have no doubt that at 100 ISO the a900 is comparable to the 1DsMkII. I'm not disputing that. But as all those samples on imaging -resource.com show, the a900 falters as you move above 100 ISO and the 1DsMkIII (and by extension the 5DMkII) surpasses it in increasing margins.
This may or may not be consistent with your quoting of "Chausser d'Images" review. When they say noise "on par with", noise at what ISO are they talking about? As I-R.com shows, it might only be at 100 ISO. I'm looking to see more comparisons done, The I-R.com is not all that favorable to the a900 -- it is only comparable in IQ performance to the 1DsMkIII at 100 ISO ( which given the a900's price is excellent). Looking for a competent comparison with the new 5D MKII when it arrives.
Let me post a quick point of comparison to the down-sampled 800 ISO crop of the a900 image I posted above, It was commented that the image was "underexposed". Well, it was actually of one of the brighter areas. However, here is an underexposed, section, in partial shadow, of a sample image from the new 5DMkII. This is also from a dpreview.com preview, is also at 800 ISO, and has also been down-sampled to the original 5D file size (2912x4368). Looking at "underexposed" areas for both cameras, on a fairly common basis (not identical, the same subject matter was not available, and therefore not ideal but still instructive), the 5DMkII is hands down better. There is a huge gap in relative performance here that some magical RAW conversion process will have a big challenge closing. http://boncratious.com/images/5DMkII-800ISO-12mp.jpg
You have got to be kidding me. Read my post again. I wasn't comparing the A900 to the 1dsiii at ISO 100, it just so happened that the 1dsiii image was there, too. My point was for you to compare the detail difference between the A900 jpeg and RAW. If you actually think we can compare different images from different cameras, you're mistaken. Bottom line, A900 jpegs don't hold up, so stop comparing them. I'm done with this thread. You'll feel silly soon enough, once real comparisons are made.
Like I said, I'm waiting for competent, fair, comparisons to be made. But you've missed the point I've been making. Based on one the evidence to date, the RAW processing you are referring to has a lot of ground to make up. It is not clear to me that it's raw processing is going to be that magical (and relatively that much more powerful than say Canon's or Nikon's) as to make up for a sensor that appears to be inherently significantly noisier. I-R.com's samples show it, and dpreview.com's samples show it. I've yet to find samples, imperfect or otherwise, that don't show it.
I hope that I am wrong. But everything I've seen and read to date (other than your so far unproven and unsupported raw processing claim), is not promising. If it is in fact comparable to the new 5DMkII in noise, I will buy one. But as it stands I am not willing to accept the loss in IQ at 200 ISO and above for a few body features that are nice but ultimately of small benefit to me. My check book is ready. I haven't pre-ordered the 5DMkII yet precisely becuase I want to give the a900 every chance to prove to me that it can match 5DMkII for IQ and noise.
douglasf13 wrote:
You have got to be kidding me. Read my post again. I wasn't comparing the A900 to the 1dsiii at ISO 100, it just so happened that the 1dsiii image was there, too. My point was for you to compare the detail difference between the A900 jpeg and RAW. If you actually think we can compare different images from different cameras, you're mistaken. Bottom line, A900 jpegs don't hold up, so stop comparing them. I'm done with this thread. You'll feel silly soon enough, once real comparisons are made.
All that is left is studio work? I guess you have been hiding under a rock!
A900 has almost the ISO performance of D3, Even sometimes little bette detailed and not so mushed noise from NR. All this when downsampled to D3s size.
Please see Davids comparisions, he have used the D3 alot and are using the A900 right now!
Lotusm50 wrote:
Brainiac's analysis is fine. However it says nothing about relative performance of the 1DsMkIII and the new 5DMkII. Both of these cameras have exactly the same pixel count so we can indeed compare them directly. As noted in many sources, the 5DMkII sensor is an update/improvement (particularly with respect to noise) of the 1DsMkIII sensor.
So there is yet another potential use the the a900 is not useful for. The market for this camera keeps shrinking! What's left, just studio work at 100 ISO? I'm not "swift boating" here at all. Just reporting what I see. So far, I've seen no evidence to date that performance with RAW with this camera is going to be significantly different with respect to noise. The RAW converter is not magically going to eliminate 2 or more stops worth of noise without applying a lot of noise reduction -- and that is going to remove detail. The noise is in the RAW file -- the RAW converter (like the jpeg out of the camera) can just apply software to reduce it but at the expense of detail and sharpness. But I'll be waiting to see the magic.
In the meantime here is comparative crops from 800 ISO samples from the a900 and the 1DsMkIII, from the recent a900 review at imaging-resource.com. These are the crops with noise reduction turned off. At 800 ISO, the a900 is clearly noisier, and shows less detail/less sharpness than the 1DsMkIII. Since the 5DMkII improves upon the noise performance of the 1DsMkIII, the difference between the a900 and the 5DMkII will be even greater. With NR turned on, the detail in the samples on the site is further obliterated in the a900. http://boncratious.com/images/a900-1DsMkIII-800.jpg ...Show more →
moire wrote:
A900 has almost the ISO performance of D3, Even sometimes little bette detailed and not so mushed noise from NR. All this when downsampled to D3s size.
