6MP prosumer?

solu

Member
Messages
17
Reaction score
0
It seems to me that almost every manufacture of 5MP prosumer cameras is using the same 2/3" CCD chip. When will we see any 6MP (or next chip generation) cameras replacing F717, Nikon 5700, 7hi etc.?? Is it a question of filling the market before releasing a new chip or is it perhaps still too expensive?? The DiMage 7 is from 02 2001 - which is incredible long time ago considering the chip technology - and I think we are only halfway trough the evolution of digital cameras!

solu
 
Are you willing to pay $100 more for only 20% increase in pixel count? The benefit of everyone using the same Sony 2/3" chip is that the sensor (and thereby the camera) prices will fall dramatically due to economies of scale.

I agree that it has been a while since a new innovation in CCD technology has come onto the scene. Sony seems content with it's current HAD CCD chip design (as it is by far the most popular type). Foveon's X3 chip seems to be stuck in a quagmire of design difficulties (I mean for the next generation of chips). Fuji has some interesting Super CCD stuff coming out... eventually... supposedly, we'll see how it goes.

--Arvin
It seems to me that almost every manufacture of 5MP prosumer
cameras is using the same 2/3" CCD chip. When will we see any 6MP
(or next chip generation) cameras replacing F717, Nikon 5700, 7hi
etc.?? Is it a question of filling the market before releasing a
new chip or is it perhaps still too expensive?? The DiMage 7 is
from 02 2001 - which is incredible long time ago considering the
chip technology - and I think we are only halfway trough the
evolution of digital cameras!

solu
 
The trend seems to be toward improved pixel quality in 3MP to 5MP
sensors. I think most people have realized that bigger file sizes
are a headache for the marginal increase in picture quality. I think
more and more people are looking for real improvements in image
quality, not just more pixels.

Just my opinion.
Darrell
 
It seems to me that almost every manufacture of 5MP prosumer
cameras is using the same 2/3" CCD chip. When will we see any 6MP
(or next chip generation) cameras replacing F717, Nikon 5700, 7hi
etc.?? Is it a question of filling the market before releasing a
new chip or is it perhaps still too expensive??
Most, if not all, of the recently introduced 5MP cameras are using 1/1.8" CCDs, which are smaller than the 2/3" CCDs. It's clear that they are cheaper, and I think the camera industry has decided that bringing prices down, rather than increasing pixel resolution is the strategy that will sell the most cameras now. Whether they're right or wrong about that is another question.

I personally have no interest in a 6MP CCD with tiny photosites (the photosites in a 5MP 1/1.8" CCD are only about 2.8µm square.) I'd love to see a lens-shutter style camera with a pro quality CCD -- a 4/3" or bigger sensor, for example. I just spent some time with the Contax TVS Digital, which is well-built and looks like its T-series film cousins but lacks a crucial ingredient in their success: top quality pictures. It uses a 1/1.8" CCD just like so many other consumer 5MP digicams. Shrug.
 
It seems to me that almost every manufacture of 5MP prosumer
cameras is using the same 2/3" CCD chip. When will we see any 6MP
(or next chip generation) cameras replacing F717, Nikon 5700, 7hi
etc.?? Is it a question of filling the market before releasing a
new chip or is it perhaps still too expensive?? The DiMage 7 is
from 02 2001 - which is incredible long time ago considering the
chip technology - and I think we are only halfway trough the
evolution of digital cameras!

solu
The price point of the Canon 10D has forced Sony et al. to lower their prosumer digicam prices. Their next generation must address the encroaching competition from the 10D (and the *ist and others to follow). I can't see Sony launching a dSLR, rather the V1 seems to have inherited the responsibility of carrying on this line, thus freeing the F7x7 to something better??? To reduce noise, they'll have to move to a larger image sensor. The first camera manufacturer that comes out with a prosumer digicam with image quality similar to 10D, negligible noise at high ISOs, negligiible shutter lag, and fast AF even in low-light would satisfy many of us who've left our 35mm SLR behind and have absolutely no desire to ever move to a dSLR.

