Chaz wrote:
Several of the posts in this thread suggest that landscape photogs who want more m-pixels from a new DSLR should shoot MF instead.
Here's why I want more m-pixels in a new D3X or whatever else comes along:
I've been a Nikon shooter since 1967 while also working with Hassy, 4x5, and 8x10 (studio and on location.) I'm quickly becoming an old fart and can't (or would prefer not to) lug heavy gear to high-country or moderately inaccessible locations for landacape work.
My prints are a minimum of 12x18 with up to 8' generally being the largest on the longest side.
The price/convenience/IQ point for me is now in the 35mm DSLR arena. A new D3X with 24mm T/S is going to suit me just fine versus the cost of a 39MP Hassy or LEAF back. And I will now have a respectable (open to further debate) trade-off against carrying the 4x5 or 8x10 into the field with the attendant holders, etc.
Conclusion: I want more M-pixels for 1)IQ, 2) Convenience and 3) Price versus other alternatives in MF or LF. ...Show more →
No doubt and nobody is suggesting that you don't need more MP. The point that I and other are trying to make is that people with your requierment are a small minority even in the professional market. A D3X would be great, but will Nikon make enough sales to justify its development costs? I think it will at 24MP, but how much further beyond that?
At 3 and 4MP, engineers were figuring out the basics and the electronics outside of the sensors were as much a part of the cost as the sensor itself, probably more. Today, the electronics are mature, but as sensor keep having more MP, the cost of the sensor goes up as does the supporting processors to handle the data. But it goes up as a power law. 2X the MP = 4X the cost but only 50% more resolution and increased noise. It's a law of deminishing returns.
HerbChong wrote:
i don't see why Canon users put up with so much lens flare. there are very few i have ever seen with acceptable flare. i'm talking about veiling flare. until you regularly see lenses without it, you just don't know what you are missing. some Nikon lenses are better, but many are not.
Herb....
OK, but that's hard to understand. I'm not sure what you're doing wrong and why you're unable to control flare, but I'm a WA guy when I'm not in the studio and if any lens will produce flare, it's a WA. Lens flare is unavoidable in some situations, regardless of which mfg made the lens. I'm not talking about diffraction problems...
I reviewed almost a thousand (970+) old images of Havana yesterday (taken in early 2006). I noticed troublesome flare on several; somewhere in the neighborhood of 3 or 4 images. No exaggeration, it was 3 or 4 images, so I'll take the higher number of 4, the lower number of images (970), divide one into the other (4/970 = .0041%) to get my perenctage of images with flare problems.
That's fewer than 1% of all images, taken in Cuba where the sun shines brightly and more intensely than many other places on earth, and I came up with less than1/2 of 1%. To top it off, every image with flare was shot using my Canon Pro1 8MP P&S. It has a great L lens, but each was without a hood in place.
Can I find other images with flare, sure, but it's hard to do and I don't want to crawl through 50-60K images. Cheap filters (UV filters especially) can add flare. I don't use cheap filters and use the best lenses. No flare...
I've been at this for more than 30 years too...I was not including you in the group that seems to think because they found religion in the last 2-3 years they're now somehow experts..
Andre Labonte wrote:
No doubt and nobody is suggesting that you don't need more MP. The point that I and other are trying to make is that people with your requierment are a small minority even in the professional market. A D3X would be great, but will Nikon make enough sales to justify its development costs? I think it will at 24MP, but how much further beyond that?
At 3 and 4MP, engineers were figuring out the basics and the electronics outside of the sensors were as much a part of the cost as the sensor itself, probably more. Today, the electronics are mature, but as sensor keep having more MP, the cost of the sensor goes up as does the supporting processors to handle the data. But it goes up as a power law. 2X the MP = 4X the cost but only 50% more resolution and increased noise. It's a law of deminishing returns....Show more →
you're basing your theory on bayer sensors, something that may not even be around in another 5-7 years...Canon was working on a 50MP sensor in the last few years. That's all anyone knows. It may never come to market but if it does it might not be a bayer sensor. If that's the case, all bets are off....
jamesf99 wrote:
you're basing your theory on bayer sensors, something that may not even be around in another 5-7 years...Canon was working on a 50MP sensor in the last few years. That's all anyone knows. It may never come to market but if it does it might not be a bayer sensor. If that's the case, all bets are off....
