M Vers wrote:
Does the 1DsIII sensor have the same microlens spacing/array the 5DII has?
I haven't seen anything to indicate anything either way. Chuck Westfall states that the 5D II sensor not the same sensor used in the 1Ds III, and that it is superior to the 1Ds III. He is quite clear in saying the 5D II provides the best IQ of any DSLR currently in the lineup.
He didn't get far enough into the technical specs in the interview I heard to explain what the technological differences were between the sensors in the 1Ds III and 5D II, but it is telling that the highest native ISO supported by the 1Ds III is ISO 1600, while the highest native ISO on the 5D II is two full stops more sensitive at ISO 6400. Those samples I've seen thus far seem to bear this out (that a difference exists). It remains to be seen whether a clear difference exists in a side-by-side comparison, though.
...it is telling that the highest native ISO supported by the 1Ds III is ISO 1600, while the highest native ISO on the 5D II is two full stops more sensitive at ISO 6400. Those samples I've seen thus far seem to bear this out (that a difference exists). It remains to be seen whether a clear difference exists in a side-by-side comparison, though.
In the second case the crops are from in-camera jpegs, whereas in the first both files were optimally post-processed with noise reduction (NX and NeatImage respectively). The 1Ds3 acquits itself quite well against the D700 at iso 12800, whereas the 5D2 seems to suffer from much worse colour noise problems than the D700 at iso 6400. Judging by Canon's statements and these two comparisons, the 1Ds3 and 5D2 are probably quite close.
braniac - thanks for your response to my question. I pre-ordered a 5DII a few days ago for the high ISO and crop-ability of 21MP's. The initial jpeg samples I've seen look really great. I may not need 21MP all the time (sRaw sounds very appealing) which is why I hope to use the 5DII to compliment my 5D. But I was really interested to see if there was a discernable difference in prints between the old and the new 5D.
brainiac wrote:
........... If you generally take pictures for display on the web, then there isn't a great need to compare by printing. ....................
That is not the world in which I live, hence my comments about print comparisons using the same RIP. This greatly helps me understand your perspective.
Thank you for the clarification.
Respectfully,
Marc
moondigger wrote:
I haven't seen anything to indicate anything either way. Chuck Westfall states that the 5D II sensor not the same sensor used in the 1Ds III, and that it is superior to the 1Ds III. He is quite clear in saying the 5D II provides the best IQ of any DSLR currently in the lineup.
He didn't get far enough into the technical specs in the interview I heard to explain what the technological differences were between the sensors in the 1Ds III and 5D II, but it is telling that the highest native ISO supported by the 1Ds III is ISO 1600, while the highest native ISO on the 5D II is two full stops more sensitive at ISO 6400. Those samples I've seen thus far seem to bear this out (that a difference exists). It remains to be seen whether a clear difference exists in a side-by-side comparison, though....Show more →
Do they both have the same pixel pitch (6.4 microns)? I'm thinking the reported higher IQ is derived from the Digic IV--then again I could be wrong seeing as no one is sure about the microlens array. It could just be some really clever in camera NR....
Moondigger wrote: He didn't get far enough into the technical specs in the interview I heard to explain what the technological differences were between the sensors in the 1Ds III and 5D II, but it is telling that the highest native ISO supported by the 1Ds III is ISO 1600, while the highest native ISO on the 5D II is two full stops more sensitive at ISO 6400.
On page 3 of this thread Steven SPencer quoted Rob Gilbraith:
the array of red, green and blue coloured filters over the sensor have been made more transmissive, which effectively bumps up the sensor's light sensitivity, plus they tweaked the way the sensor's signal (the light it has gathered during the exposure) is amplified and then read out.
So the 5D2 sensor appears to be a 1Ds3 sensor with a thinner rgb filter and different read-out electronics.
Don Clary wrote:
So the 5D2 sensor appears to be a 1Ds3 sensor with a thinner rgb filter and different read-out electronics.
