Gary's Parries 10/09/06

September 10, 2006 | Mark Goldstein | Gary's Parries | Comment |

Gary's ParriesThis week’s Gary’s Parries topics are:

1. What Goes Up, Must Come Down
2. Changing The Laws Of Physics
3. Venus And Mars Are Alright Tonight

Introducing this week’s Gary’s Parries column. Everything you always wanted to know about digital cameras, but were afraid to ask. No question too difficult, or too easy. As a Senior Principal Software Engineer, and a former Assistant Professor of Computer Information Systems, as well as a recording studio owner/operator, inventor, and now, a digital camera enthusiast, GARY has more digital camera knowledge in his entire brain than most people have in their little finger. In the unlikely event that GARY would not know the answer to your question, he will answer it anyway, true to the spirit of the word “Parries”, a fencing term which, in this context, implies “cleverly evasive answers”. So let your imagination run wild. Email all your nagging digital camera questions to: , and then, En Garde!

You may also attach to your email an ORIGINAL PHOTO of your choosing. A preview of the photo will be displayed with your question, and a full-sized version will be just a click away. No personal information will be published with your question unless you specifically include it in the text or attached photo of your email, which may be further edited for grammar, content, or other reasons.



Digital cameras get more and more pixels on the same sensor size. The latest 1/2.5” sensors have 7.1 million pixels, and yet they have higher ISO ratings than previous sensors with bigger pixels. Why? Shouldn’t smaller pixels mean more noise, and consequently a lower ISO rating?

Is it the newer sensors that have less noise, or is it that noise reduction can be more aggressive on larger files?

Anders Tiberg

*** ANSWER 1

Anders, so you’ve noticed that phenomenon as well. :)

To the less astute observer, it might appear as if CCD image noise were decreasing with decreased pixel size. But as we all know, that is physically impossible without at least some other CCD technological improvements. So what might those be?

The short answer is there really have been no CCD improvements that would account for this phenomenon. Keep in mind, we are not talking about any of the ‘specialty’ image sensors that have recently emerged, such as the Fujifilm Super CCD, the Foveon X3 CMOS Image Sensor, the Planet82 SMPD nanotechnology-based CMOS Image Sensor, as well as the more traditional Canon CMOS Image Sensor. We are talking only about ‘standard’ CCD technology, which really has not changed much in recent years.

Yes, there have been improvements in CCD fabrication techniques, the most significant of which was the addition of microlenses to the individual photosites, whereby the microlenses enable the capture of a higher percentage of the incoming light at each photosite, and then direct that light onto the ever-shrinking photosensor for that site. But microlens technology has pretty much peaked as to the percentage of light that can be captured, and yet, the maximum ISOs for the latest crop of higher megapixel cameras continue to climb.

So, Anders, how is it that ISO ratings continue to rise while pixel sizes decline? The answer has little to do with CCD improvements; rather, as you alluded to in your question, it has to do with improved, in-camera, noise reduction. As microprocessors used in digital cameras have become evermore powerful, so too have the in-camera algorithms used for image noise reduction, thereby enabling more aggressive noise processing, while at the same time preserving much of the image detail normally lost to such processing.

But that’s only half the story. Add to this the fact that camera manufacturers, in order to enhance the marketability of their product by claiming higher ISO ratings, are now willing to use even ‘more’ aggressive noise reduction (i.e., of the image blurring variety), and you have a pretty clear picture (no pun intended) of how ISOs go up, while pixel size comes down.


garys_parries_100906_q2.jpgHi Gary,

In the attached document, you will see that I have both raised a question and, to the best of my knowledge, answered it (well, sort of). However, this is a topic I feel will be of interest to others, so I would very much appreciate your viewpoints on the issue of sensor size in compact digicams.

Andy Martin

[The following is an abridged version of Andy Martin’s attached document. – Ed.]

With more and more pixels being crammed onto the tiny image sensors of compact digicams, why haven’t digital camera manufacturers tried putting a DSLR-sized sensor into one of those digicams? I suppose I could always buy an entry-level Canon or Nikon DSLR with a decent lens for around $1000, so why even make an issue of it?

Good point (talking to myself here), but for the kind of photography I do, I don’t want a large camera that draws attention to itself (or me). I have been taking photographs for nearly 35 years, and I have used almost every film camera imaginable. I have also done my own darkroom work, including archival printing, selenium toning, the whole deal. But these days, I don’t have time for all that. However, I still have an interest in producing large detailed prints that are as free as possible of purple fringing and the other digital artifacts characteristic of small sensors.

I’m not so much into ‘pixel peeping’ (i.e., blowing up portions of an image and whining about the defects) as I am into viewing A3-sized prints and cringing at the high levels of noise, purple fringing, and overall softness. Yes, I’ve done RAW files on, for example, a Panasonic Lumix DMC-LX1, boosted the sharpness, reduced the purple fringing, etc., but I would much prefer to deal with the relatively artifact-free detailed prints that can be obtained from a DSLR-sized sensor with its larger, less densely packed pixels.

