Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Canon EOS dSLR

Reply
 
Thread Tools Search this Thread
Old May 1, 2003, 6:39 AM   #1
Junior Member
 
Join Date: Apr 2003
Posts: 5
Default Enough Pixels on my D30 to effectively print at 13x9"

Is 3 megapixels on my D30 enough for a 13x9 print? or do you need 6 megapixels?

Kind Regards,

Eric
etsmith is offline   Reply With Quote
Sponsored Links
Old May 1, 2003, 10:59 AM   #2
Member
 
Join Date: Dec 2002
Posts: 47
Default

A photo from a D30 will give you less than 115 pixels per inch when printed at 13x19. This is considered a low resolution image. It is better than most video monitors and will look fine when viewed from a distance of 3 feet or more, but fine details won't look sharp when viewed up close.
kenww is offline   Reply With Quote
Old May 1, 2003, 4:00 PM   #3
Senior Member
 
Join Date: Aug 2002
Posts: 610
Default

You need at least 300dpi for printing quality picture, take the maximum res of the D30 (probably 2048 x 1536) divide them by 300, in this case is about 8 x 6, so 8 x10 is probably the maximum size for quality print on the D30, preferably the 5 x 7 size.
tuanokc@hotmail.com is offline   Reply With Quote
Old May 2, 2003, 2:20 AM   #4
Junior Member
 
Join Date: Apr 2003
Posts: 5
Default

Thanks.
etsmith is offline   Reply With Quote
Old May 2, 2003, 5:41 PM   #5
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default Some corrections and misconceptions........

The simple answer is yes. Do you really mean 13x9 or 13x19?

One of the problems with getting help on the forums is that there are numerous helpful, but sometimes misinformed readers who are sincere in their answers, but haven't the experience or overall knowledge to give accurate and complete answers. I don't mean this in a malicious way, it's simply a problem with so much going on and a wide variety of experiences in the areas under consideration.

How large a print can be made with a given capture resolution depends almost entirely on the amount of actual geography in the frame. The D30 is routinely used to print beautiful 8x10's and "can" print beautiful billboards as well. By "billboards" I mean exactly that. Images measured in feet rather than inches. Some portrait photographers routinely print beautiful 60 to 100 inch (measured on the diagonal) portraits from the D30. I myself have printed 30 inch posters which, even when viewed from arms length, hold up beautifully. How? Let me explain.

Enlargement in the film world is done through optics. Enlargement beyond the optimal possible with a native file in the digital world is done through a process called interpolation.

Interpolation happens by a software process which examines adjacent output from the photodiodes in the sensor which sample the environment and create a number which, to make a rather complex process greatly simplified, represents values of light and color. By examining adjacent pixel information (numbers) intermediate values are "created" and added to the file. In this way the file size (number of pixels in the horizontal and vertical axis) grows and satisfies print requirements for larger prints.

To make this happen, there must have been sufficient sampling sites on the original capture so that the boundaries of fine detail in the image were properly defined. This can be done two ways. Larger numbers of photosites sampling a larger frame, or field of view, or smaller numbers of photosites sampling a smaller field of view. To make this easier to understand, think this way. A closeup or macro shot takes fewer pixels (lower resolution sensor) to properly sample than a far away landscape with great detail.

The 3.2 million pixels in the D30 can provide all the necessary capture resolution to capture a beautiful head and shoulders portrait. It can "not" properly sample a detailed landscape at infinity. Of course, neither can a Canon 1Ds or even a medium format film camera. It's all very relative. If there were "sufficient" sampling sites to properly define boundaries of fine detail, there is no practical limit to the size print which can successfully be made with the D30's capture. So, a head and shoulders portrait may be printed at even 100 inches from a D30 capture, as long as the proper interpolation software is used to process the file.

The simple answer to your original question is that the D30 can print about any capture at 8x10 and produce beautiful results. When you start printing larger, the results will vary according to the nature of the subject and the amount of actual geography in the frame. There is no hard and fast "rule," and only experience will teach you what can and can't be done. The same goes for the D60/10D the EOS-1D and EOS-1Ds. The more resolution the camera has, the more geography you can hold detail on.

The human brain does a lot of its own "interpolation." We are perfectly happy to accept a few brush strokes on an oil canvas as "pine needles or leaves" - at least until we examine it under strong magnification. Strong magnification reveals the "deception" and interpolation software acts like a magnifying glass for the file. When it finds accurate data, it reproduces it at a much larger size faithfully. When it finds "marker pixels" it reproduces it likewise and the results will be disappointing. "Marker pixels" are simply a few pixels in the proper location which when viewed from a distance or at a small enough print size "fool" the brain into "seeing" our expectation. Take a strong magnifying glass and examine the eyes of one of a group of people in a newspaper print. Do the eyes still look like eyes under strong magnification? Of course not. Now examine the eye of a head and shoulders portrait from a fairly large newspaper picture of a single individual. It still looks like an eye under strong magnification. Not a "pretty" eye, but still an eye.

That's the difference between adequate capture resolution and inadequate capture resolution. One could be greatly enlarged (the single head and shoulders portrait) and one would fail badly when interpolation revealed the "deception" of marker pixels.

