APS-C sensor in new compact - trend?

In view of the imminent (tongue in cheek) arrival of the Sigma compact (DP1?)

with the same APS-C sensor as fitted in the SD 14 - 14 meg.

 

Should we expect this camera to set new trends for compacts, will it sell in

sufficient volume to make the "Big ones" sit up and pay attention, or will it

only be purchased by a limited audience who need a compact that produces DSLR

quality images?

 

It will be interesting to see if this will be a trend setter, or a touch of

white elephant.

 

Either Sigma will price it out of range for most of us requiring a high

quality compact, and sales will be very limited resulting in a market "own

goal", or they will market it well at a reasonable price and sell millions,

thus pushing the big boys into reacting and producing APS-C sensor compacts

for sensible prices.

 

All of the above depend of course on the camera/lens combination living up to

expectations, and being capable of delivering the goods!

 

Thoughts, or are these just my inane ramblings scattered with an element of

hope and desire?

I think the purpose of DP1 is (1) the maximize returns for the development costs as it, most likely, share with SD14 the sensor integration, the focus system and the in-camera/ PC software; and (2) to sell the sensor in larger quantities and thus make it's implementation more cost effective or, perhaps, even possible.

 

The big guns of the industry are unlikely to notice it in my opinion. They are big enough to dictate the market and to annihilate any independent development. Am I too cynical?

 

I for one am eagerly awaiting the entrance of the SD14/ DP1 and hope the cameras to be a commercial success.

sell millions?

 

I dont think sigma will become a household name because of the aps sized sensor (not so ) compact, even if sigma sells it for $300-,

 

Why? because it doesnt even look cool, sexy or something similar. The market trend is going toward smaller size with higher capability( or gadget factor).

 

IMO, the design of the sigma aps P&S looks ugly and odd.

<blockquote>Should we expect this camera to set new trends for compacts, will it sell in sufficient volume to make the "Big ones" sit up and pay attention, or will it only be purchased by a limited audience who need a compact that produces DSLR quality images</blockquote>

<br>

Most likely the latter. Even then the DP1 isn't likely to be that popular as a 28mm f4.0 lens is a compromise lens. Yes it is relatively wide but it is also relatively slow. <br>

<br>

Maybe the Foveon will have excelent high ISO performance but you still wont have the depth of field control of a FujiFilm Natura [24mm f1.9] or even a Fujifilm Klasse W [28mm f/2.8]. Even the little Fuji compact zooms Natura Classica and Silvi F2.8 are more flexible cameras.<br>

<br>

The catch is all four are film cameras so not as convenient as a digital compact, and compared to digital compacts they are a little bulky.<br>

<br>

If Sigma had the sense to provide a digital equivalent to the Natura [24mm f1.9] or the Silvi F2.8 [24mm-50mm f2.8], that looked at least as good as the Fujis (consumers care about appearances) and maybe a tad smaller than the Fujis then they would sell like hotcakes and they wouldn't know what to do with all the money they would make.<br>

<br>Maybe Sigma ahould have Fuji build the bodies for them and use their own sensor, then they would have a fine little compact.<br>

<br>

And before anyone asks<br>

1. I don't work for (or own any shares in) Fujifilm.<br>

2. I don't own any of the cameras listed above (<i>but will accept any as a donation :)</i>)<br>

3. I think that the Foveon sensor is an innovative design.<br>

4. I own a few Sigma lenses that I think are lovely.<br>

<br>

Well enough rambling, final thoughts:<br>

<br>

I don't think that there will be a new trend for APS-C compacts (although I'd like to see to one) until there is a company out there that offers a really good camera with a decent wide angle lens (as the first consideration), with a decent APS-C sensor (as a secondary consideration), at a price that engages the mass market.<br>

<br>

I don't think that the DP1 is that camera, but then again I'm just making a decision based on the published information (which may be all that is ever produced at this rate).

I'm waiting for an APS pro-sumer which isn't reputed to be as slow as a carthorse.

