[biz-engineer]

No, that's not a typo -- they use thirty dots (pixels) per inch.

One dot is like one bit. SQRT((30 x 30)/8bits) = 10 pixels per inch.

[Dartmoor Dave]

One dot is like one bit. SQRT((30 x 30)/8bits) = 10 pixels per inch.

Someone more knowledgeable than I am will need to explain the logic of that for me. My understanding is that if one dot was one bit then it would be either a solid black dot or a solid white dot. Binary 1 or 0. I'd assume that at least 16 bits are used to describe the colour of your average dot. (Edit: or maybe 8 bits if the billboard is printed from a jpeg? I dunno. . . but surely you need more than one bit to define any sort of a colour?)

Billboards are usually designed at 300dpi at a scale of 1/10th of the dimensions of the intended printing size, then printed at 30dpi to achieve the intended size in the resulting annoying advertisement.

Darn, I keep editing myself while I'm trying to figure out the logic of this. . .

Later. . .

Ah, I think I'm getting it now. You mean it takes at least 8 bits to describe each dot. The basic fact remains though. To quote myself: "Billboards are usually designed at 300dpi at a scale of 1/10th of the dimensions of the intended printing size, then printed at 30dpi to achieve the intended size in the resulting annoying advertisement."

[photoptimist]

Well yes, but the overwhelming majority of non-camera-obsessive photo-looking-at people don't respond primarily on the basis of how much detail they can see. If fine detail in huge prints was their main aesthetic concern then they wouldn't all be using smartphones.

The overwhelming majority of people respond to photos on the level of what it makes them feel. They respond to the subject of the photo, to the composition and the light, and even. . . dare I say it. . . to the meaning.

I've been in galleries with non-photographer friends, and I've observed them looking at huge and technically perfect prints of subjects and light that don't evoke any sense of an emotional response in them. They just shrug and move on along to the next thing on the wall. More often than not, the photograph that makes them stop and look for the longest time is a smaller print full of technical imperfections that grabs their eyes because it makes them feel something that has a meaning for them.

This is so true!

And it also highlights a huge difference between types of viewers (and types of photographers) and the role of detail in creating meaning.

There are many viewers (and photographers) who seek out pictures of people such as portraits, street scenes, and action sports. Such images don't depend on detail to create meaning because humans are so adept at recognizing human emotion, the mood of the scene, and the story behind the shot. Such images could be done with a pocket 110 camera or 6 MPix camera and still communicate the feelings.

However, there are many other viewers (and photographers) who strongly enjoy the wonder of nature, the grandeur of geography, the beauty of biology, and the magnificence of cities. Such images often benefit from intensive levels of detail -- the more the merrier!

I certainly pause for pictures of people and the fascinating stories latent in such images. But I can become far more engrossed in a high-detail landscape, cityscape, or macro shot because there is so much to see, explore, and notice. Like those "Where's Waldo" or Dorling Kindersley books, the more there is to see packed into the image, the better. Although pixel peeping gets a bad rap (and it is over-used), it's wonderful for leisurely arm-chair tourism of a high-detail scene. Today's (and tomorrow's) high-megapixel cameras let me take a quick shot of a canyon, forest, city, or lichen-covered rock and then get to luxuriate in the image and all the things I never noticed while at the actual place.

[Wheatfield]

Has anyone else seen these Photoshop ads touting some feature to squeeze more pixels into your image? No idea how that works and no interest in finding out, but it seems it's still something people fall for.

You are speaking out of ignorance, I'm afraid to say.

[biz-engineer]

Ah, I think I'm getting it now. You mean it takes at least 8 bits to describe each dot.

My reasoning might not be completely correct because tricks may be used to convert pixels to dots. One bit has a value 0 or 1. On paper for the smallest paper area increment, a dot of ink is dropped or not, so for simplicity I consider 1bit = 1 dot site on paper. I consider how many printer dots fit into a 1x1 inch square, say this is the total number of bits, divide by 8 (e.g. 1 primary color), that's the number of pixels information per square inch of paper. I take the square root of it to get back pixel per inch, assuming the vertical DPI and horizontal DPI of the print head are the same. For a Canon Pro printer 2400x1200 DPI, the printer can't print more than 600 PPI per color [ sqrt((2400x1200)/8) ], for Epson would be 700 PPI max, in all cases the number of PPI is much lower than the number of DPI.