Please see Davids comparisions, he have used the D3 alot and are using the A900 right now!
Look at it from my perspective. "almost the ISO performance of D3" doesn't do me any good. My current 5D already has comparable ISO performance of the D3 at ISO settings that I use If I am going to upgrade to a 21+ mp body, it better be able to do better than my current 5D at 12.7mp. I want to see all 21+ mp working at all ISO's, I want something better than what I am using (the 5D) at ISO's I normally use. Otherwise why bother?
I am happy to look a David's comparisons. But who is David and where do I find his comparisons? Do you have a link? With what cameras does he compare the a900? To be honest, a comparison that shows it is comparable at 12mp to a D3 really doesn't do all that much for my decision making. But, at the very least it will be good to see his a900 samples (hopefully he used the magic raw conversion that is essentially claimed to eliminate 2-3 stops of noise, without loss of detail).
But you will have to wait a week or two to get a review of the a900, this is just his findings after using the a900.
You will have to wait a really long time before we will see double the efficiency of each pixels. That is what you require.
Todays sensors are using already very officiant microlenses and readout electronics.
Only a new more radical design approach can double the quantum efficiany as you require and I doubt that we will see that this decade .
Better software NR is another question tho. Better algorithms that can distinguish detail from noise will slowly improve
moire, I love the "you require" part in your post...
We humans can never have enough.
Sony made A900 at 2.5 times cheaper than 1Ds3 and there are complaints still. (is it 2.5 times worse in IQ?)
Then Canon improved their flagship sensor in the 1Ds3 and stuff it into the prosumer body 5D2 then price at A900 level to counter head-on. I bet there will still be complaints . (in fact there are already: lack of 50 AF points like in D3 and lack of weather sealing - which is false - I wonder how often they shoot in the rain per year, or lack of gapless design etc.)
But you will have to wait a week or two to get a review of the a900, this is just his findings after using the a900.
Well, that's not a real comparison, and it is all down-scaled to 11.2 mp. Not exceptionally helpful to my needs. He has one sample done at 3200 ISO plus one stop exposure (effective 1600 ISO) and it is pretty noisy at 11.2 mp -- I would say easily worse than my current 5D (at 12.7 mp)
I will wait to see a real comparison. Finally, I try not to rely on reviews posted on fan sites of the camera in question. Their somewhat less than unbiased enthusiasm usually comes through in their reviews. Sometimes the sample images produced are helpful but often their assessments of those images and conclusions drawn are less than objective or reliable. This one might be different. I'm not sure. But generally I prefer to look at, or rely on, more objective sources, and sources that are not explicitly biased to a particular product line.
moire wrote:
You will have to wait a really long time before we will see double the efficiency of each pixels. That is what you require.
Todays sensors are using already very officiant microlenses and readout electronics.
Only a new more radical design approach can double the quantum efficiany as you require and I doubt that we will see that this decade .
I don't "require" double the efficiency, but there is no reason to assume that all sensors, of a given pixel count, will have the same efficiency, or than a manufacturer can not increase efficiency or increase the signal to noise ratio of their products over time. I am just looking for "better" than what I currently have. It is reasonable to assume that different products will have different performance characteristics and that include noise and efficiency. Three years since the original 5D was introduced and over a year since the 1DsMkIII was introduced, given the fairly steep technology curve we are on, it is not unreasonable to expect that Canon has improved performance further with it's now current generation of its sensor in the 5DMkII. It appears, from samples available to date, to noticeably outperform the original 5D (and the 1DsMkIII) sensor in terms of noise when output to the lower 12.7 mp file size. I don't expect the Sony to be the same just becuase it was introduced at the same time as the new Canon. It might be better, it might be worse, or even the same. What is available to date is less than ideal, but it doesn't appear to hold out great promise for the relative noise performance of the a900. I await more ideal, competent and objective comparisons so I can be convinced conclusively.
And with respect to Leon's comment, none of this is complaint. I agree that the a900 is a fine camera a great price. I am just waiting to see if it is competitive enough to get me to switch, and noise is an important performance characteristic for me.
Hard to understand why this thread limps on: anyone with eyes can see that a fairly conducted comparison between the A900 and the 5D II reveals a significant advantage for the Canon at high ISO. Probably two stops. QED.
By 'fairly conducted comparison', of course I mean comparing RAW files with NR turned off processed through identical workflows and sampled up or down to the same size. Heaven knows why some are still confusing image noise with per-pixel noise.
Think back to the first Sony digital cameras: always with the megapixels. Noise was never a priority.
