Wayne; I think you should rent that A900 just to try it out. If you were ever considering Scandinavia for a visit you could borrow my outfit just for the heck of it.
I am not saying you would like it better, not even coming from Canon myself, but I like it and for me the viewfinder is as much a contribution to the decision of getting the A900 as it has bought me a couple of more years focusing alt glass manually, it is simply so much better. The last two years with Canon 5D it was a hit and miss experience.
As for the other aspects I can't say Sony is VASTLY superior in any way it is just that the files look better to me.
Henrik,
I agree with you that DXO and pop-photo testing/benchmarks are far from perfect and need to be better.
I agree that the A900 have more richer saturated color look different than Canon and Nikon.
How different is hard to tell without same scene shots and how close you could tweak the color of the Canon to the A900 would all be interesting to see.
What contributes to these color differences, CFA design vs. image processing pipeline differences would be interesting to see.
Also, how accurate is the color of the A900 relative to the actual scene.
Now some people don't care and don't mind if the colors are enhanced over the real colors. Alot of people prefer more vivid color and saturated colors over less and I can see in magazines and nature shows on Discovery channel and how TV sets are set to VIVID mode more often that this is a trend.
I will see about renting an A900 and 135 and doing some comparison shots.
That still won't answer what is causing the difference.
For that I will see what info Iliah has.
Oh, and the last thing which started all this, how is the A850 better than the A900 one whole stop in high iso noise and how is the color different?
Andrey just said the color was different, thats all. Nothing more.
edwardkaraa wrote:
I think Wayne is on the way to switching to Sony. It always starts with denial, then doubt, then total conversion
You forgot the part about making friends with the whole Sony user community!
I am in trouble now!
Tariq Gibran wrote:
A lot of monitors, yes, but EIZO is not the only one that can equal or even surpass what modern state of the art printers are capable of. There is at least one NEC that offers an amazing 107% of Adobe RGB. Originally selling for a whopping $6500!, it can now be had for $999, complete with color calibration software and hardware. It is the LCD 2180WG, the first LED lit monitor. Nothing can likely touch it for still photography but it's no good for other uses with lots of motion - games for instance - due to it's slow refresh rate.
Many of the WUXI series NEC panels are exceptional and will get you to 96% of Adobe RGB (the top of the line EIZO's claim 97%) much cheaper than an EIZO monitor. Although one might be able to measure a slightly wider gamut than this form current printers, the difference is too small to be readily noticeable. ...Show more →
That's similar to the Dell I currently use 2408WFP -- "110% color gamut", 96% AdobeRGB, 100%sRGB. Recently rep[laced with the U2410, with similar specs. I suspect they use the same panel as the NEC's. Compared to the EIZO's and even the NEC's, the Dell is a good value.
kosmoskatten wrote:
Wayne; I think you should rent that A900 just to try it out. If you were ever considering Scandinavia for a visit you could borrow my outfit just for the heck of it.
I am not saying you would like it better, not even coming from Canon myself, but I like it and for me the viewfinder is as much a contribution to the decision of getting the A900 as it has bought me a couple of more years focusing alt glass manually, it is simply so much better. The last two years with Canon 5D it was a hit and miss experience.
As for the other aspects I can't say Sony is VASTLY superior in any way it is just that the files look better to me. ...Show more →
Well, as mentioned above I just bought an a850 to see. Still don't have a lens for it though :-(
I can tell you at this point that the a850 viewfinder is brighter and a bit bigger than the 5D (classic, Mkl version). With lenses off both cameras, the Sony is noticeably brighter at the viewfinder.
I have a chipped M42 adapter coming so I can compare directly with a 5D and probably a 5Dll with the same lens (removing lens color issues from the comparison). I don't doubt that there will be differences, or even that the Sony might look better, whether that be in a Velvia sort of way, or any other way. The question I have, which has not been conclusively resolved by this discussion of any other I've read, is whether is it is objectively better -- that is, measurably more accurate and/or measurably providing a greater range of tones (which is, of course, related to accuracy as we don't necessarily want the camera to create colors not in the scene), and whether the differences in the output can be approximated by calibration in RAW processing with say Canon or Nikon files -- provided, of course, that you subjectively prefer the color output of the Sony. Of course, in the process I can decide if I like the Sony color sufficiently to sacrifice the significantly better high ISO of the Canon (or the Nikon).
Lotusm50, just curious as to which M42 lens you will be using on both cameras? Do you currently have both the 5D and 5DII?
All this talk about the Sony's "Velveeta" like color is a bit out there in my experience with both. You would only see that if you had some fairly dramatic creative styles set (such as 'Deep' instead of say 'Neutral') with out-of-camera jpegs or using the Sony converter to get those in-camera settings from raw files.