<I>Sigma compact (DP1?) with the same APS-C sensor as fitted in the SD 14 - 14 meg</I><P>

 

This statement contains two misconceptions: (1) that the Sigma / Foveon sensor is APS-C size (it is not), and (2) that the SD 14 is a 14 megapixel camera (or whatever "14 meg" means--it is not really).<P>

 

The Foveon sensor is 13.8 x 20.7 mm, just barely bigger than the 4/3 sensors in the Olympus E-500 and E-330, wich are 13.5 x 18.0 mm, and appreciably smaller than, e.g., the 15.8 x 23.6 mm sensor in the Nikon D80. Seen another way, the Sony has a 1.52x factor, the Sigma a 1.74x factor, and the Olympus has a 1.78x or 2.00x factor (depending on which side you want to look at, given the different aspect ratio). In terms of functional size for typical 8x10 and 11x14 inch prints, the Sigma / Foveon is much closer to 4/3 size than to APS-C.<P>

 

As to the resolution, 14 million 'detector sites' or whatever Sigma calls them is not the same as 14 million pixels, because the sites are stacked three-deep in the same place. The specs I've seen give the top resolution of 1760 x 2640 pixels, which is 4.6 MP. However, because there is no loss of resolution due to Bayer sensor issues, a 4.6 MP Foveon-equipped camera should give about the same resolution as a 7.5 MP Bayer-sensor-equipped camera.<P>

 

I wish Sigma luck--I actually like the Foveon sensor on a theoretical level. But there is a lot of misinformation on Sigma cameras, so much and of such character that one has to wonder whether it is to some extent (at least on Sigma's part) intentional.

I wouldn't like the thread to deteriorate into a pointless exchange of Foveon vs. Bayer but the whole issue of the MPs resolution is a complete nonsense perpertuated to a no end.

First of all, a sensor has no PICture ELements but photo sensing sites.

Second, it's absolutely misleading to represent the sensor linear resolution (or resolutions summary) as the true resolution. Correct me if wrong but as far as I am aware, in the typical Bayer filter array sensor half of the registered data from the monochrome photo sites is used for the luminance (those under the Green filter) and 50% of the data is used for the chrominance ( 25% under the Red and 25% under the Blue filter respectively). Sophisticated algorithms fill the gaps (pun intended). Nothing wrong indeed - in practice the end is important not the route to it - as long as it's described correctly as Interpolated Resolution.

Do the maths to what really equal the common 6, 8 and 10 MPs sensors.... in reality only 1/3 of that (approximately) is based on registered at the photo sites data.

 

If there is something to blame Sigma/ Foveon it is the fact that they joined the rest of the manufacturers chanting the nonsensical oversimplification called Megapixels (Bayer) resolution. Hence the SD10 became 10MPs and the SD14, well, you guessed. In fact hey are 3.3/ 4.7 millions full colour (at the) photo sites sensors. I see nothing be ashamed about.

I think the main problem with this type of a camera is that you cannot make a compact camera with that size sensor, because the larger sensor demands a larger lens. Maybe I am wrong, but I think it's true. For example, Sony R1 is much larger than Nikon 8400, and their main difference is the sensor size.

"Momo Vuyisich , feb 07, 2007; 08:14 p.m.

I think the main problem with this type of a camera is that you cannot make a compact camera with that size sensor, because the larger sensor demands a larger lens."

 

True. A brighter 28mm lens, f2.8 or more, for the DP1 sensor size would be considerably bigger. I suppose, one of the DP1 design briefs was to be pocket sized camera.

I was under the impression that the Sony DSC-R1 has an APS-C sized sensor. I realize it isn't a pocket-cam, but neither is it a dSLR. I would guess that from time to time, we might see a digicam with a similar-sized sensor. Probably aimed at the enthusiast market rather than Joe Sixpack. Just my guess, though.

Who is Joe Sixpack? What is he known for?

"Who is Joe Sixpack? What is he known for?"

 

He is the subject of middle class jokes. In the US it is not politically incorrect to denigrate the working class, unlike other "minorities" of race, ethnicity, or gender.

 

It should come as no surprise that the behaviors and attitudes ascribed to Joe Sixpack are those of the middle class.

 

The proper term of art for this phenomenon is called, I think, 'displacement'.

 

--

 

Don E

Interesting. I used it to mean "the average consumer." Or does that have some sort of connotation I'm unaware of as well?