[Dartmoor Dave]

My reasoning might not be completely correct because tricks may be used to convert pixels to dots. One bit has a value 0 or 1. On paper for the smallest paper area increment, a dot of ink is dropped or not, so for simplicity I consider 1bit = 1 dot site on paper. I consider how many printer dots fit into a 1x1 inch square, say this is the total number of bits, divide by 8 (e.g. 1 primary color), that's the number of pixels information per square inch of paper. I take the square root of it to get back pixel per inch, assuming the vertical DPI and horizontal DPI of the print head are the same. For a Canon Pro printer 2400x1200 DPI, the printer can't print more than 600 PPI per color [ sqrt((2400x1200)/8) ], for Epson would be 700 PPI max, in all cases the number of PPI is much lower than the number of DPI.

Thanks for that. I love it when things like this come up in a thread, because it makes me realise that there's a subject that I really don't understand properly. And now I can see that the actual workings of a printer on a dot-by dot basis is one of those things.

I've always blithely assumed that one pixel equals one dot, so that say a 6 megapixel camera with exactly 3000x2000 pixels printing at 300dpi would produce a 10" x 6.7" print. And I think I've always assumed (on the basis of no actual evidence) that the printer somehow mixes the primary coloured inks in the print head and then squirts out a dot of the exact colour required. But like I say, that's just been an ignorant assumption of mine that I've never bothered to actually check.

Now I'm thinking -- based on what you're saying here and hoping I've understood you right -- that perhaps a printer actually squirts out dots of primary colours only. So that you need one dot each of Red, Green and Blue from the printer to additively get the colour of one pixel from the camera. Is that the basis of your calculation that 30 dots-per-inch is actually 10 pixels-per-inch?

I'm happy to confess that I'm utterly confused, which means that this a chance to find out about something that I've never investigated in any detail. Have we got an expert here who can explain -- on the basis of actual knowledge rather than internet mythology -- how exactly a printer makes a colour print on a dot-by-dot basis?

(Ironically, my favourites of the prints that I've got on my own wall were actually made by a lab that digitally projects the image onto old school photographic paper then develops it in chemicals. Much cheaper, and to my eyes much nicer, than the fancy his-res inkjet printing service that they also offer. Typical of me to prefer the simplest and cheapest technology available.)

[biz-engineer]

I'm happy to confess that I'm utterly confused, which means that this a chance to find out about something that I've never investigated in any detail.

That's even more confusing that not all display media uses dots.
- A basic HD (1920 x 1080) LCD display RGB made of 3 (or more) primary color dots per pixel, which appears finer grained than the display resolution would let us to believe
- Inkjet printers use clouds of tiny dots (2400dpi/2880dpi) to create continuous tones
- C-type printers use modulated tri-chromatic laser light to create continuous tones
We had a similar thread about this topic a year earlier, and other members of this forums are more knowledgeable than me about that topic.

[photoptimist]

Now I'm thinking -- based on what you're saying here and hoping I've understood you right -- that perhaps a printer actually squirts out dots of primary colours only. So that you need one dot each of Red, Green and Blue from the printer to additively get the colour of one pixel from the camera. Is that the basis of your calculation that 30 dots-per-inch is actually 10 pixels-per-inch?

Yes, the printer can only make dots of full color. Pale colors (e.g., gray, pink, light green, etc.) require leaving white space between the dark dots. (if you look carefully at an inkjet print with a jewelers loupe or microscope, it looks like a pointillist painting.)

As for the colors, the inks are actually known as subtractive colors: cyan, yellow, and magenta. Cyan dots absorb red light. Yellow dots absorb blue light. Magenta dots absorb green light. Thus, for example, a cyan dot with a yellow dot on top (subtracting both red and blue light) creates a saturated green dot. In theory, a cyan+yellow+magenta dot would be black but limitations of the inks mean it's not a very deep black, plus it uses a lot of expensive color ink, plus it makes the paper very soggy. Thus, color printers also include black ink. The resulting color system is known as CYMK, where the K is the black "Key" ink and avoids confusion with using "B" for black which could be confused for blue.

