Pixels -vs- Dpi?

I'm a little confused. Is the mega pixel number just a function of size of the image created? In other words are there always the same number of dots per inch and more mega pixels will just make a larger image or is there in fact better resolution and finer detail with a higher mp camera? If mostly printing 4x6 or 5x7 can you tell a difference between a 2 and 3mp camera or does it only matter when going larger than 8x10?

Thanks,
Peter--www.pbase.com/optionguru

Megapixel cound that is quoted comes from the product of the height and width in pixels that you shoot the image at. So if you shoot at 2048 x 1536, the megapixel count is 2048 x 1536 = 3145728 or 3.15 megapixels.

dot per inch is usually used within the context of how you will present the pixels on screen or in print. Higher dpi will mean a smaller overall image size with the same number of megapixels.

2mp and 3mp will not normally reveal any difference at 4x6 or 5x7.

There will be a slight difference at 8x10 (depends also on the subject matter and difference in quality of lens used on the 2mp and 3mp camera), and the difference in quality will become larger as you go to larger print sizes, until the 3mp maxes out. According to Nikon, 2048 x 1536 at FINE quality will produce acceptable 11x14 prints.

Jason S Tay

Peter B. Nightingale said:

I'm a little confused. Is the mega pixel number just a function of
size of the image created? In other words are there always the
same number of dots per inch and more mega pixels will just make a
larger image or is there in fact better resolution and finer detail
with a higher mp camera? If mostly printing 4x6 or 5x7 can you
tell a difference between a 2 and 3mp camera or does it only matter
when going larger than 8x10?

Thanks,
Peter
--
http://www.pbase.com/optionguru

Click to expand...

I think I had the same question...just to make sure let me rephrase it: So if I take a 3MP camera even though it has more MP than a 2MP camera it doesn't take better 4x6/5X7 pictures?

I was holding out for a 4MP camera because I thought that since it had more MP it would look better...but if that's not the case I'll go out an buy a 3MP today!!!

Thanks

Jason S Tay said:

dot per inch is usually used within the context of how you will
present the pixels on screen or in print. Higher dpi will mean a
smaller overall image size with the same number of megapixels.

2mp and 3mp will not normally reveal any difference at 4x6 or 5x7.
There will be a slight difference at 8x10 (depends also on the
subject matter and difference in quality of lens used on the 2mp
and 3mp camera), and the difference in quality will become larger
as you go to larger print sizes, until the 3mp maxes out. According
to Nikon, 2048 x 1536 at FINE quality will produce acceptable 11x14
prints.

Jason S Tay

Peter B. Nightingale said:

I'm a little confused. Is the mega pixel number just a function of
size of the image created? In other words are there always the
same number of dots per inch and more mega pixels will just make a
larger image or is there in fact better resolution and finer detail
with a higher mp camera? If mostly printing 4x6 or 5x7 can you
tell a difference between a 2 and 3mp camera or does it only matter
when going larger than 8x10?

Thanks,
Peter
--
http://www.pbase.com/optionguru

Click to expand...

Click to expand...

Jimi said:

I think I had the same question...just to make sure let me rephrase
it: So if I take a 3MP camera even though it has more MP than a
2MP camera it doesn't take better 4x6/5X7 pictures?

I was holding out for a 4MP camera because I thought that since it
had more MP it would look better...but if that's not the case I'll
go out an buy a 3MP today!!!

Click to expand...

Hi Jimi,

There is more to get good pics: The most excellent pics had been made with the Canon EOS D30 with 3.11 million effective pixels. Smoothest and sharpest pictures without noise. As a rule of thumb I would say that you can't go wrong with any camera on te market with > 3 Mpixels.

You can crop with more pixels and still be at the max resolution for the common printers but don't overestimate the number of pixels. After few years that number will be higher.

If the number of pixels was the only thing wich held you back till now: Run to the shop then....
Hans.