For me, I don't think one can talk about color without also linking it to dynamic range and the overall tonal (film curve) response of the sensor. The effect both have on color is what I see as what sets the a900 apart (which is similar, though not as dramatic, as what one sees with MFD backs). Getting those subtle, gradations of color the closer up the tonal scale one goes, approaching pure white or clipping, is visible and measurable. It's the difference between that tinge of color in highlights, or lack of it, that often screams digital or film to many of us. In film curve parlance, the curve has more of a shoulder and rolls off more gently at the top. This is why with the a900, one can expose to the right. With my old Canon 5D, to mimic such a roll off, I had to underexpose and later perform more post production processing to come close to this look. I don't think there is any voodoo or magic here. Anyone can see this through testing themselves.
Whether the a850 reacts in exactly the same way, I do not know. It appears that it is being suggested in this thread that the a900 and a850 are not as similar as I had thought.
douglasf13 wrote:
you2, I didn't really mean to imply that you or anyone here were the typical user. I think this forum is generally helpful and nice. As far as the A850, I'm just not really sure as to how noticeable the color differences are between it and the A900, but, I certainly take stock in what Andrey from RPP said in that they are different.
I would suggest that you go over to getdpi and ask around. I know there are at least a couple of DMR to A900 converts.
Not to be contentious here Douglas... but I think you may be reading your personal preferences into reasons who a couple guys over on the getdpi forum switched from the DMR to the Sony.
I was in communication with several former DMR shooters who moved to the Sony and both indicated they needed auto focus and better high iso (less noise) than the DMR. Here's a quote from one of the new A900 owners:
"With the DMR, it stayed at home because of the size and weight. It also makes me think I may have a M9 in my future.
As for the Sony, yes, I'm quite happy with it. As I have said repeatedly, the Leica mops the floor with the Sony on image quality, but the Sony is good enough for most applications. I shoot it like my old F3 - spot meter for the most important detail, place it about where I think it should be on the tonal scale, focus manually and shoot."
Might be useful to "bone up" on your information before you make assumptions like that.
Lawrence Beck wrote:
As for the Sony, yes, I'm quite happy with it. As I have said repeatedly, the Leica mops the floor with the Sony on image quality, but the Sony is good enough for most applications.
It would be great to see and have access to two identical shots taken with both the Sony and the Leica using two known good lenses on both to see this. Since it sounds like you have the Leica and the Sony, perhaps you could do this with a landscape shot and provide the raw images. I have not seen a valid comparison done by someone with real experience shooting both systems yet - one that got the most out of both.
I agree completely Tariq.
I was very interested in the A900 given the death of the R system. The SSS could be invaluable for my hummingbird photos taken with the 800 modular lens as this beast is difficult to keep vibration-free with the best of tripods. The larger file size the A900 provides would be most welcome for making oversized landscape prints as well. I'm not, however, inclined to bail on the DMR for something that is not as good at rendering fine detail. This was the prognosis I received from several former DMR owners. Both agreed that the DMR provided better IQ, albeit with smaller files.
The A900 is superior in so many areas that I may still go that route just for it's landscape possibilities. Douglas is probably the most knowledgeable A900 shooter I'm aware of and I have the greatest respect for his opinions. I just think the implication from his comment was a bit misleading, given the feedback I've received from DMR owners who really took the time to get the maximum IQ from the DMR. I'm personally hoping Sony will release a successor to the A900 with 30+ megapixels, SSS and Live View. If and when that happens I'll be one of the first in line.
I have a friend in Hawaii who shoots with the A900. I fly over to Oahu several times each year to photograph waves. My DMR is presently at Leica getting adjusted... but once it's back I plan to fly over and shoot some images side by side with the A900 and DMR. I'd even buy a Leitax mount for the 100 APO macro for this test just to settle the issue for myself. Then again... given the crop factor of the DMR, a better matchup might be comparing the 100 APO on the Leica to the Zeiss 135 on the Sony with a landscape as the subject.
Lawrence, I made a big mistake. you2 was talking about contax lenses, and, for some bizarre reason, I forgot that and thought he was talking about Leica R lenses (i need to stop posting with my cellphone .) Duh. I was simply pointing out that there were some dual A900 and DMR users on getdpi, so that may be a good resource for him to ask around about the differences in camera bodies (for his imaginary R lenses.) I know full well that you prefer the DMR color, and I wasn't meaning to imply that the A900 was better. In fact, I've personally considered trading in all my A900 gear for an M9 quite a few times, and haven't done it mostly because of the cost. Sorry to both you2 and Lawrence for causing such a misunderstanding. I just got a 50 Summicron, so I must have had Leica on the brain.
p.s. on a side note, I bummed about the 50 Summicron. My uncle gave me a Leicaflex SL and the Summicron to "borrow" indefinitely, but I didn't realize that the first edition 50 Summicron needed a little filing to work on the A900, and I can't do that to his lens, so it looks like I need to buy some film. The Leicaflex is a real beauty, so I guess I can't complain.