As Michael Reichmann of Luminous Landscape notes:

“The 10MP sensor in the Panasonic FZ-50 is 7.18 X 5.32mm in size, while the 10MP sensor in the 10MP Nikon D80, for example, is 23 X 15.8mm. Do the math. The Nikon’s 10MP chip has 373 sq/mm of surface area while the one in the Panasonic has 38 sq/mm. This gives the Nikon’s sensor nearly 10X the recording area for the same number of pixels.”

“So what do we get as a result when a digicam tries to match megapixel count with a DSLR? Noise, or possibly worse – noise reduction artifacts.”

The above conclusion would seem to support my argument that inadequate sensor size, combined with the continuing ramping up of a sensor’s pixel count, has little to do with making any real attempt to improve the image quality of compact digicams.

Here are four major reasons why digital camera manufacturers HAVE not, and WILL not, put DSLR-sized image sensors into compact digicams.

(1) A larger image sensor would increase the camera’s manufacturing costs, which would then be passed on to consumers. Furthermore, such a camera would require considerable retooling for lenses to match the larger sensors, and would therefore cost as much, if not more, than an entry-level DSLR.

(2) These days, the trend is towards DSLRs aimed at pros, semi-pros, and advanced enthusiasts, with most digicams aimed at the less discerning snapper. Manufacturers that produce both DSLRs and digicams would likely prefer to maintain these two distinct product lines. If, however, manufacturers started producing ‘super sensor’ digicams, what would be the impact on their DSLR sales? Once DSLR users realized they could get close to the same image quality with a pocket-sized camera, who would want to tote around a huge camera with a lens the size of a siege gun? The result would be a drop in the sales of DSLRs and lenses.

(3) With the current state of technology, if a large sensor were mounted in a digicam, the resulting camera would have to be quite large. Case in point, the Sony DSC-R1. You would end up with a digicam the size of a DSLR, which defeats one of the main purposes of having a DSLR-quality digicam in the first place.

(4) There just is not sufficient consumer demand for a large-sensor, compact digicam. This stems from the fact that the average consumer is unaware that prints from a 10 megapixel digicam will have nowhere near the quality of prints from a 10 megapixel DSLR. Heck, most consumers, if they even make prints at all, will order only postcard-sized prints from their local DPE shop, and for the most part, they will be quite happy with the results.

Maybe someday sensor technology will advance to the stage where the image noise that results from cramming so many pixels onto such a tiny chip will drastically improve. I personally do not think that is possible. As Scotty, Chief Engineer of the Starship Enterprise would say, “Ye canna change the laws of physics.”

*** ANSWER 2

Andy, it is obvious that you put a great deal of thought into this question, and as a result, your four reasons as to why digital camera manufacturers HAVE not, WILL not, and (I would add) CAN not put a DSLR-sized sensor into a compact digicam, were all right on the money. However, having said that, there is no reason why camera manufacturers cannot squeeze current DSLR image sensor performance from a 1/2.5” chip. And I say this, Andy, despite Mr. Scott’s claim to the contrary. :)

CMOS image sensor technology is just one way such performance can be realized. Due to the ease of fabricating CMOS circuitry, it is not only possible, but also cost-effective, to add digital signal processing circuitry to image sensors at the pixel level, thereby providing orders of magnitude better noise characteristics and dynamic range than is currently possible with conventional 1/2.5” CCDs.

[Note: In all fairness to Chief Engineer Scott, the advantages of CMOS image sensor technology had not yet been realized at the time of his aforementioned claim. :)]



Here are some initial impressions of the Panasonic Lumix DMC-LX2.

Its Venus Engine III addressed the sky blotches (prevalent with the LX1) very well, but showed some blurring of faces. Overall, it’s a slight improvement over the LX1, and enables ‘cautious’ use of ISO 400.

Wonderful 2.8” widescreen LCD (Panasonic must have read our LX1 thread :)). Very easy to see compared to the LX1’s LCD, and its 16:9 aspect ratio is a perfect match for the camera’s 16:9 aspect ratio CCD.  Well designed and, overall, a step up from the LX1.

Semper Fi,
Nick and Family

*** ANSWER 3

Nick, thanks for the info, and for submitting the accompanying photos. I always like to see people submitting photos with their questions.

I see what you mean in the LX2 photo on the left about the effect of the Venus Engine III on those purple blotches in the sky. I can’t tell you how much they bothered me in those otherwise beautiful LX1 photos you took by the water.

I really like the narrow depth of field you achieved in the LX1 photo on the right (which appears to be self-explanatory :)).

[Column photo “The Photographer” by Brenda LaFleur of Brenda LaFleur Photography.]