Sony Zone


Right and Wrong: A New Perspective

Digital Photo Myth #3: Pixels Are Perfect.

Digital Photo Myth #4: Bigger = Gooder.

By Peter iNova  
Design and photography by Peter iNova

You purchased an 8 megapixel camera, and you want eight million details in your shot. It's only natural. And at about $1000.00 for the camera, each pixel costs you a very expensive $0.000125 each. Eight thousand to the dollar. Hey that's not too bad. Last year's camera only brought you five thousand pixels to the dollar...

With every pixel comes a hidden cost. Saving that pixel on the camera's storage medium is cheap. Camera media is recycled.

Saving the pixel on your computer's hard drive is relatively cheap, but at eight million details per picture, even that multi-gig wonder will dry up some day, or you'll have to find room inside your computer for another and another and another.

Terabyte anyone? Drives that big ARE available and they'll set you back about as much as your camera cost.

You laptop's 60 gig drive is going to fill up pretty quickly. Along with all the other files in your life, so anything you can do to save file space means more pictures in your local quick-access archives.

Most people think that means choosing the option to increase compression and lower file sizes, and while this is a worthy strategy, the nagging Fear is that you will be ruining your images by using Standard compression instead of that far more virtuous Fine compression.

So the first thing to do is see exactly and unambiguously what the difference is between these two compression settings.

Test your own perceptions.

Roll your mouse over the image above and see Fine versus Standard compression in action. It's a 100% scale crop out of a pair of shots made back to back. A and B; but which of them is the Fine and which is the Standard? There are differences in the two shots, but are they simply the differences one might expect from any two images or do they have something to do with the difference of compression?

Compression artifacts tend to echo into the adjacent areas near contrasty detail. Like frozen waves near beaches of sharp transitions, they follow contours and introduce false details into the image. Can't you see that happening in either A or B above?

Keep in mind that your computer screen is probably displaying an image with about 100 pixels per inch, and at that size this image would make a print over two feet wide. Or for you more advanced, metric folk, it would be around 2/3 of a meter wide. Shrink the image above down to about 33% of the size you see here (where it would occupy 1/9th of the area seen above) and you can imagine how much detail would be in every square inch (or centimeter) of an 8x10 print.

You can do this sort of test for yourself, too, with any Photoshop that permits layers. Shoot from a tripod and layer your different images, then blink the upper layer off and on with its visibility icon (eyeball).

To get a real sense of what compression choices might force a picture to gain or lose, print a Fine and Standard full frame pair of exposures onto your image size of choice. 8x10 or 11x14--something like that--and pass the pair around to coworkers and image enthusiasts for critical review. Can they see the difference in quality between F and S?

The two shots seen here represent a 55.3:100 ratio of storage requirement. The more compressed image takes up only 55.3% of the Memory Stick that the Fine file requires.

See? Isn't it obvious which one requires all that extra room?

(Answer at bottom, but don't look yet.)

Now let's get more controversial.

How big is "Good?"

It's one of those questions that if you heard it from a five year old, you'd laugh and engage the kid in a dialog to find out more. What on earth is he thinking about?

But from digital photographers, the answer you hear to that same sentiment is this: "Full frames : Good / Smaller frames : Bad!"

Say that with a Frankenstein accent and it becomes a parody of itself.

Are full size files Gooder than smaller, Badder files? Intuition snaps your answer into the Yes column faster than you can smile upon learning you just earned a tax refund.

"Of COURSE bigger files are more Good, you nerd, what kind of question is THAT? Geez. What a moron."

Certainly the Bigger = Gooder formula is at the heart of digital photography. A digital image with 75 x 100 pixels isn't as good to look at as one with 300 x 400 pixels. Study this pair until that idea sinks in:

Rock Diva Ninette of the group, Powder. DSC-F717 in Burst 3 mode under stage lighting.

We each have our own preferences in life, but for my money, I get a lot more communication out of the diva on the right. In this case the larger pixel count = Gooder. But is it always so cut and dried?

Nearly every model of Sony digital camera gives you the choice to shoot pictures at full size and several smaller sizes. Cleverly, Sony has included a next-to-full-size file option that stores images at about 75-80% of full size.

On the 707/717, full size is 2560 pixels wide, and the next size down is 2048 pixels wide. On the 828, full size is a whopping 3264 pixels wide, and the next smaller file is 2592 pixels in width.

NOW which one is Gooder? The answer to that may shock and amaze you.

In fact, I predict that you will eventually slap yourself in the forehead so many times and with such violent force that you could do some real damage to your brain, so perhaps you could practice other methods of registering miraculous surprise. Shouting "Yikes!" is safer.

Here's the drill: These next-smaller files occupy less memory space. No mystery there. Typically they take up around 60% of the room a full frame consumes with every shot.

That means you can shoot around 67% more images per unit of storage space if you were to use the smaller frames. Not a small saving at all. If it were gas mileage, you'd jump for joy. If it were votes, it would declare a landslide. If it were a raise, you'd pass out from hyperventilating.

Here are two images. One is a crop from the center of a full frame image made with the 717. The other is a crop made from the same locked-down camera but it is shot at 2048 pixels wide, the camera's 3M setting (to use the current shorthand vernacular).

Which is which? More importantly; Which is Gooder?

Your whole value system is at stake, here. Because as every five-year-old knows, Good = Right and Bad = Wrong.

Shooting conditions: DSC-F717 at mid zoom, 1/250 sec @ f/8. ISO 100. Partly cloudy.

Oh, did I forget to tell you that the smaller frame has been enlarged in Photoshop CS to the exact dimensions of the full frame? Bicubic interpolation was the driving engine, but no extra tweaks or sharpening were added during enlargement.