"Interesting. I used it to mean "the average consumer." Or does that have some sort of connotation I'm unaware of as well?" Wigwam

 

In its current denatured usage, you are correct. It is related, I think, to "Joe Lunchpail". Once upon a time, before we became consumers, we were, instead. workers and citizens.

 

Both are language used by middle class people who are unlikely to refer to their doctor, accountant, lawyer, fellow "suits" or professionals as a joe sixpack or a joe lunchpail, even though such people typify the average consumer, and are the target of much advertising due to their disposable income (or credit/debt limits). But neither are working people inclined to describe themselves their family members, co-workers, or neighbors as a joe sixpack or joe lunchpail. At least, I have never heard it and I am embedded in the industrial working class.

 

It is a term of contempt and disdain, and to use it is, at the least, bad manners.

 

--

 

Don E

Although I do my best to avoid terms of derision and racially-charged phrases which, although once part of the common lexicon, are now considered offensive, I have a hard time finding the term 'consumer' offensive in any way. If I can't refer to a consumer (a class I am a member of) as "Joe Sixpack," I'm going back to referring to the average whatever as a 'mook'. Political correctness taken to such levels makes my skin crawl.

Two things:<P>

 

(1) Yes, the problem with a "compact" camera with an APS-C-size sensor (or one nearly so, like the Sony R-1's) either will be fairly big (like the Sony, in which case you might as well carry a compact DSLR) or have a fixed focal length lens (like lots of compact cameras, many of them rangefinders, of 30-40 years ago). Not too many camera buyers will forego a zoom lens, and to get a decent-quality zoom with a decent zoom range in a reasonably compact package, you have to use a small sensor. Those who would settle for a fixed lens of, say, the equivalent of 28mm or 35mm in order to get an appreciably bigger sensor are a tiny market.<P>

 

(2) <I>Do the maths to what really equal the common 6, 8 and 10 MPs sensors.... in reality only 1/3 of that (approximately) is based on registered at the photo sites data.</I><P>

 

This is completely incorrect. Bayer sensors use clusters of four pixels, 1 red, 2 green, and 1 blue. They use a de-mosaicing algorithm to turn the R-G-B-G data from 4 elements into 4 color pixels. In the process, they do not lose 67% of the resolution, as was suggested in the comment; in typical cameras, they lose about 35% or 40% of the resolution. I absolutely stand by my earlier statement that a 4.6 MP Foveon image is likely to have about the same real resolution as a 7.5 MP Bayer sensor image.

"I'm going back to referring to the average whatever as a 'mook'. Political correctness taken to such levels makes my skin crawl."

 

I'm told the feminine of 'mook' is 'midriff'.

 

You illustrate my point that political correctness cannot extend to class. We can blithely refer to sixpacks and lunchpails without a twinge of concern, unlike a reference to his race or ethnicity -- class unconsciousness, to coin a term.

 

Anyway, Joe's job has been offshored; he can't afford a sixpack and his lunchpail sits unused in the cellar. Let's hope he's not the average consumer anymore.

 

On topic: although an APS-C sized sensor in a compact is unlikely to satisfy the desire for a digital '70s rangefinder' due to size issues, perhaps a 4/3". Has any digital used a 1" sensor? That would be slightly more than half the diameter of an APS-C and about double the diameter of the common compact sized sensors.

 

--

 

Don E

Why does an APS-C sized compact camera have to be BIG? The Yashica T4 and Olympus Stylus Epic FILM cameras were tiny and their "sensor" was full frame.

"(2) Do the maths to what really equal the common 6, 8 and 10 MPs sensors.... in reality only 1/3 of that (approximately) is based on registered at the photo sites data.

 

This is completely incorrect. Bayer sensors use clusters of four pixels, 1 red, 2 green, and 1 blue. They use a de-mosaicing algorithm to turn the R-G-B-G data from 4 elements into 4 color pixels. In the process, they do not lose 67% of the resolution, as was suggested in the comment"

 

 

Really? I thought that the algorithms employed in the case of the Bayer sensor type work a bit diferently.... 1 "pixel" use the the information of ALL of the surrounding pixels to determine the correct colour value. In that case it's not "4 elements into 4 color pixels" as stated above but more in the lines of 16 elements sharing data to get 4 colour pixels. Each filter element discart 2/3ds of the chroma information from the incoming light as it records only one of the primaries.