My :)
---------------------------------------
Digital Camera Fact Sheets
http://www.photoxels.com
'A Smile Is Forever'
 
The trend seems to be toward improved pixel quality in 3MP to 5MP
sensors. I think most people have realized that bigger file sizes
are a headache for the marginal increase in picture quality. I think
more and more people are looking for real improvements in image
quality, not just more pixels.
I agree. That's why the Foveon X3 in in the Sigma SD9 is so great.

You get the sharpness of 6.9 MP (same number of luminance samples as 6.9 MP Bayer mosaic sensor, that is, 3.45 million) and the color resolution of 13.8 MP (same number of red and blue samples as 13.8 MP Bayer mosaic sensor, that is, 3.45 million of each color) and generating a file size of only 3.45 MP (10.3 megabytes TIFF, one byte per photodetector).

Of course, you'll see the 6 MP consumer cameras soon, too, but don't expect better pictures than the 5 MP; just bigger files. Just like some of the 5's are worse than the 4's and some of the 4's are worse than the 3's (in the same optical format).

j
 
As long as the size of the ccd does not increase, more pixels may be difficult because small er pixels have more noise given other things the same. At some point the noise increase negates the advantages of a few more pixels. I suspect we are at that point with the current consumer ccd technology. Something new might be indicated. Some of the new Fuji technology like the two sites per pixel idea might be a better approach. Leon
 
In principle, the Oly 4/3 system should fill the bill. In many respects what people are asking for is DSLR capability in a smaller size where smaller size should result in lower costs particularly for the glass with no quality penalties. Unfortunately, it is hard to figure out how to get there. Designing a high quality (Any attemp at a lower quality level would be severely panned and maybe dead on arrival. Note criticism of X3 camera.) body with new high quality lenses is a very tough proposition to do in a single step. The best DSLR stuff around today resulted from an evolution through prior models. The 4/3 models shown by Oly are somewhat disappointing in that they don't appear all that much smaller than DSLRs. I guess time will tell. Leon
The price point of the Canon 10D has forced Sony et al. to lower
their prosumer digicam prices. Their next generation must address
the encroaching competition from the 10D (and the *ist and others
to follow). I can't see Sony launching a dSLR, rather the V1 seems
to have inherited the responsibility of carrying on this line, thus
freeing the F7x7 to something better??? To reduce noise, they'll
have to move to a larger image sensor. The first camera
manufacturer that comes out with a prosumer digicam with image
quality similar to 10D, negligible noise at high ISOs, negligiible
shutter lag, and fast AF even in low-light would satisfy many of us
who've left our 35mm SLR behind and have absolutely no desire to
ever move to a dSLR.

My :)
---------------------------------------
Digital Camera Fact Sheets
http://www.photoxels.com
'A Smile Is Forever'
 
The trend seems to be toward improved pixel quality in 3MP to 5MP
sensors. I think most people have realized that bigger file sizes
are a headache for the marginal increase in picture quality. I think
more and more people are looking for real improvements in image
quality, not just more pixels.
I agree. That's why the Foveon X3 in in the Sigma SD9 is so great.

You get the sharpness of 6.9 MP (same number of luminance samples
as 6.9 MP Bayer mosaic sensor, that is, 3.45 million) and the color
resolution of 13.8 MP (same number of red and blue samples as 13.8
MP Bayer mosaic sensor, that is, 3.45 million of each color) and
generating a file size of only 3.45 MP (10.3 megabytes TIFF, one
byte per photodetector).
Bayer demosaicing ain't as simple (or as inefficient) as that. Take a sample from the SD-9 and one from, say, the 10D, up-rez the SD-9 one to 6.3MP or down-rez the 10D one to 3.34 MP, apply USM, and look at the fine detail. They will be as good as identical. (OK, you can design engineer a test scene that'll give the Foveon a slight edge beyond that.)
Of course, you'll see the 6 MP consumer cameras soon, too, but
don't expect better pictures than the 5 MP; just bigger files.
Just like some of the 5's are worse than the 4's and some of the
4's are worse than the 3's (in the same optical format).
I'll join the chorus for better, rather than more, pixels. I hope Fuji's extended dynamic-range sensors sell really well: I hope the market understands the distinction. I'm not too optimistic, though. "But this one has 6MP!" could be the new rallying call...