Jimi said:

I think I had the same question...just to make sure let me rephrase
it: So if I take a 3MP camera even though it has more MP than a
2MP camera it doesn't take better 4x6/5X7 pictures?

Click to expand...

Let's do a little calculating to find out.

I'll use my CP990 as a reference 3mp camera it takes images with maximum pixel dimensions of 2048 x 1536 pixels.

Now if we were to print this image out on a printer that has an output resolution of 300dpi (Most photo laser printers have at least 300 dpi res. for example the Fuji Frontier Laser Photo used by some of the best online photo finishers.) we would map each pixel to each dot in the output media, mathematically we want the number of pixels that maps to a real world inch so we do a straight division of the pixel dimension by the dpi:

2048 300 = 6.82 inches
1536/300 = 5.12 inches

So a CP990 image if printed out at 300dpi will yield a physical print with dimensions of 6.82 x 5.12 inches without scaling. It covers a 4 x 6 with room to spare, in fact the image would need to be cropped or scaled down further to fit on a 4 x 6 print. Let's see how a 2mp camera fairs.

Let's use the 2mp CP900, with pixel dimensions of 1600 x 1200:

1600/300 = 5.3 inches
1200/300 = 4 inches

So its revealed that a 2mp image without scaling is printed at 300dpi at 4 x 5.3 inches which is less than the standard 4 x 6 print size. Most photofinishers automatically scale up the image so that there is no border unless you ask for a print "to fit". If it were printed at 5 x 7 there would be even more scaling, remember that scaling increases the size of the pixels on the output media which results in blurring of detail and increased pixelization. The 3mp image in comparison has its native print size very close to 5 x 7 : 6.82 x 5.12 as calculated previously, the scaling that would be performed to have it fill the print edge to edge would be minimal compared to the 2mp camera which was already being scaled at 4 x 6 size. You can see why more MP translates to a better image, since more pixels are available to map to real world inches under the output mediums print dpi rating.

300 dpi is considered the standard high resolution print dpi by many graphic designers, most designs that are meant for high res. display are proofed at that dpi or highier. We can then look at the resolution of our camera in terms of the minimum dpi we would like to see at a particular print size.

For example, if we wanted an 8 x 10 print edge to edge at 300dpi we would need a camera with:

8 x 300 = 2400 pixels
10 x 300 = 3000 pixels

The just announced Canon D60 has a max. resolution of 3072 x 2048 thus even its images would have to be scaled/cropped to fit an 8 x 10 print edge to edge. It's dpi along the long side is:

3072/10 = 307.2 dpi

along the short side:

2048/8 = 256 dpi

You'd crop along the long side and scale up along the short side to fit the page. Since the pixels are square we can find the normalized print resolution at the scaled size by:

307.2 - 256 = 51.2/2 = 25.6 dpi normalized.

So the final print will have 25.6 dpi cropped off along the long side and 25.6 dpi added (scaled) to the short side. Final effective print resolution of:

281.6dpi

very close but not quite maxing out the print mediums 300dpi capability since the short side only had 256dpi of res. to begin with.

Since the long side is cropped we can find the number of pixels trimmed to correspond to the scaling done on the short side.

10 x 281.6 = 2816 pixels along 8 inch side. (crop of 3072 - 2816= 256pixels)

I've printed out D1x images (3008 x 1960) at 8 x 10 without scaling, because there is more than enough pixels to achieve the 300dpi print output on the long side(because its available dpi is 300.8 but since the printer can only output 300, the extra .8 can't show up in the print), side bars are visible on the print since the number of short side pixels is insufficient to cover the full 8 inches at 300dpi. It's easy to figure out how big those side bars are:

1960/300 = 6.53 > > 8 - 6.53 = 1.46/2 = .73 inches

I just measured the print and sure enough, the side bars are .73 inches. The D60 and the newly announced D100 (3008 x 2000) would reduce the side bars for an 8 x 10 "to fit" print since they both have more pixels along the short side.

