megapixel count frauds, serious and less serious

[dave_redmann]

I know I'm a contrarian, but am I the only one who is bothered by the

fraudulent and borderline-fraudulent claims regarding megapixel

counts that most digital camera manufacturers make?

 

They fall into separate categories, some worse than others.

 

IMHO, the worst is Sigma, which now claims, regarding the SD-10, "NEW

10.2 Million Pixels (3.4 Mp Red + 3.4 MP Green + 3.4 Mp Blue)".

Anyone who knows anything about the SD-10 and its Foveon sensor knows

that these are NOT SEPARATE PIXELS and cannot record different

details since the occupy the same light path. Sigma's prior claim of

10.2 million "photo-detectors" was more true, if not much less

misleading. And since the highest recording mode is only 2268x1512

(i.e., 3.43 million pixels), I think this is fraudulent.

 

Next comes Olympus, which only talks about sensor pixels, and not

effective pixels, on the E-1. They say 5.5 million. Well, the thing

only records from 2560 x 1920, i.e., 4.92 million pixels. To me, any

mention of the total pixels on the sensor, not right next to a figure

for the effective pixels, is also seriously misleading.

 

Fuji calls the Finepix S3 a 12.3 MP camera, even though only through

interpolation does it reach 12.12 million (i.e., 12.1, not 12.3

million) pixels. Again, pretty bad--I don't think anyone thinks

Fuji's interpolation gives you anywhere near full resolution.

 

And then of course are three that seem like nit-picking, after the

above frauds and near-frauds:

 

- "Rounding" that doesn't follow the mathematical rules of rounding,

like Nikon's 12.21 million pixel D2x being called 12.4 MP and almost

everyone else (with Canon usually but not always being the

exception), saying cameras that record 3008 x 2000 pixels (6.02

million) and are called 6.1 MP, instead of 6.0 MP;

 

- quoting the number of pixles recorded on a Bayer (R-G-B-G) or

similar-type sensor, despite the fact that it is theoretically

impossible for such sensors to achieve full resolution (typical

results are true resolutions of around 60% of claimed pixels), and

again, what they use to get their claimed resolutions is basically a

type of interpolation;

 

- using the capital-M for pixels counts despite the fact that they

all mean multiples of 1,000,000, not 2^20 (remember from computers: 1

KB is 2^10 or 1,024 bytes, and 1 MB is 2^20 or 1,048,576 bytes, and

the lower-case k is used for 1,000 and m for 1,000,000).

 

I guess I must just be a crank. Or does anyone agree?

[francois_gauthier]

I would mention also advertising the zoom power including the useless 'digital zoom' wich is just a cropping function.

[User_1172872]

You haven't even mentioned claims about DMAX for film scanners.

[beauh44]

And don't get me started about an "80 Gig hard disk" that ya' only get 70-some gigs of space out of! ;-)

[paulr]

Its all marketing baloney.

 

For instance, in the hands of a person that knows how to print 3-4mp is enough. Lots of work, but one photographer Linda Butler used a 3.3MP point and shoot for some of the photographs in her book. She converted them to negatives so she could print then in her wet conventional darkroom on regular photographic paper.

Usually she uses 4x5.

 

If she did not say which photos in her book were done with that Sony digicam, I wouldn't have known.

 

 

<a href="http://www.lindabutlerphoto.com/">Linda Butler</a>

[astcell]

I don't get the "effective" megapixels part either. If you have ineffective megapixels, what are they doing there and why am I paying for them?

[mark u]

Much more important than the overblown marketing dependence on pixel count is the total lack of emphasis on sensor size, pixel well capacity, noise levels and fill factor and their consequences for dynamic range, higher ISO performance and depth of field control, and image quality (or even the ability to capture the desired image at all in the first place). The same could be said for claiming that a digicam with say an f/2.8 lens is the equal of an SLR or even a film P&S - when the reality is that there is no depth of field control, and the limitations of useable sensor ISO mean that reasonably fast shutter speeds in low light are completely unachieveable - at least without a very powerful slave flash (not supplied with the camera, of course). Fortunately the pixel count wars seem to be drawing to a close, though your obsession with it reflects the way the market developed.

 

Hopefully the more savvy consumers look for critical reviews that evaluate picture quality (there are big differences in in-camera image processing which affect even RAW images), and general useability, etc. rather than relying on one number to define where a camera should stand in the pecking order. After all, even an SD10 or 6MP DLSR will completely outclass an 8MP digicam in anything but the easiest of picture taking circumstances.

[godfrey]

Yes, I agree. You're a crank.

 

Godfrey

[nstock]

Its the ineffective megapixels that give you the better bokeh....