Look carefully at the two shots. Try to see things like fine detail, noise, artifacts of compression, artifacts of digital sharpening and stuff like that. Take your time and be very picky.

Perhaps it would be easier if I put the two images on top of each other so you could use the mouseover technique to see each image in direct comparison to the other. Here:

By now you have formed an opinion. Either one image looks a whole lot Gooder to you than the other or you have lost your ability to tell Right from Wrong.

Don't feel bad, prisons are full of people who can't tell Right from Wrong, and in this case the prison may be the mental conclusions you've latched onto like the knee-jerk reaction to the Bigger Is Better question.

How come you can't tell which is which? (Perhaps I assume too much here. Perhaps it's only moi who can't see the absolute differences in these two shots. Perhaps it will rain beer.)

When your Sony camera shoots any picture except the movie modes, it gathers every pixel's detail from every photosite on the image chip.

That chip is covered with a Bayer Pattern Color Filter Array that allows full color images to be inhaled by a single graphic surface.

Compromises happen, but not as strongly as you might think. Two green and one each red and blue filtered sensors make up the repeating pattern. The 828 camera uniquely uses a near-green, "Emerald" sensor that catches a bit more of the blue-green section of the spectrum.

Detail is a product of luminance and contrast. Without differences in contrast from pixel to pixel, there is no detail. Here's a detail-free image to illustrate that point:


No doubt of it. This is one of the most thrilling digital images around. Clearly a panoramic telephoto picture of a section of sky during a dramatic sunset. Ahh.

Wait a minute. There's something in my perfect shot! Dang. It's one of those stray orange photons. How did it get in there? At least it's well composed.

Without the small difference in contrast between the color background and the tiny photonic spot, you would not be able to see any detail.

But with a color filter array that requires the coordination of four color filtered areas to make up a perfectly detailed spot of color, you might think that detail has been compromised down to half of its potential. Thanks to computer image processing, that is not the actual case.

Your eyes, and digital images, see luminance images with color overlays. Green contributes 60% to your appreciation of luminance (red and blue contribute 29% and 11%, respectively) so those two green positions in the color filter array are chess-boarded. It's not a big stretch for the camera's computer to predict the appropriate values in the other adjacent squares, and it does just that. It reconstructs most of what might have been lost and gives an image about 80% as detailed as a pure monochrome, unfiltered image chip would gather.

80%? Aha! That happens to be about the size of the next-to-full-size frame you can select as your file option. For an 828 camera, that's the 5M image. For a 717/V1, that's a 3M image. As you shoot these smaller frames, the camera gathers everything it can and reduces the image with an intelligent shrinking operation.

In doing so, it crams 25 pixels into 16 newly rendered pixels that attempt to hang onto as much detail and color as is possible. And since the original 25 pixels were a little on the soft side to begin with, the newly computed cluster of 16 is sharper on a per-pixel basis.

It's called down-sampling and it really works. Here's that very rough shot of the diva, above, presented at 25% size. Notice how it doesn't look so pixel-compromised?

Open it in its own new page, and you'll see that it's the exact same file displayed above, only shown here instantly down-sampled by your Internet browser to 75 x 100 pixels.

Blowing up smaller images in Photoshop re-computes each group of 16 pixels into a new group of 25 while attempting to not lose detail. Some loss is inevitable, but in a practical sense, very little is actually completely lost in each process.

As you can see from the house example pictures, what I mean to convey with the words "very little" is not an exaggeration. I take "very little" to mean the loss of pictorial information at 100% scale is on the order of 1%. At any reasonable print scale, it is so far below the threshold of perception as to be totally invisible.

Bottom lines:

Shooting with the next-to-full-size frame is a goodness. It keeps so much of the original frame's detail that neither you nor I can tell the difference in a print, and even viewing the files directly on our computer screens won't reliably show artifacts, flaws or loss of detail.

As you shoot with reduced files, you can gather 100 pictures in the same amount of memory card that used to hold only 60 shots before.

Shooting with greater compression, using Standard instead of Fine, has a similar effect on your storage requirements. As it turns out, Standard shots take up about 55% of the space occupied by Fine shots. Now your memory stick is 80% larger than you thought.

But what happens when you shoot full frames at Fine and compare them directly to reduced frames at Standard? Here's that test, too:

Roll your mouse over this image and tell me you can see which one was gathered at about 36% of the storage space of the other.

Now sit down with a pencil and paper and write me a 1200 word essay on how Very Much you need to shoot full size, Fine compressed images instead of next-size-smaller, Standard compressed images. Points will be awarded for making sense only. Literary style, humor and sarcasm will not get you an A.

The Goodest thing about these two concepts, compression and file size, is that you don't have to take anything presented here at face value. You can shoot pictures in your own camera that show you the same things. You can print out your own shots and ask people to decide which is which.

Here's the same shot scaled down to simulate the effect of printing it on a letter page:

(Actually, detail on the page would be greater than seen here because the printer would be capable of over 700 ink spots per inch.)

Is it worth giving up a tiny percent of image quality just so you can shoot over 250% of the shots you would otherwise have gathered?

Then again, perhaps you have terabytes of storage space in your computer and you simply don't need to conserve space at all.

The choice, as always, is yours.

ANSWERS: Which is which. In every case, the full-frame, Fine compressed image is A. B is the greater compressed, Standard version that takes up 55.3% of the file space of the Full/Fine image, and it is the 3M/Fine image that occupies 60% of A's space.

In the last example, B is the 3M/Standard compressed image that takes up just 36.1% of the memory required to hold A.

More will be here as the Sony Zone pages evolve.


PS: As experience grows with these cameras, so will this review and the Sony eBook. Available now on the order page.

© 2004 Peter iNova. All rights reserved. Do not replicate or link to images without permission. All photos by Peter iNova unless attributed to others. Photos are the copyright of their originators.