That constitutes a different sensor, in my mind. Who's to say what specific kinds of changes to a sensor constitute a 'real' change? The RGB filtering is an important component of sensor design and (at least in this case) the sensor's light sensitivity has been improved by altering the filters and electronics.
Two stops higher native (not "expanded") ISO is not something you can get by tweaking software in Digic IV. You need a real change (or changes) to the sensor to achieve that. Whether the changes are to the filters and electronics or the microlenses and gaps between them is immaterial -- in both cases the physical sensitivity of the sensor to light has improved.
As I mentioned previously - I'd like to see some side-by-side tests (of the same subjects in the same light) to determine whether these changes will translate into a real-world advantage or not.
These arguments tend to get heated... Hehe.
I agree with the threadmaker. In all aspects, photographic detail (and noise) should be compared with the apparent magnification being equal. 5700x3800 pixels in digital, or a good quality print with 300DPI resolution at size 19" by 13" is the norm for what we can faithfully reproduce with DSLR equipment today. When the time comes for higher MP amounts, we will have to increase those values.
The most important thing to remember here is HOW DO WE USE the output of our cameras? Do we all sell poster size originals to be printed at x feet by x feet? Do we all sit and zoom 4x into screen-filling pictures to go "ooohhh" and "aaahhh" at the sharpness of them, every one of them that we see?
If my norm is screen representation at 1920x1200 pixels at 24" diagonally, viewers will see my end-results at that resolution and magnification. This is what I normally do in my own workflow as an amateur photographer. Give/take some cropping/straightening in the picture, and I would want maybe 2500x1600 pixels to choose from. But I want those pixels to be pristine, to be just perfect... This can be achieved from downsizing 12million VERY GOOD pixels, or from downsizing 21million MERELY GOOD pixels, the end result will for me as a viewer of my picture on my screen, be exactly the same.
My workflow:
1 - RAW-process, crop/level/WB in LightRoom, export 16-bit proPhoto .tif, no resize.
2 - (Maybe) details and effects in PS. export 16-bit proPhoto .tif, no resize.
3 - Batch resize to 1920 x something and convert to an AdobeRGB .jpg with ImageMagick (better resize and compression quality than Adobe's products)
4 - Save pictures to my own album, publish a few on the Net, and show/share with my friends at home.
I archive the RAW's if I would like bigger prints/higher resolutions of something later.
My guess is that working photographers don't even HAVE discussions like these. Those I know (I work in the pre-press / printing business) never mention it. Quality per pixel at a certain resolution (as in my example - 1920x1200) matters in digital, quality per square cm at a certain print size matters in physical form. Nothing much else.
This is why a Sinar or a Phase picture looks better than a DsIII or a D3, even in a "tiny" magazine print or a web-sized digital copy - it has almost 3 full stops of dynamic range more and therefore a more nuanced and EXACT colour and highlight/shadow representation (IF the rest of the repro-chain can handle it). This will be more apparent than if a 12MP or a 21MP camera was used.
My ideal camera, for my intents and purposes, would be a Foveon sensor with adequate resolution (~10MP?), the PER-PIXEL iso capabilities of the D3/D700 and the dynamic range of a MF digital back. In a rangefinder body. Could someone please do that for me? :P
> The most important thing to remember here is HOW DO WE USE the output of our cameras? Do we all sell poster size originals to be printed at x feet by x feet? Do we all sit and zoom 4x into screen-filling pictures to go "ooohhh" and "aaahhh" at the sharpness of them, every one of them that we see?
I think if you know in advance everything about the future use of the image, that argument is fine, but for me at least that is seldom the case. I see it as my duty as a photographer to capture at the highest quality that I reasonably can, so that my clients have the FACILITY to use the picture in the broadest range of ways. For example, it's impossible to know in advance whether there might be a use of the picture which involves extensive cropping. Or the client might suddenly ask if the shot that was made for a postcard could also be used for a publicity poster. In order to provide that kind of flexibility in the future, I don't think it makes a lot of sense for many photographers to settle for a lesser capture quality than the best current DSLR can give you.
brainiac wrote:
Which 5.25 megapixel camera are you using? Most 5D users are used to 13 megapixels, so the mkII will be about double the number of pixels. In order to double resolution, you would need to quadruple the number of pixels to 52 million. Read what I said more carefully.