[kelly_flanigan1]

This is why one must place a larger weight on real performance; than BS marketing specs. A heck of alot of "stuff" is marketed with inflated specs. This has happened for thousands of years. Entire industries are built on marketing of "stuff" for folks to buy. Our lawn mower we got in 1952 was rated at 1.10 HP; and rated at 1 to 3 acres per day. This was before the Horsepower wars. This mower was rated as the real HP at the blades; after the selfpropelled HP was subtracted. It really had a motor of 2.5 HP on a test bench. <BR><BR>Long ago a 1.3 Megapixel slr was thousands of dollars; and a "pro digital camera"; a high end consumer camera was VGA; a starter 300 dollar camera a 240x320 pixel. Today one of my cellphones is a 1.2 Megapixel camera. The cellphone is vasty less in real quality and resolution than the old 1.3 Megapixel Nikon Fuji dslr camera; of eons ago. Both have the about the same pixel output. <BR><BR>If you are so ticked at normal digitals; you will be in heart attack mode if you test a dozen cellphone cameras; like I have recently been doing. Most all UPSIZE the image; well beyond the actual real sensor size. Some have sensor sites that are not even X-Y semimetrical. An object can be sharper if horizontal or vertical; even if the image is on the optical axis. <BR><BR>In vacuum cleaners; no earthly 120volt 15 amp outlet is going to put out 6 Hp; on a constant basis. The marketers use "peak HP"' where the inertia of the motors rotor goves these big numbers for a few seconds. <BR><BR>Notice in beers; very few have both the calories and alcoholic content on the container. In some states it is against the law too. <BR><BR>In hard disk drives for you "digital darkroom"; notice the actual media transfer rate is NEVER on the packaging. Only the interface spec is. This is like a YUGO and a Corvette both having a 140MPH speedometer; but the top speed or acceleration is never mentioned. Todays UDMA class 133 disk drives can have media transfer rates of 70 to 20 megs per second; all not known unless you test them. The UDMA133 spec is the interface; not the actual drive spec. <BR><BR>In inkjet cartridges; the trend is to not label the capacity in ml.<BR><BR>The marketers usage of 1024 for 1000; or vise versa goes back to magenetic tape; and drum memory. This is a 40 year old "trick"; and goes back before Ralph Naders Corvair book. <BR><BR>Picture tubes for your Philco TV in the 1950's were once round; and always abit smaller in diameter than the brochure. A 10 inch round TV had a viewable area smaller than 10 inches; when IKE was President.<BR><BR>In farming; some stuff is sold by the pound; some by volume. When to sell and how to measure your goods to get the best deal go back centuries.<BR><BR>The federal trade commission gets involved when alot of folks get confused; and real frauds are committed. The lesser camera brands have alway had more baloney claims in cameras; than the main stream brands. An Argus C3 and a Leica M3 are both good cameras; the Leica has a better lens. The asa and iso specs are used to label film speeds. Note Kodaks P3200 is not labeled an iso of 3200. It is really a iso 800 to 1200 film.<BR><BR>If all the Ralph Nader types create a huge committee to control all consumer goods specs; the prices must rise. All those fluffed up rolls of toilet paper will be sold by the pound.

[kelly_flanigan1]

If you were buying 12.3 acres of land; and only got 11.3; the surveyor who made the plat would legally be in a pickle. In cameras; there is no fraud; you are not buying megapixels; but a camera. What if the camera had the exact megapixel count; and had a bad warpage; so they were not all in focus? The megapixel war is like the "number of transistors in your pocket radio" race of the 1960's. What really matters is performance; not a marketers fancy labeling.

[dean_g]

What if you ignored all of it. Then checked out a camera that interested you and decided if you like the way the images look for the kind of photos you make? What difference do the numbers make? If you need the manufacturers marketing dept to make that decision, you're in trouble.

[dave_redmann]

Well it looks live I've touched a nerve (or several). Absolutely, I agree that pixels aren't close to everything. For overall quality, many factors mean that I'd much rather have a D30 than some 7 MP mini digital P&S. And I also realize that many specs on many things are borderline-BS if not outright BS.<P>

 

But actually, it is <I>because</I> of the mass-market emphasis on pixel counts that I think these frauds and borderline frauds are a problem. Hey, if they spouted BS on specs that only the really knowledgeable cared about, and the knowledgeable weren't fooled, who'd care? But, for better or (more likely) worse, many people buy on megapixels. And so to have so much BS about a stat so many relatively unknowledgeable people rely is a problem, IMHO.