Lin
Lin Evans is offline   Reply With Quote
Old May 5, 2003, 5:17 PM   #6
Senior Member
 
Join Date: Jun 2002
Posts: 143
Default

Lin is right on this. I would of course use a very good photo and preferably shot in raw. I have made 4x blowups of sections of photos and they look fine printed out. I prefer Genuine Fractals for interpolating up.
sasc is offline   Reply With Quote
Old May 7, 2003, 9:11 AM   #7
Member
 
Join Date: Dec 2002
Posts: 47
Default

When a digital picture is enlarged, the software reduces pixalation by adding false detail. This detail may be a smooth gradient of color where there may have been none in the original scene or, in the case of fractal techniques, the manufactered detail may be more sophisticated. For me, the appearence of the result is almost always more important than the pixel by pixe accuracy (if your camera has a Bayer sensor, you don't start that kind of accuracy anyway).

I'd say that the amount you can enlarge a given picture depends on the viewer's expectation of accurate, sharp detail, how close they are viewing the print, and how much detail there was in the image to start with.

As an example, a couple years ago a friend was giving a talk on iris, but did not have a slide projector available. I made some 12.5x17 prints from 1.3 megapixel images at about 80 pixels/inch. From the viewing distance of the audience, the prints looked great. Even up close most of them looked pretty good, because you don't necessarily expect to see finely detailed, sharp patterns on flowers. The exception was the edge of a white petal against a dark background. It was very pixelated.

The figure of 300 pixels/inch is intended to handle even the most demanding situations where you need to display fine, sharp detail to a discrimating person viewing the image at a distance of a few inches.
kenww is offline   Reply With Quote
Old May 7, 2003, 11:45 AM   #8
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default

Hi Ken,
A good interpolation algorithm doesn't "add" false detail, it faithfully reproduces what it finds by an examination of the numeric values of existing adjacent pixels. If the algorithm finds a numeric value of (for representative purposes) 247 at position "A" and position "B," it then creates another value of 247. On the other hand, if it finds value 247 at position "A" and 249 at positon "B," it produces an intermediate value of 248 with the logical assumption that in nature changes are more frequently gradual than abrupt.

Obviously, this is not always the case but it is more often than not. The primary problem with interpolation is that when it encounters "marker" pixels, which are pixels which simply delineate a position of detail rather than faithfully representing the detail itself, it reproduces exactly what it "sees" and the result is what you are interpreting as "false" detail.

At what is referred to as "normal viewing distance," and when these marker pixels are small enough, the human brain interprets them as the expected detail. When seen for what they really are, the deception is revealed and dissapointment in the enlargement results.

Let's look at some examples which show a similar situation. First image is a clip from a newspaper image with a few people. Second image we select one girl and observe her features. She appears to have recognizable features which most would say are eyes, nose, mouth, etc.




Now let's see what the interpolation algorithm actually "sees" and reproduces when an enlargement is made from an original captured with too few sampling sites to properly define boundaries of fine detail. Text continued below image....



As you can see from the above, in this case what we really have are "marker" pixels which represent little true detail. Had the original been captured at a much greater resolution, the enlargement would hold the detail because the interpolation algorithm would have faithfully reproduced detail rather than "false" detail.

The point then is that the interpolation doesn't "create" false detail, it "reveals" false detail.

Best regards,

Lin
Lin Evans is offline   Reply With Quote
Old May 31, 2003, 2:59 AM   #9
Junior Member
 
Join Date: May 2003
Posts: 14
Default Re: Enough Pixels on my D30 to effectively print at 13x9&quo

etsmith wrote:
Quote:
Is 3 megapixels on my D30 enough for a 13x9 print? or do you need 6 megapixels?
Hi,

With some simple calculations you can understand the exact output capacity of your camera.
Your D30 max resolution is 2160 x 1440 pixels.
One pixel translates to one dot.
So if your print requirements are 300 dots per inch(dpi) then you have:
(2160/300) x (1440/300) = 7.2 x 4.8 inch.

If you want a larger print you just interpolate the original file.

Here is a list of interpolation software:
http://pixtool.info


Theo
theodor is offline   Reply With Quote
Old Jun 1, 2003, 3:06 PM   #10
Senior Member
 
Join Date: May 2003
Posts: 577
Default

Interpolation attempts to create new pixels in between exisiting pixels. Interpolation can be used to zoom into an image, or make an image bgger. Interpolation will work if the digital image has sampled the geometry in the scene at a sufficiently high rate. That rate is also referred to as the Nyquist rate in the signal processing field. If your image indeed has been sampled above the Nyquist rate, then interpolation can create new pixels that are exactly correct. In theory interpolation can therefore enlarge an image to any size, and not lose detail.

In practice this is never possible, for various reasons. You cannot sample your geometry at a high enough rate, especially if there are sharp transitions (from black to white, for example). Lin's marker pixels are another good example of undersampling. Only an interpolation filter that uses all pixels available can create a new pixel that is exact. Typical filters used, for example a linear one, only use 4 pixels. Which is not nearly enough. (The reason linear filtering is so popular is because it smooths, it leaves out some high frequencies, and that is pleasing to the human eye).

Thus the answer to the original question is "it all depends'', as the other posters have already mentioned. Experiment!

Barthold
barthold is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 6:41 AM.