The spatial resolution, conveniently ignored by you, is derived from the green channel only, which is 50% of the filter array. Hence, the actual resolution differs for the luminance and the chroma. At best, the bombastic resolutions should be halved (luminance); for the actual recorded colour information they drop to the modest 1/3.

 

"in typical cameras, they lose about 35% or 40% of the resolution."

 

The typical cameras can't lose something they never had in the first place. Period.

In my original comment I stated that they create missed data (through interpolation). "Creating" and "Lose" are not interchangeable.

 

"I absolutely stand by my earlier statement that a 4.6 MP Foveon image is likely to have about the same real resolution as a 7.5 MP Bayer sensor image."

 

You may stand by your earlier statement as much as you like but it does not correspond to the facts! The Foveon type of sensor as employed in the SD14/ DP1 has a real resolution of 4.6 MP. The bigger numbers are simply a speculation of how it may compare to the interpolated linear resolutions of the commonly used Bayer type sensors.

Don E.,

 

I am afraid I do not share your 'idee fixe' regarding either the vanishing middle class or the nobility of labor versus any other means of employment. I appreciate your Biercian wit, however, and so I dip my banner and retire.

"Why does an APS-C sized compact camera have to be BIG? The Yashica T4 and Olympus Stylus Epic FILM cameras were tiny and their "sensor" was full frame."

 

The way film and sensors capture the light is different. The angle of incidence of the light striking film is not an issue and fast wide and compact lenses are possible at a reasonable price-point. It is different with a sensor where the angle is critical.

 

--

 

Don E

"I am afraid I do not share your 'idee fixe' regarding either the vanishing middle class or the nobility of labor versus any other means of employment. I appreciate your Biercian wit, however, and so I dip my banner and retire." Wigwam

 

The first sentence above refers to nothing I wrote, Wigwam. I do not want to continue this off-topic, and I did not intend a criticism of you.

 

I'm posting this because I did not express any "'idee fixe'", nor anything on a "vanishing middle class", nor on the "nobility of labor". To have responded that way seems to confirm a 'nervousness' the subject of "class" can evoke.

 

I believe the politcally correct term for "working class" these days is "Middle class working families". It may be a Clintonism, but is a purely politician-originated phrase, derived from "working families" as used in news stories about "latch key kids" (which you may recall) popular a decade or more ago.

 

--

 

Don E

<I>Why does an APS-C sized compact camera have to be BIG? The Yashica T4 and Olympus Stylus Epic FILM cameras were tiny and their "sensor" was full frame.</I><P>

 

The Stylus Epic has a fixed 28mm lens, right? And doesn't the T4 also have a fixed-focal-lenght lens? I pointed out that you could have a larger sensor on a compact digital if you're willing to forego the zoom lens. Also, today's compact digital is much smaller than the compact 35mm of ten years ago; we have a Fuji Discovery 1000 Zoom, ca. 1994, that has a decent 35-80mm f/3.8-7 (or thereabouts) lens, but it is probably 3-4x the size of today's average digital compact.<P>

 

<I>re real resolution, Foveon and Bayer sensors, and Ivaylo's bizarre assertions</I><P>

 

Ivaylo, if you don't believe me, why don't you go to some reputable independent testing groups and look up their test results for cameras with Foveon and Bayer sensors? The results are typically expressed in lines-per-picture-height for vertical and horizontal resolution. To get the real, effective resolution, you multiply the vertical resolution by the horizontal resolution by the aspect ratio (to compensate for the normalization of vertical resolution to lines per picture height). You will find that good Bayer-sensor cameras have <I>linear</I> resolutions averaging nearly 80% of thier pixel counts (e.g., 6 MP DSLR with 2000 pixel vertical resolution will typically have vertical resolution of close to 1600 lines), so their effective resolutions in megapixels are about 60-65% of the nominal pixel count. You will also find that Foveon sensor cameras average about one line per pixel. Go ahead, go read up at DPReview or Popular Photography or some other source that actually does instrumented lab testing of resolution. You will find I am correct.<P>

 

How can this be? Well, sophisticated signal processing can do a pretty good job at, e.g., estimating the red and green components of a blue pixel in a Bayer sensor, based on the values of the adjacent red and green pixels. This estimation is not perfect, of course, but it is good enough so that linear resolution is, as stated above, close to 80% of the pixel count.