Petteri
--
Portfolio: [ http://www.seittipaja.fi/index/ ]
Photo lessons: [ http://www.seittipaja.fi/lessons/ ]
 
Unfortunately, teh industry is handful with the wide demand/penetration to make more compact cameras improving upon the sensor material. The care is not that much for what we are asking here. Fuji opted to retain smaller size while responding to better dynamic range. however, this is necessary only for ccd's linear response. Why are not there any cleaner cmos alternative (like in dslrs) , which can benefit from onchip logic and price competiveness?

I am looking for slr-like again. I am still hoping that industry should gather together,

a sensor of x3 sharpness in smaller file with better and neater than existing iso.

an internediate size sensor like 4/3 would be good idea for prosumer as it would add shallower DOF which is missing presently. Some people get good result however using add-on zoom from compact cameras, but noise are still far inferior to DSLRs.

The dissappointing industry trend in the near future if force me to be more realistic, I would look at 602's replacement with SR, but as a compensation :) I would like them to add 30flops mpeg video support with zoom and RAW support, which i guess would be more crucial as there are more data per site.

for now, i am trying to look away for something more suitable by the time hopefully.
The price point of the Canon 10D has forced Sony et al. to lower
their prosumer digicam prices. Their next generation must address
the encroaching competition from the 10D (and the *ist and others
to follow). I can't see Sony launching a dSLR, rather the V1 seems
to have inherited the responsibility of carrying on this line, thus
freeing the F7x7 to something better??? To reduce noise, they'll
have to move to a larger image sensor. The first camera
manufacturer that comes out with a prosumer digicam with image
quality similar to 10D, negligible noise at high ISOs, negligiible
shutter lag, and fast AF even in low-light would satisfy many of us
who've left our 35mm SLR behind and have absolutely no desire to
ever move to a dSLR.

My :)
---------------------------------------
Digital Camera Fact Sheets
http://www.photoxels.com
'A Smile Is Forever'
 
Bayer demosaicing ain't as simple (or as inefficient) as that. Take
a sample from the SD-9 and one from, say, the 10D, up-rez the SD-9
one to 6.3MP or down-rez the 10D one to 3.34 MP, apply USM, and
look at the fine detail. They will be as good as identical. (OK,
you can design engineer a test scene that'll give the Foveon a
slight edge beyond that.)
A slight edge beyond the 10D is what I'm talking about when I say the luminance resolution is like 6.9 MP Bayer. I've done that comparison enough times to know that the difference is readily apparent, even though going from 6.3 to 6.9 is "slight". Not just "as good as identical" as you say.

And yes, better-engineered tests than B&W res charts will show the additional advantage of the additional color resolution. Any picture that shows color aliasing is such a test. See Yamada's test shots with the 1Ds for a good example.

j
 
As long as the size of the ccd does not increase, more pixels may
be difficult because small er pixels have more noise given other
things the same.
That's another point in favor of the Foveon X3. Three times more pixel sensors per unit area, for a given pixel size.
. . . Something new might be indicated. Some of the new Fuji
technology like the two sites per pixel idea might be a better approach.
That seems backwards. Now you cut up the area more finely, packing more pixel sensors per area by making them smaller? It might help with highlight latitude, but it can't help with resolution or ISO.

By the way, in Fuji's terminology, it's not two sites per pixel, it's two photodiodes per photosite. They prefer to reserve pixel to refer to the inflated number of pixels in their interpolated output file (pretty much like what the other Bayer sensors do, but interpolating by another factor of two beyond the usual factor of three).

j
 
The dissappointing industry trend in the near future...
This view is bizarre. How can you think the trend will disappoint you, when you can't see what it is yet? The rate of progress has been incredible, and there's no reason to think it won't stay that way. All sectors, low-end to high-end, will be well served as the technology and markets mature and companies compete every way they can for your business.