Lotusm50 wrote:
Well, as mentioned above I just bought an a850 to see. Still don't have a lens for it though :-(
I can tell you at this point that the a850 viewfinder is brighter and a bit bigger than the 5D (classic, Mkl version). With lenses off both cameras, the Sony is noticeably brighter at the viewfinder.
I have a chipped M42 adapter coming so I can compare directly with a 5D and probably a 5Dll with the same lens (removing lens color issues from the comparison). I don't doubt that there will be differences, or even that the Sony might look better, whether that be in a Velvia sort of way, or any other way. The question I have, which has not been conclusively resolved by this discussion of any other I've read, is whether is it is objectively better -- that is, measurably more accurate and/or measurably providing a greater range of tones (which is, of course, related to accuracy as we don't necessarily want the camera to create colors not in the scene), and whether the differences in the output can be approximated by calibration in RAW processing with say Canon or Nikon files -- provided, of course, that you subjectively prefer the color output of the Sony. Of course, in the process I can decide if I like the Sony color sufficiently to sacrifice the significantly better high ISO of the Canon (or the Nikon).
+1
Cool look forward to your comparison shots and analysis.
One thing I like about this forum is that the camera base is diverse and I find that more preferable to the isolationistic,more fanatical camera brand forums.
Henrik,
thanks for the offer and I do plan on making a trip in the future to your beautiful spot in the world, maybe along with Iceland.
Douglas: The "s" has a slightly different CFA, unfortunately I can't publish the response curves I've measured since they're the intellectual property of my current employer. Now don't mistake all "wide" filters for being "worse". In fact, metamerism areas can be smaller with a wider filter, if the filter is made with carefully considered ramifications. Canon's biggest drawback is the swingback in the red channel, a kind of dual-peak response which makes the metamerism area quite large/wide just smack in the middle of skintones. This also affects the transition from blue to purple by inserting an uncertainty factor that's quite large. Very small amounts of noise can throw the colour interpolation way of mark - hence the strange colour noise behaviour of most recent Canons. Noise reduction on top of this gives luma contrast a very good look at very high ISOs, but also totally kills any colour information in the image - they go "flat".
And regardless of this, the sad part is that most users really don't get what colour is all about. Colour quality in a camera/sensor is NOT about getting accurate colour straight out of raw-files - this is quite simply not physically possible. Accurate colour comes from the camera profile. The camera/sensor part in the system is to provide ACCURATE and NON-METAMERIC placement of hues on a hue-circle - they don't need to be colorimetrically perfect as long as they do this. It is not in any way part of the CFA responsibility.
One of the major reasons that the A850 is better at higer ISO's than the A900 is that they reinforced the botched and underdimensioned powersupply lines going into the sensor. The support structure around the chip is totally different from the D3x implementation.
"Narrow" or "sharp" filters have another drawback, and that is that noise increases. If you don't have large, uniform patches of colour, noise is just as important to colour fidelity as the CFA. So you can only get "better" up to a certain point.
And BTW:
NO MDFB IS 16-BIT ACCURATE READOUT!
Don't buy that BS. The files may be 16 bits, but that's just interpolation from the 12-bit AD converters. In fact, no MDFB amplifier/AD chain comes even close to the signal quality of D3/D3s/D3x or the Canon counterparts. In fact, pixel for pixel even the P65+ is worse than my lowly "toycam" - the Nikon D40. The only thing that the P65+ has got in its favour here is a base ISO of just over 40, an almost 6x higher sensor area, and a massive amount of pixels. Measured for pixel quality and colour accuracy it's quite a lot worse than even the worst entry-level DSLR today. It just beats this by piling on area and pixels. The best MDFB in existence isn't in the least hindered or kept down by a 12-bit converison readout.
So in your experience which consumer cameras produce reasonable colours ? On the surface it sounds like you are suggesting that canon is not the place to look; what about the alternatives ?
theSuede wrote:
Douglas: The "s" has a slightly different CFA, unfortunately I can't publish the response curves I've measured since they're the intellectual property of my current employer. Now don't mistake all "wide" filters for being "worse". In fact, metamerism areas can be smaller with a wider filter, if the filter is made with carefully considered ramifications. Canon's biggest drawback is the swingback in the red channel, a kind of dual-peak response which makes the metamerism area quite large/wide just smack in the middle of skintones. This also affects the transition from blue to purple by inserting an uncertainty factor that's quite large. Very small amounts of noise can throw the colour interpolation way of mark - hence the strange colour noise behaviour of most recent Canons. Noise reduction on top of this gives luma contrast a very good look at very high ISOs, but also totally kills any colour information in the image - they go "flat".