Indeed, I wouldn't be horribly surprised to see that the 21-24mp range of DSLR's are the last to use the bayer array. If the manufacturers are going to impress anyone after that point with more resolution using the same tech they'll have to jump to the 36-40mp range and that's pushing the limit of what you can do with that size sensor and that tech. The jump from 16mp to 21mp with the 1Ds line is a bit underwhelming. I hope the manufacturers don't see fit to continue that sort of incremental resolution increase at the high end.
Perhaps we'll see something Foveon like in the next round or two.
As far as sensor development costs increasing with resolution increases, I'm not so sure about that scenario. I've got a feeling we'll see something like what happened in the world of microprocessors. As the technology matures the fabrication and RD costs could go down dramatically. Does anyone have any idea how much it costs to make something like a 386 proc when it first came out versus a modern 2ghz processor?
May 23, 2008 at 03:45 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
jamesf99 wrote:
you're basing your theory on bayer sensors, something that may not even be around in another 5-7 years...Canon was working on a 50MP sensor in the last few years. That's all anyone knows. It may never come to market but if it does it might not be a bayer sensor. If that's the case, all bets are off....
Weather it's bayer or foveon, there basic physics is the same as are the limits. If it's 50MP on a 35mm format, the individual photo sensative sites will have to get smaller and all the same physics comes into play. The bayer pattern simply turns the camera from a basic spacial intensity meter (i.e. B&W) into a color picture. The foveon uses depth to capture the different colors, but pixel density limitations are the same as is the resolution in B&W; the only advantage there is CA issues are much reduced and there's no interpolation for color.
To date, there are two basic semiconductor devices for light capture, CMOS and CCD. The Foveon is just a stacked CCD so the bayer pattern is not needed. In the interest of full disclosure, I work in the semiconductor industry and have worked on the foveon chip.
Andre Labonte wrote:
To date, there are two basic device for light capture, CMOS and CCD. The Foveon is just a stacked CCD so the bayer pattern is not needed. In the interest of full disclosure, I work in the semiconductor industry and have worked on the foveon chip.
It was my understanding the Foveon chip has a higher apparent resolution than each of its actual layers would lead one to believe it would?
I've thought the Sigma DP1 was resolving around the same detail as ~8mp 1.6x DSLR but it's sensor's layers are only around 4.7mp. I could of course be incredibly wrong
Edited on May 23, 2008 at 04:08 PM
May 23, 2008 at 04:08 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
Jammy Straub wrote:
It was my understanding the Foveon chip has a higher apparent resolution than each of its actual layers would lead one to believe it would?
I've thought the Sigma DP1 was resolving around the same detail as ~8mp 1.6x DSLR but it's sensor's layers are only around 4.7mp. I could of course be incredibly wrong
Yes, a wonderful marketing gimmick!
In B&W, resolution is equal the the # of megapixels. In color, the "effective" resolution of a bayer sensor is "about" 1/2 the sensor resolution but is a little better because you have the off-color pixel in between for intensity correction and color interpolation ... this is where image processing comes in. The net result is that you can "see" any color at any pixel even if that pixel does not detect that color ... yea, wierd but it work well.
The cool thing about the Foveon chip is the color resolution and the B&W resolution are the same without any interpolation! So, they equate the color resolution to being equal to what you would get from a similar bayer pattern sensor ... thus claim that 4.7 MP = 8 MP. The reality is that in B&W the Foveon is 4.7MP and is 4.7MP at the 3 primary colors.
Andre Labonte wrote:
Yes, a wonderful marketing gimmick!
In B&W, resolution is equal the the # of megapixels. In color, the "effective" resolution of a bayer sensor is "about" 1/2 the sensor resolution but is a little better because you have the off-color pixel in between for intensity correction and color interpolation ... this is where image processing comes in. The net result is that you can "see" any color at any pixel even if that pixel does not detect that color ... yea, wierd but it work well.
The cool thing about the Foveon chip is the color resolution and the B&W resolution are the same without any interpolation! So, they equate the color resolution to being equal to what you would get from a similar bayer pattern sensor ... thus claim that 4.7 MP = 8 MP. The reality is that in B&W the Foveon is 4.7MP and is 4.7MP at the 3 primary colors....Show more →
Ohh, ok. So does the joining of those 3 primary color layers actually give any tangible increase in resolution over the 4.7mp B&W image? (in terms of actual resolution like lp/ph) (from your explanation I'm guessing it wouldn't)
Wouldn't it also mean there's a real advantage of having a 24mp (per layer) Foveon chip in a camera compared to the 24mp bayer chips were using now?