Who would have ever thought that grown men would engage in lengthy debates about "pixels." Our founding fathers would have frowned upon such an expenditure of intellect. (On the other hand, "pixel peeping" sounds more tintillating, even somewhat naughty).
National Geographic print some RIP's at 300lpi, the highest resolution commercial volume print that I know of. This translates to a 4500x6000 pixel digital copy. The normal (think Vogue and so on) is 120-133lpi - that's 2100x3200 pixels for a full frontpage. A high-end blow-up (more expensive than most people would EVER consider) is 66-80lpi - that's 5700x8600 for a 36"x48" poster. Roll-based dye-sublimation printers with 120lpi capabilities are available in 24" widths (5700 pixels).
All prints below RIPped offset in quality produce lower colour-fidelity and substantially lower resolutions than they state. A 4800dpi inkjet will realistically produce 200-240dpi measured usable output, with typically quite low colour fidelity, especially in the already shoddy highlight and deep shadow ranges.
Bear in mind that my experience is from commercial printing, not fine-arts printing. Higher resolution and more colour precise printing can be achieved with supplemental colour laser lithographies, but the costs... Very few (not in the printing business) will ever see a print like this in his/her lifetime outside of the great museums.
If assessing picture quality means that I have to scroll 3 screens sideways on my calibrated 107% AdobeRGB gamut 2.400$ screen to see the whole picture, then the pictorial value has been, lets see... Kinda lost. It's just a technical exercise. Work. I assess a picture when I can see it in full. This means backing away from a big print, or downsampling on a screen.
The best original capture quality attainable is always a philosophical goal, of course. But for "normal" use, enough is enough - reasonably. Even in commercial works. If you suspect your work will be published BIG then you don't use a DSLR if you can afford not to. But I see your point about the customer wanting to crop heavily, I do...
Other than this, more resolution right now is mostly for your own ego's sake - for knowing that there's more detail than you can see, not for the potential customer who in (approaching) zero cases will be able to get a better end product from just having more MP. Downsampling high-ISO pictures to make noise more fine-grained is an exception to this argument.
I have (with certain limits) access to P45's with battery packs and almost pocketable Sony Vaio laptops with 320G available HD space, but I don't borrow them to go snapping away at weekends. At my normal picture count (as I mentioned, I'm in pre-press, not a working photographer) that would anyway fill several archive DVDs per month, easily. Not worth it for me, and it wouldn't make me a better photographer - or the pictures any better (I think).
theSuede wrote:
> Other than this, more resolution right now is mostly for your own ego's sake - for knowing that there's more detail than you can see, not for the potential customer who in (approaching) zero cases will be able to get a better end product from just having more MP. Downsampling high-ISO pictures to make noise more fine-grained is an exception to this argument.
With respect, there are a great many photographers for whom well executed 21 megapixel captures from 35mm type gear will obviate carrying a second cumbersome and time-consuming kit to the job: http://cyberphotographer.com/m8v5d/images/393.jpg
> I have (with certain limits) access to P45's with battery packs and almost pocketable Sony Vaio laptops with 320G available HD space, but I don't borrow them to go snapping away at weekends. At my normal picture count (as I mentioned, I'm in pre-press, not a working photographer) that would anyway fill several archive DVDs per month, easily. Not worth it for me, and it wouldn't make me a better photographer - or the pictures any better (I think).
I suspect that you are being disingenuous here: the reason you don't take the P45's and Vaios with you is because they are cumbersome, time-consuming and inconvenient, not because they have too many pixels.
I am glad that the majority of people are prepared to upgrade to higher megapixel cameras and can understand the intrinsic value of that. A3 and A2 pigment ink printers are better and cheaper than ever now, and improving every year. Taking photos is to some extent an investment for the future. Even if the average photographer doesn't have access to higher resolution printing and display technology right now, she very likely will in the near future.