Just because our simple extrapolations aren't what vendors choose to do doesn't mean the future will disappoint. Give it a few months.

j
 
It seems to me that almost every manufacture of 5MP prosumer
cameras is using the same 2/3" CCD chip. When will we see any 6MP
(or next chip generation) cameras replacing F717, Nikon 5700, 7hi
etc.?? Is it a question of filling the market before releasing a
new chip or is it perhaps still too expensive?? The DiMage 7 is
from 02 2001 - which is incredible long time ago considering the
chip technology - and I think we are only halfway trough the
evolution of digital cameras!

solu
You really don't "need" a six megapixel consumer or prosumer fixed lens digicam - if it's like the dSLR's, it wouldn't give you any more usable resolution. Why? It's very simple. The aspect ratio on nearly all five megapixel cameras gives you a nearly perfect 8x10 print. With all six megapixel dSLR's the aspect ratio is designed like the 35mm film frame which works out to 8x12. The additional pixels used to make six megapixels are nearly all at each end of the frame with only a minor number contributing to the vertical aspect. Here is an overlay sample I made about a year ago using some of Steve Saunder's photos from a variety of cameras from 2 megapixel up to 6. Notice that nearly all the difference between the 5 and 6 megapixel frames are vested in the areas on each side of the frame. There would be no practical advantage for doing this since nearly everyone crops an 8x12 to 8x10 anyway.

Lin



--
http://208.56.82.71
 
As long as the size of the ccd does not increase, more pixels may
be difficult because small er pixels have more noise given other
things the same.
That's another point in favor of the Foveon X3. Three times more
pixel sensors per unit area, for a given pixel size.
True, but the light energy capture efficiency (in the current implementation anyway) is much worse than on a Bayer sensor (IIRC, about 25% against about 40% for a Bayer sensor). It also appears to get worse for the lower layers of photosites (which is pretty much what you'd expect). This renders the advantage more or less moot.
. . . Something new might be indicated. Some of the new Fuji
technology like the two sites per pixel idea might be a better approach.
That seems backwards. Now you cut up the area more finely, packing
more pixel sensors per area by making them smaller? It might help
with highlight latitude, but it can't help with resolution or ISO.
That's correct, but IMO highlight latitude is currently the biggest weakness of digital over film. I think Fuji's got their head on straight abou this one.

Petteri
--
Portfolio: [ http://www.seittipaja.fi/index/ ]
Photo lessons: [ http://www.seittipaja.fi/lessons/ ]
 
Bayer demosaicing ain't as simple (or as inefficient) as that. Take
a sample from the SD-9 and one from, say, the 10D, up-rez the SD-9
one to 6.3MP or down-rez the 10D one to 3.34 MP, apply USM, and
look at the fine detail. They will be as good as identical. (OK,
you can design engineer a test scene that'll give the Foveon a
slight edge beyond that.)
A slight edge beyond the 10D is what I'm talking about when I say
the luminance resolution is like 6.9 MP Bayer. I've done that
comparison enough times to know that the difference is readily
apparent, even though going from 6.3 to 6.9 is "slight". Not just
"as good as identical" as you say.

And yes, better-engineered tests than B&W res charts will show the
additional advantage of the additional color resolution. Any
picture that shows color aliasing is such a test. See Yamada's
test shots with the 1Ds for a good example.
Better engineered for what?

I can engineer a test that will make the Foveon approach look terrible. I can also engineer a test that will make the Bayer approach look terrible. Neither test is a very good one, IMO, as it doesn't address the real-life performance of the system at all. A well-engineered test, IMO, is one that attempts to replicate a series of real-world test situations: this will give a realistic idea of how the system performs in practice.