Getting non-saturated colors (e.g., pale green) means putting down a cyan dot, a yellow dot, and then leaving some space of white paper. Getting full color with 8-bits per color means dividing the page into 8-dot by 8-dot squares such that one square (one pixel) can any number of dots from 0 to 256 of each of the inks to create anything from 100% white to 100% of any color including black. That's why printers talk about DPI (dots per inch) instead of PPI (pixels per inch) and printing full 24-bit RGB color at 300 PPI requires a printer with 2400 DPI of CYMK dots.

Modern-day high-end inkjet printers are bit more complex. Some can create dots of a couple of different sizes and include added inks (lighter versions of C, Y, M, and sometimes K). The result is that maybe a 4-dot by 4-dot square can produce the full gamut of colors and shades so that a 1200 DPI printer can create good 300 PPI prints.

[texandrews]

Yes, the printer can only make dots of full color. Pale colors (e.g., gray, pink, light green, etc.) require leaving white space between the dark dots. (if you look carefully at an inkjet print with a jewelers loupe or microscope, it looks like a pointillist painting.)

As for the colors, the inks are actually known as subtractive colors: cyan, yellow, and magenta. Cyan dots absorb red light. Yellow dots absorb blue light. Magenta dots absorb green light. Thus, for example, a cyan dot with a yellow dot on top (subtracting both red and blue light) creates a saturated green dot. In theory, a cyan+yellow+magenta dot would be black but limitations of the inks mean it's not a very deep black, plus it uses a lot of expensive color ink, plus it makes the paper very soggy. Thus, color printers also include black ink. The resulting color system is known as CYMK, where the K is the black "Key" ink and avoids confusion with using "B" for black which could be confused for blue.

Getting non-saturated colors (e.g., pale green) means putting down a cyan dot, a yellow dot, and then leaving some space of white paper. Getting full color with 8-bits per color means dividing the page into 8-dot by 8-dot squares such that one square (one pixel) can any number of dots from 0 to 256 of each of the inks to create anything from 100% white to 100% of any color including black. That's why printers talk about DPI (dots per inch) instead of PPI (pixels per inch) and printing full 24-bit RGB color at 300 PPI requires a printer with 2400 DPI of CYMK dots.

Modern-day high-end inkjet printers are bit more complex. Some can create dots of a couple of different sizes and include added inks (lighter versions of C, Y, M, and sometimes K). The result is that maybe a 4-dot by 4-dot square can produce the full gamut of colors and shades so that a 1200 DPI printer can create good 300 PPI prints.

So, everyone mark and heed this post. It's the clearest and most succinct explanation of the relationship between PPI and DPI that I have ever seen, And I've been reading these descriptions for at least a decade.

[SOS7]

36mp this year as that what I have in the K1.

[Wheatfield]

Yes, the printer can only make dots of full color. Pale colors (e.g., gray, pink, light green, etc.) require leaving white space between the dark dots. (if you look carefully at an inkjet print with a jewelers loupe or microscope, it looks like a pointillist painting.)

As for the colors, the inks are actually known as subtractive colors: cyan, yellow, and magenta. Cyan dots absorb red light. Yellow dots absorb blue light. Magenta dots absorb green light. Thus, for example, a cyan dot with a yellow dot on top (subtracting both red and blue light) creates a saturated green dot. In theory, a cyan+yellow+magenta dot would be black but limitations of the inks mean it's not a very deep black, plus it uses a lot of expensive color ink, plus it makes the paper very soggy. Thus, color printers also include black ink. The resulting color system is known as CYMK, where the K is the black "Key" ink and avoids confusion with using "B" for black which could be confused for blue.