And regardless of this, the sad part is that most users really don't get what colour is all about. Colour quality in a camera/sensor is NOT about getting accurate colour straight out of raw-files - this is quite simply not physically possible. Accurate colour comes from the camera profile. The camera/sensor part in the system is to provide ACCURATE and NON-METAMERIC placement of hues on a hue-circle - they don't need to be colorimetrically perfect as long as they do this. It is not in any way part of the CFA responsibility.
One of the major reasons that the A850 is better at higer ISO's than the A900 is that they reinforced the botched and underdimensioned powersupply lines going into the sensor. The support structure around the chip is totally different from the D3x implementation.
"Narrow" or "sharp" filters have another drawback, and that is that noise increases. If you don't have large, uniform patches of colour, noise is just as important to colour fidelity as the CFA. So you can only get "better" up to a certain point.
And BTW:
NO MDFB IS 16-BIT ACCURATE READOUT!
Don't buy that BS. The files may be 16 bits, but that's just interpolation from the 12-bit AD converters. In fact, no MDFB amplifier/AD chain comes even close to the signal quality of D3/D3s/D3x or the Canon counterparts. In fact, pixel for pixel even the P65+ is worse than my lowly "toycam" - the Nikon D40. The only thing that the P65+ has got in its favour here is a base ISO of just over 40, an almost 6x higher sensor area, and a massive amount of pixels. Measured for pixel quality and colour accuracy it's quite a lot worse than even the worst entry-level DSLR today. It just beats this by piling on area and pixels. The best MDFB in existence isn't in the least hindered or kept down by a 12-bit converison readout....Show more →
theSuede, thanks for such a clear an informative post. Feel free to expound more if you have time, as I'm sure that a lot of us here are pretty interested in gaining more understanding on the subject.
p.s. are you saying that the CFA of the A850 and A900 is the same, and Sony just changed the support electronics? If so, that's very interesting. Also, there's been two main varying opinions about the differences in the D3x and A900. One is that the D3x has a similar sensor to the A900, but it has 14bit onboard ADCs. The other is that the sensors (outside of toppings) are the same, and Nikon does its 14-bit magic off chip derived from the 12-bits. Are you saying the latter is correct? thx.
Well, the digital communication lines out from the sensor in the D3x are only twelve lanes wide, just as the 850/900... And I seriously doubt they "hid" the missing two lanes somewhere else. There's a definite correlation between the slower read-out times 12>14bits and this. This is not 100% certain proof, but a very good indication.
The CFA's in the A850 and the A900 seem to be slightly different, but I've no hard data there yet. I have to take the time to go through the 850. This will be soon, as I'm personally interested in that camera. In fact the difference 900>850 seems to be about the same as the difference D3>D3s, with the D3s being (very) slightly better on all accounts.
you2:
Most cameras can be "good enough" with good light, and the right profile. Canon and Olympus have placed their level of acceptance slightly higher than some other players, both for the same reason - both the 4/3-system and the later Canons have a quite high MP/cm2 ratio. But for "normal" circumstances the problem is not as bad as it would seem, you only run into real-world, real-picture noticeable problems when shooting in "spiky" spectrum lights like fluorescents or sodium or stadium-type lights. Otherwise, no-one except the discriminating photographer who actually saw the scene in reality will notice any larger differences if you use the right camera profile for the situation (in my personal opinion). For higher ISO shots, the problems lies mainly with the blue primary, as that is highly dependent on the available surface fill factor.
You have to protect the silicon surface (as the pure silicon crystal is quite susceptible to chemical contamination), and the easiest way to do this is to oxidize the surface into SiO2 a few A deep, this is called the "passivation". You also have to insulate the areas around the lanes transporting charge in/out to the individual cells to keep them from contaminating the measurement. Both the insulation and the passivation are highly absorbant in the blue spectrum, so there is a considerable loss of blue Qe if you make the cell areas smaller. This is also why the noise on higher ISOs is mostly red/green - blues have long since vanished below the detection threshold, only thermal/read noise remains, almost no actual photon detection occurs.
Green has less problems with this, and long red passes through almost everything except the metal parts of the chip. Canon has expanded the blue response a bit to counteract this, but unfortunately this affects the blue-purple-red transition in a bad way. Trying to get the blue-to-purple shift just right in a Canon takes some serious work.
So, in good light with the right profile, any camera will do. A slightly "worse" camera will need a slightly "sharper" profile to be used for accurate correction, but it's definitely doable. The differences only surface when you get into "difficult" light, or less light. In my opinion, this is also why different raw-conversion engines work better with certain cameras, and worse with others - the algorithms that interpolate the two missing colours per pixel can use many different schemes of weighing luma/chroma (mis)information against eachother to get as close to reality as possible. And some schemes work better for some types of CFA compromises, worse for others.