"Ivaylo, if you don't believe me, why don't you go to some reputable independent testing groups and look up their test results for cameras with Foveon and Bayer sensors? The results are typically expressed in lines-per-picture-height for vertical and horizontal resolution. To get the real, effective resolution, you multiply the vertical resolution by the horizontal resolution by the aspect ratio (to compensate for the normalization of vertical resolution to lines per picture height). You will find that good Bayer-sensor cameras have linear resolutions averaging nearly 80% of their pixel counts (e.g., 6 MP DSLR with 2000 pixel vertical resolution will typically have vertical resolution of close to 1600 lines), so their effective resolutions in megapixels are about 60-65% of the nominal pixel count. You will also find that Foveon sensor cameras average about one line per pixel. Go ahead, go read up at DPReview or Popular Photography or some other source that actually does instrumented lab testing of resolution. You will find I am correct.

 

How can this be? Well, sophisticated signal processing can do a pretty good job at, e.g., estimating the red and green components of a blue pixel in a Bayer sensor, based on the values of the adjacent red and green pixels. This estimation is not perfect, of course, but it is good enough so that linear resolution is, as stated above, close to 80% of the pixel count."

 

Hi Dave,

 

I am afraid you misread my comments, or at least there is a misunderstanding between us. In essence you make a reference to the produced points in the image whereas my point was the discrepancy between captured and produced points. To assume the two are equal when the later in case of the Bayer type sensor is partly software generated, and to use the numbers for calculated reference (implying it's sensor capture) is simply incorrect.

 

Perhaps would have been better if the lines-per-picture-height comparisons were referring only to the cameras as a whole units and were clearly differentiated from the actual sensors capture.

 

Would you assume a vocabulary test result valid if somebody used highly sophisticated predictive text input software to produce the papers? Indeed I am simplifying a bit for the purpose of illustration but I hope you get my point.

 

If one is so much interested in the sensor generated resolution and insist of expressing it in lines-per-picture-height a pre-requisite for a meaningful result is all of the the photo sites to behave equally, that is, luminosity and chroma to get their measurement at each photo site. At present that apply only to photo sites with vertically stacked filter arrays (Foveon and the experimental work of Fuji, as far as I know).

 

Otherwise what is the point of an exercise of multiplying the number of photo sites when they measure the light in highly differential manner and produce qualitatively unequal data?!?

 

Frankly I would prefer not to go to DPReview or Popular Photography as I regard such sites as the online presence of the industry (retail included) PR machine. Their entire raison d'être is Spin and the quasi scientific instrumental testing is one of the biggest, if not the biggest, tools in their hands.That's why we are blessed with bombastic resolutions, the day dream of the full frame golden standard, noise neurosis, you name it.

<I>Perhaps would have been better if the lines-per-picture-height comparisons were referring only to the cameras as a whole units and were clearly differentiated from the actual sensors capture.</I><P>

 

The lines-per-picture-height comparisons <B>are</B> referring to the cameras as a whole units. In other words, when testing a camera, they take a picture of a test chart, download the picture file from the camera, and examine the picture to see how much resolution the camera produces. The whole point is that such tests don't care <I>how</I> the resolution is achieved (Bayer sensor, Foveon sensor, three separate sensors as in some camcorders, etc.); the tests just tell you how much resolution the camera actually achieves (under the test conditions, usually a high-contrast black-and-white test target etc.).<P>

 

Such tests have generally shown that Foveon-equipped cameras produce roughly the resolutions that their pixel counts (<I>not</I> their 'photo detector' counts) imply; and that Bayer-sensor cameras produce linear resolutions of 75-80% of what the sensor pixel dimensions imply, and therefore about 60-65% of what their pixel counts imply. So a Foveon-equipped camera with 13.9 'photo detectors' or whatever Sigma calls them, producing a file with 4.6 MP, will very likely give you a real-world resolution very similar to a Bayer-sensor camera with a sensor with 7.5 million Bayer (R-G-B-G) pixels, which also produces (through de-mosaicing software) a file with 7.5 MP.