Getting non-saturated colors (e.g., pale green) means putting down a cyan dot, a yellow dot, and then leaving some space of white paper. Getting full color with 8-bits per color means dividing the page into 8-dot by 8-dot squares such that one square (one pixel) can any number of dots from 0 to 256 of each of the inks to create anything from 100% white to 100% of any color including black. That's why printers talk about DPI (dots per inch) instead of PPI (pixels per inch) and printing full 24-bit RGB color at 300 PPI requires a printer with 2400 DPI of CYMK dots.

Modern-day high-end inkjet printers are bit more complex. Some can create dots of a couple of different sizes and include added inks (lighter versions of C, Y, M, and sometimes K). The result is that maybe a 4-dot by 4-dot square can produce the full gamut of colors and shades so that a 1200 DPI printer can create good 300 PPI prints.

This would be how Epson does it. My 8 tank Epson has light yellow, magenta and cyan tanks as well as what they call light black (gray).

Correction: It has one yellow tank and 3 black tanks. They call them black, light black, and light light black.

[Dartmoor Dave]

Yes, the printer can only make dots of full color. Pale colors (e.g., gray, pink, light green, etc.) require leaving white space between the dark dots. (if you look carefully at an inkjet print with a jewelers loupe or microscope, it looks like a pointillist painting.)

As for the colors, the inks are actually known as subtractive colors: cyan, yellow, and magenta. Cyan dots absorb red light. Yellow dots absorb blue light. Magenta dots absorb green light. Thus, for example, a cyan dot with a yellow dot on top (subtracting both red and blue light) creates a saturated green dot. In theory, a cyan+yellow+magenta dot would be black but limitations of the inks mean it's not a very deep black, plus it uses a lot of expensive color ink, plus it makes the paper very soggy. Thus, color printers also include black ink. The resulting color system is known as CYMK, where the K is the black "Key" ink and avoids confusion with using "B" for black which could be confused for blue.

Getting non-saturated colors (e.g., pale green) means putting down a cyan dot, a yellow dot, and then leaving some space of white paper. Getting full color with 8-bits per color means dividing the page into 8-dot by 8-dot squares such that one square (one pixel) can any number of dots from 0 to 256 of each of the inks to create anything from 100% white to 100% of any color including black. That's why printers talk about DPI (dots per inch) instead of PPI (pixels per inch) and printing full 24-bit RGB color at 300 PPI requires a printer with 2400 DPI of CYMK dots.

Modern-day high-end inkjet printers are bit more complex. Some can create dots of a couple of different sizes and include added inks (lighter versions of C, Y, M, and sometimes K). The result is that maybe a 4-dot by 4-dot square can produce the full gamut of colors and shades so that a 1200 DPI printer can create good 300 PPI prints.

Huge thanks for that explanation. I think it makes sense to me now, thanks to your usual skill at getting ideas across with a concise clarity that makes things understandable even for non-technically-minded folk like me.

[biz-engineer]

Yes, the printer can only make dots of full color. .... The result is that maybe a 4-dot by 4-dot square can produce the full gamut of colors and shades so that a 1200 DPI printer can create good 300 PPI prints.

Next time I want to express/explain something, before doing so, I might consult you for suggestions about how to do it the right way

[125px]

How many megapixels do you need?

Sounds like the old boring rant.

This topic has been debated Ad nauseam, yet most people get it wrong.

Please world wide web, stop spreading wrong information.

The megapixels needed has nothing to do with print size (or screen) DPI or PPI.

The ultimate secret that no one ever told is that the megapixels needed is entirely relative to how much detail worth looking by human being, is present in an image, i.e how many pixels (image samples) are used to describe an image detail.

This secret trick has been largely used, and is still being used by David Yarrow, who produced gallery prints the size of pools table, of images captured with a D810 and D850. In all his images, the subject occupy a large part of the image frame, such that even printed 10 feet wide, the subject details still look well defined to the human eye.

Image quality is all important. I will stick with K3-III and film scans 35mm on SP/x and LX/MX/KX/1000 and whatever Tukumar/K/M lens. :-)

[weverka]

I like to be able to look up close as well as take in the whole picture. The more detail the better.
For example, today I took a picture of a cherry tree. It is inviting as a whole as well as inviting to come in close to see the cherries. Yummy.