Phil's studio scene is an example. Careful examination of a sufficient number of test shots taken by a competent photographer is a better example. Resolution charts are not, and neither is the color aliasing test that you mention (yes, I've seen it). (In fact, the latter doesn't even have the merit of being a well-accepted test for lens resolution; it's clearly a custom-engineered situation that doesn't tell us anything we don't already know, and is only engineered to make Bayer look bad.)

You're also making a very blanket statement about "6.9 MP Bayer equivalence." Bayer sensors and systems aren't created equal. Take four representative scenes, one from, say, the Minolta 7i, one from the Canon 10D, one from the Canon 1Ds, and one from the Kodak 14n (the SLR ones taken with the sharpest lens you can find). Examine them carefully. You'll see that pixel-level sharpness goes up at every step (and moiré shows up in the highest-resolution one).

Petteri
--
Portfolio: [ http://www.seittipaja.fi/index/ ]
Photo lessons: [ http://www.seittipaja.fi/lessons/ ]
 
Agree completely for single shot images. I just went from a 5 MP Oly E20 to a Canon 10D. The short dimension went from 1920 to 2048, a small change. On the other hand, I do panoramas and, unless I want to do 2 dimensional panoramas (a pain), the larger the number of pixels in one dimension, the better. Thus going from 2560 to 3072 was a significant gain. While I will have a slightly larger image count for a given panorama, it is still worth it to get a bit more resolving power without going to 2-d. Panorama people tend to like the 3:2 ratio. Leon
 
I'm not looking to restart the bayer v foveon debate but peddling these theoretical figures around doesn't help inform that much. The SD9 has been around a while now and it's pretty clear that while it does some things brilliantly, it also has a number of limitations which mean that overall the foveon approach is not going to supplant bayer anytime soon unless they can improve on the current problems or ramp up the pixel count whilst maintaining the per pixel quality.

The SD9 image is remarkable for the pixel count but it isn't good enough to show a clean pair of heels to 6MP bayers at the moment at least.
The trend seems to be toward improved pixel quality in 3MP to 5MP
sensors. I think most people have realized that bigger file sizes
are a headache for the marginal increase in picture quality. I think
more and more people are looking for real improvements in image
quality, not just more pixels.
I agree. That's why the Foveon X3 in in the Sigma SD9 is so great.

You get the sharpness of 6.9 MP (same number of luminance samples
as 6.9 MP Bayer mosaic sensor, that is, 3.45 million) and the color
resolution of 13.8 MP (same number of red and blue samples as 13.8
MP Bayer mosaic sensor, that is, 3.45 million of each color) and
generating a file size of only 3.45 MP (10.3 megabytes TIFF, one
byte per photodetector).

Of course, you'll see the 6 MP consumer cameras soon, too, but
don't expect better pictures than the 5 MP; just bigger files.
Just like some of the 5's are worse than the 4's and some of the
4's are worse than the 3's (in the same optical format).

j
 
Fuji marketing may have tried a trick or two with the quoted pixel counts from super CCD but I think you need to bear in mind that the x2 interpolation is apparently a necessary part of the demosaic process for these diagonal array sensors so you can't blame the specs for that...
As long as the size of the ccd does not increase, more pixels may
be difficult because small er pixels have more noise given other
things the same.
That's another point in favor of the Foveon X3. Three times more
pixel sensors per unit area, for a given pixel size.
. . . Something new might be indicated. Some of the new Fuji
technology like the two sites per pixel idea might be a better approach.
That seems backwards. Now you cut up the area more finely, packing
more pixel sensors per area by making them smaller? It might help
with highlight latitude, but it can't help with resolution or ISO.

By the way, in Fuji's terminology, it's not two sites per pixel,
it's two photodiodes per photosite. They prefer to reserve pixel
to refer to the inflated number of pixels in their interpolated
output file (pretty much like what the other Bayer sensors do, but
interpolating by another factor of two beyond the usual factor of
three).

j
 

Keyboard shortcuts

Back
Top