[kelly_flanigan1]

Its like rating amplifiers in the 1960's in watts; where every maker used a different test! :)

[dave_redmann]

re: <I>"Its like rating amplifiers in the 1960's in watts; where every maker used a different test! :)"</I><P>

 

Kelly, exactly! And maybe a Watt meant a Watt only a 1 kHz, with only one channel driven, into 3 Ohms, at 10% total harmonic distortion, for 1 s. Then came the FTC (?) rule and all was relatively good (continuous output, 8 Ohm load, generally full-bandwidth, or at least specified bandwidth, etc.). But then something happened--maybe the rule did not apply to multi-channel systems?--and the BS in power amp ratings is making a comeback, especially with some of the surround receivers.<P>

 

Now I'm fairly libertarian and am not a fan of government regulation, but it is tempting to wish for a counterpart to the old FTC hi-fi power amp rating rule for digital camera specs.

[astcell]

It's like a 2x4 piece of wood is not 2" x 4" unless it is 80 years old. Now they are 1 7/8" x 3 1/2" and still called a 2x4.

 

WHY!?

[jim_gifford]

Robert, nowadays a 2x4 piece of lumber is two inches by four inches BEFORE it is planed for flatness on all four sides... you lose about one-eigth to one-quarter inch from each side in that process. That's why the 2x4 is only 3.5 inches wide.

 

The same logic applies to larger boards. A one-by-six isn't an inch thick and it isn't six inches across.

 

As for megapixels (back to Dave's question): It's not something that keeps me awake in the mddle of the night, but yes, manufacturers should be content to advertise the true number of pixels that contribute to images from a camera. Those who exaggerate are just being silly.

 

Regarding the difference between 1,000,000 pixels and 1,048,576: the prefix mega refers to a million of something. Pixel counts are accurately described in multiples of 10^6. It is data capacity that is INcorrectly described in 1,024 for K's and 1,048,576 for M's. Data capacity borrowed (and bent) the prefixes kilo- and mega- et cetera.

It does not matter whether the k or m is capitalized... kilo- means 1,000 of something (10^3) and mega- means a million (10^6). When you buy a kilogram of butter you do not get 1,024 grams.

 

Let's not get started on the trans-Atlantic differences in meaning of billion and trillion!

 

Be well,

[mark u]

As a small P.S. I actually think Sigma's description is the most honest - it makes clear exactly how many pixels of each colour are present, unlike any other manufacturer that I know of.

[jack paradise]

Dave,

If you think pixel count is bad, you should see what mfg's of film and flatbed scanners are claiming for specs. Thruth in advertising is a very abstract concept to them.

[basscheffers]

Some of the mentioned crimes are worse than others (although I have never seen Oly call it 5.5MP without mentioning 5 effective) but that is what reviews and your own tests are for when you decide to drop a few grand on a camera.

[scott_eaton]

The standard should be based on the actual out-put file size minus software based camer interpolation.

 

If a camera produces a 4mp native file, then it's a 4mp camera, not a 10mp camera. In this case I've considered the Foveon/Sigma issue to have undertones of false advertising. Fuji does the same.

[mark u]

Frankly, as resolution is the issue being addressed, I think it makes much more sense to look at lines resolved per image height or width under standardised test conditions. I think Phil Askey at DPreview has a far superior approach to any pure megapixel comparison, however doctored for different sensor configurations such Foveon's or Fuji's. The only major shortfall of his tests is that they relate to black and white resolution charts, rather than real world colours (e.g.blades of grass in a landscape?). But resolution is far from the only aspect of image quality that needs to be reviewed...

[bill_tuthill]

I say you should give the Fujifilm S3 credit for having 6 million

sensors to improve dynamic range and 6 million to record color.

Although I agree that Sigma and Foveon claims are exaggerated, their

3.4 million sensors are better per-pixel than Bayer-pattern sensors.

[e_m3]

I think you're a crank. But a nice one. :) The fact of the matter is the peformance of a file from a camera is effected by many variables, as many posters have pointed out. Is it really 'honest' to call a 6mp bayer camera 6mp when not all of those photosites are capturing the actual color, but rely on a CFA? You've got me. I don't have the answer. I know the ISO standards organization is wrestling with the question of how to define a pixel in relation to digital capture. It's not as easy as simply saying its a spatial location on one dimension, as single pixel capture (like Foveon or scanning backs) give very different per-pixel results than a CFA design.

 

Probably the best thing to do at this point is test cameras using a variety of color resolution charts. To use B&W charts doesn't seem terribly productive.

[astcell]

Well if we played the same game with film, then 35mm film could use the total sixe of the film, sprockets and all, and that area that does not get the light can be counted as "innefective silver hallide" but still count towards the grand total measurement of the film size.

 

Anyone want to start a class action lawsuit to sue for the extra pixels we paid for? Hey since 5% of the pixels are inefective, how about if 5% of my credit card payment becomes ineffective? :-)