The A900's extra 3MP may or may not make a significant difference in resolving ability: of course, much more is involved than the pixel count, and Canon's sensor hard/software design has always been class leading: you might also recall comparisons between the Kodak SLR and the 1Ds Mark 1 that rarely showed the extra pixels mattering by the end of the signal path.
Unless the 5D II reveals some thus far unsuspected Achilles Heel when in production, it seems likely to restore their dominance of the high-end DSLR market for another 18 months.
By the end of the year, we'll see a real challenge from a Sony system fielding killer Zeiss 16-35, 24-70 and 135/1.8 lenses (with effective IS) plus some very attractive Minolta G glass, but I wouldn't bet against Canon L + ZE and adapted lenses looking equally appealing by then, too.
Hubsand, please point me to this RAW comparison, because I haven't seen it. Chausser d'Images is saying that the A900 noise is on par with the 1DsIII, so if the 5Dii is two stops better in noise than the 1Dsiii, then that is truly incredible.
Can I just say that I plan on buying the a900 based on what I've seen on other sites. The RAW conversions I've seen are beautiful, the details are fantastic, the viewfinder is amazing, the camera body is what I expected, and the price is pretty damned good for what you get. All this bickering about what little noise there is seems so petty and lame.
No camera is perfect and no camera will ever be perfect. Some people will never be happy. Buy what you want and be done with it.
On the left, the Sony A900 image is reduced to 12 megapixels using Bicubic Sharper - this emphasises, rather than reduces, the grainy appearance of noise and matches the file size precisely to the Nikon D3 on the right. The actual light levels involved here (1/800 at f6.3 for the Sony, 1/1600 at f4 for the Nikon) are within 1/3rd of a stop of being the same, and both images were overexposed by one effective stop in camera, then set to -1 exposure using Adobe Camera Raw. In both cases, no sharpening, no luminance NR, and 50 chroma NR have been used. The black point of the Sony shot is at 12 while the Nikon is at 2 - this may tend to emphasise the Sony noise in the darker areas by increasing its contrast (but was a better setting for the overall look of the image).
While the Nikon file shows a smoothed luminance structure, the Sony applies almost no in-camera NR to the raw file, resulting in a very crisp and clear grainy structure. However, the Nikon does not appear to lose fine detail, indication that an adaptive NR process is happening.
Taking this comparison and others, I conclude that while the A900 is not identical in 6400 ISO performance to the D3, in practical applications it will prove as robust and deliver similar levels of fine detail and noise. Each file would normally be custom processed - applying some luminance NR to the Sony file makes it closer to the Nikon 'look'.
moire wrote:
While the Nikon file shows a smoothed luminance structure, the Sony applies almost no in-camera NR to the raw file, resulting in a very crisp and clear grainy structure. However, the Nikon does not appear to lose fine detail, indication that an adaptive NR process is happening. {emphasis added}
I guess the operative word here is "almost". The review of the a900 on imaging-resource.com does indicate that there is some noise reduction going on in-camera (apparently to raw and jpegs). This is they say occurring on the chip. They call this "good" NR. The "High ISO" NR that you are talking about can be turned off, which is appreciated. Downloading a few RAW files I found on the web, my trials suggest that noise reduction can be done a little better than is apparent in some jpegs, using ACR (latest version 4.6) or through Neat Image.
Uh? there is NR going on in all cameras. Much more so in D3 then A900 for example.
Do you really think that the raw image that you open is unprocessed?` Guess why Nikons d300 has somewhat less and different structured noise then A700 even tho they use the same sensor
moire wrote:
Uh? there is NR going on in all cameras. Much more so in D3 then A900 for example.
Do you really think that the raw image that you open is unprocessed?` Guess why Nikons d300 has somewhat less and different structured noise then A700 even tho they use the same sensor
I rather prefer to have absolutely no NR applied. It's very easy to remove chroma noise with the appropriate software. Luminance noise is more difficult to remove without making the picture softer, but this is something that never annoys me. I never remove luminance noise, I actually like the look compared to plasticky images. The problem with manufacturers is that they compete against eachother about noise performance and we don't know exactly what's happening between the analog capture and the raw file. Sony's approach is commendable, even though it got flamed for being honest. Maybe they won't repeat this mistake next time
moire wrote:
Uh? there is NR going on in all cameras. Much more so in D3 then A900 for example.
Do you really think that the raw image that you open is unprocessed?` Guess why Nikons d300 has somewhat less and different structured noise then A700 even tho they use the same sensor
Actually, no I don't. It only makes sense to filter noise as it comes off the photosites. I only made the statement in the interest of having complete information in the discussion.
I am not so keen on the Nikons precisely because so much of the noise reduction is done in-camera where there is no control over it. That said, with the D3 and D700, Nikon must have found the "special sauce" becuase they do am excellent job with their in camera NR.