If you would ever feel the need to make a 1-4 person crop from your picture above, I stand corrected. Having the MP to be able to do that, and still present the end result in large formats is, of course, a justifiable end mean. I take it that that picture is 5616 pixels wide (haven't checked the exif), and using those pixels to the current maximum REAL performance of normal commercially available printers, that would be app. 30" (~2.5feet) wide. Printed at that resolution, you would still loose a lot of actual colour resolution, so you could go twice the size by upsampling with care and the right algorithms for the job - without loosing any picture acuteness (if you don't compare by loupe), and actually gain more colour information. this means that you could get a 10" wide of a four person crop from the picture above right now - if it is approaching pixel perfect in capture. As blowups go, that's of course better than the 7" wide available from a camera with half the resolution.
My point is still that I'd rather take the P45's dynamic range and colour resolution than its pixel count if I got to choose what parts of it's performance I'd like to have in a DSLR. When we have DSLRs that can do those things as good as the MF-backs (it's not that hard at lower ISO's, actually - we just have to stop the MP race for a while and allow the rest of the performance to catch up..), THEN by all means continue piling on the pixels!
More pixels of the same per pixel quality is of course better than fewer. No argument there, as long as you have the computing capacity!
theSuede wrote:
My point is still that I'd rather take the P45's dynamic range and colour resolution than its pixel count if I got to choose what parts of it's performance I'd like to have in a DSLR. When we have DSLRs that can do those things as good as the MF-backs (it's not that hard at lower ISO's, actually - we just have to stop the MP race for a while and allow the rest of the performance to catch up..), THEN by all means continue piling on the pixels!
More pixels of the same per pixel quality is of course better than fewer. No argument there, as long as you have the computing capacity!...Show more →
OK - but what evidence is there that CaNikon have compromised on image quality for the sake of pixel count? My latest DSLR's seem to have better colour, tone, and DR than the ones I used to use in the bad ol' days, despite ever increasing numbers of pixels. I really don't see any necessary connection between numbers of pixels and the other aspects of image quality, and luckily nor does CaNikon. MFDB's tend to be CCD, which is a slightly different technology, and some of the colour and tone advantages at low isos may be due to that. For me, in a 135 format DSLR, high iso is more important than dynamic range, and the sacrifice of dynamic range that CMOS _perhaps_ requires, isn't enormous. At the end of the day, it's my belief that the pictorial advantages of MFDB's over 135 format cameras, such as they are, (and quite small in the case of the 1Ds3), are due to the sensors' larger size, not the number of megapixels per square inch. So it's not about pausing the megapixel race to gain better quality. Compact cameras have far far higher pixel densities, so we are nowhere near the limits in DSLR. That's why I think the Leica S2 is an exciting development. It's all about sensor size, not how many pixels it has.
See those larger D700 pixels paying off? Me neither. More pixels are not necessarily an obstacle to great image quality. The pixel race ends with the lens.
Brainiac wrote:
>>See those larger D700 pixels paying off? Me neither. More pixels are not necessarily an obstacle to great image quality. The pixel race ends with the lens.
BenV wrote:
the 1Ds definatly has the better image quality and detail, due to the more megapixels in the image when you downsized :-)
Don't judge the sharpness. Focus was not in exactly the same place for both shots. Just look at the noise, colour, and dynamic range. The D700 is better because it has less banding in the deep shadows, but there is very little in it. Those who think that more megapixels necessarily spoil the party haven't done their homework. Everybody is raving about the D700 sensor. The 1Ds3 sensor pulls the same stunts but with 9 extra megapixels of detail in the lower iso range. That ought to be almost impossible if pixel density has anything to do with it.
So, the D700 upressed to the 1DsIII "magnification" they both have about the same noise-characteristics.
Now, if noise cancels out when downsizing (without taking sharpness or other image-quality impacts taken into account because of it) to the size of the D700, shouldn't that result in a lower noise 1DsIII image?
I'm guessing the answer is no, but I don't know about the noise being cancelled out aspect.