It would indeed mean that. You'd then have 24MP, where each individual pixel sees all colors and no interpolation is needed. Match that kind of sensor to N or C electronics and it should beat the pants off a 24MP Bayer array.
May 23, 2008 at 04:46 PM
Andre Labonte Offline • • • • • • Upload & Sell: Off
I am probably the less qualified person in this forum to discuss megapixels. I do not do professional work nor do I sell my prints..at least often enough. What follows are my personal impressions:
When I was shooting with the D100 I found that my 13x19 enlargements were superb. I can say the same about the D70S. To my surprise, the 4.1 megapixel files from the D2H also make beautiful enlargements. I know I can go larger than that but I usually print to those sizes.
I am in agreement with others that the D2H would have been a superb tool with 8 megapixels. Even with 4 megapixels I feel very comfortable with it for my casual shooting.
It seems that more megapixels is what many photographers are after. Files will be larger will use more memory and processing those files will be, I guess, more complicated.
My personal opinion on this is that if I need 24 megapixels I could be better served with a medium format sensor.
Just my opinion.
William Rodriguez
Miami, Florida.
Andre Labonte wrote:
To date, there are two basic semiconductor devices for light capture, CMOS and CCD. The Foveon is just a stacked CCD so the bayer pattern is not needed. In the interest of full disclosure, I work in the semiconductor industry and have worked on the foveon chip.
Actually, Sigma DP1 has a Foveon CMOS sensor The resolution indeed is higher than that of Bayer's because no interpolation is required. However, the low frequency response, which determines the size of print, for example, is still set by the MP and not by MPx3.
gfiksel wrote:
Actually, Sigma DP1 has a Foveon CMOS sensor The resolution indeed is higher than that of Bayer's because no interpolation is required. However, the low frequency response, which determines the size of print, for example, is still set by the MP and not by MPx3.
Actually, you are correct ... it's the photo-diode that is stacked and the information is collected by a CMOS based circuit. My bad .... it's been a few years since working on it and I'm not a circuit designer, I'm a process developer.
Resolution in color is higher, as I said, but for B&W the resolutions is the same. But this brings out a great point, it may not be more MP that buys us advancement in the future, but improvements like the foveon design. They have a long way to go before they get the leakage of the stacked diode down to the level of the standard diode, but they'll solve it and then the big boys will convert.
I haven't read the other posts on this occasion, but I am responding to the OP's question:
12Mp (D3 ones) are enough, for me, for commercial (brochure and advertising, some books) work. I'm not getting into any theoretical discussions re. Bayer vs. Foveon here, but I can tell you from looking closely at the DP1 files (JPEGs only, for now) that little 4.7Mp sensor punches above its weight.
re. 12Mp: if you really do need to print above the 300ppi native size of the D3's files, Qimage or similar programs can do this really easily and well—IF the pixels are clean. The D3 qualifies in this regard.
Andre Labonte wrote:
Actually, you are correct ... it's the photo-diode that is stacked and the information is collected by a CMOS based circuit. My bad .... it's been a few years since working on it and I'm not a circuit designer, I'm a process developer.
Resolution in color is higher, as I said, but for B&W the resolutions is the same. But this brings out a great point, it may not be more MP that buys us advancement in the future, but improvements like the foveon design. They have a long way to go before they get the leakage of the stacked diode down to the level of the standard diode, but they'll solve it and then the big boys will convert....Show more →
Foveon is not the only answer and exactly why I specifically didn't mention it. True, it is a "somewhat" commercially viable product today, but it can't deliver across the board as a standard bayer CMOS does. The technology does a few things very well, but if it could/did deliver what's required for commercial success, you'd have the companies converting in mass. That's not happening and I don't see any change on the "visible" horizon.
What most of us should keep in mind is that there may be ways of producing new sensors in the next 5+ years that we're not even considering right now. Very few people know what's buried in the labs at Kodak (see "Truesense" info for recent press), Dalsa, Canon, Sony, et. al., and there are too many factors at play to believe that only Foveon has a chance.
Personally, I think that Foveon will prove a genetic dead end for many reasons but that's only a guess. The 3 layer idea is sound for film substrates, but until they can make a true 18-24MP sensor (which would upstage a similar sized bayer sensor - a 30 year old technology - using interpolated or "made up" data in a head to head competition), the product will remain a little known niche product and not commercially successful.
IMO we're going to see bayer variations for a while longer, and possibly something new. I don't know what it will be, but odds are it won't be Foveon...