D100:
2000/300 = 6.66 > > 8 - 6.66 = 1.33/2 = .66 inches

D60:
2048/300 = 6.82 > > 8 - 6.82 = 1.17/2 = .58 inches

As illustrated earlier if they had short sides of 2400 pixels there wouldn't be any scaling, it would be a perfect mapping to 8 inches 2400/300dpi.

I think that my hunger for increasing megapixels will end once I am able to print an 11 x 14 inch print at 300 dpi. We can use the same principles applied above to find out the minimum sensor dimensions needed for this.

11 x 300 = 3300
14 x 300 = 4500

3300 x 4500 pixels or about 15 MP. (14,850,000) I figure that's about 2-3 years away. If the sensor dimensions are larger we can crop away the excess and still print at the output mediums max. assuming its 300dpi.

Jimi said:

I was holding out for a 4MP camera because I thought that since it
had more MP it would look better...but if that's not the case I'll
go out an buy a 3MP today!!!

Click to expand...

As illustrated above it may look better depending on the resolution of the output medium. For Fuji Frontier prints its 300dpi, but that would vary depending on what you use, for example the numbers on a home inkjet would be different. The lower the output resolution the larger size print you can get before seeing a difference in quality between the two sizes. However, given the same output dpi you can always calculate the print size and effective print resolution of an image of a given MP size. For 3 and 4 mp cameras, 4 x 6 and 5 x 7 will roughly look the same but 8 x 10 and highier will look better for a 4mp camera than a 3mp at 300dpi.

Hope that helps,

--DSL

Wow,

I think you should get an award for putting the most time into a response. I must say that this site is great, I ask a question and now I have the answers.

Just like my saltwater fish tanks, half of the fun in a good hobby is the learning.

Thanks,
Peter

David SL said:

Jimi said:

I think I had the same question...just to make sure let me rephrase
it: So if I take a 3MP camera even though it has more MP than a
2MP camera it doesn't take better 4x6/5X7 pictures?

Click to expand...

Let's do a little calculating to find out.

I'll use my CP990 as a reference 3mp camera it takes images with
maximum pixel dimensions of 2048 x 1536 pixels.

Now if we were to print this image out on a printer that has an
output resolution of 300dpi (Most photo laser printers have at
least 300 dpi res. for example the Fuji Frontier Laser Photo used
by some of the best online photo finishers.) we would map each
pixel to each dot in the output media, mathematically we want the
number of pixels that maps to a real world inch so we do a straight
division of the pixel dimension by the dpi:

2048 300 = 6.82 inches
1536/300 = 5.12 inches

So a CP990 image if printed out at 300dpi will yield a physical
print with dimensions of 6.82 x 5.12 inches without scaling. It
covers a 4 x 6 with room to spare, in fact the image would need to
be cropped or scaled down further to fit on a 4 x 6 print. Let's
see how a 2mp camera fairs.

Let's use the 2mp CP900, with pixel dimensions of 1600 x 1200:

1600/300 = 5.3 inches
1200/300 = 4 inches

So its revealed that a 2mp image without scaling is printed at
300dpi at 4 x 5.3 inches which is less than the standard 4 x 6
print size. Most photofinishers automatically scale up the image so
that there is no border unless you ask for a print "to fit". If it
were printed at 5 x 7 there would be even more scaling, remember
that scaling increases the size of the pixels on the output media
which results in blurring of detail and increased pixelization. The
3mp image in comparison has its native print size very close to 5 x
7 : 6.82 x 5.12 as calculated previously, the scaling that would be
performed to have it fill the print edge to edge would be minimal
compared to the 2mp camera which was already being scaled at 4 x 6
size. You can see why more MP translates to a better image, since
more pixels are available to map to real world inches under the
output mediums print dpi rating.

300 dpi is considered the standard high resolution print dpi by
many graphic designers, most designs that are meant for high res.
display are proofed at that dpi or highier. We can then look at the
resolution of our camera in terms of the minimum dpi we would like
to see at a particular print size.

For example, if we wanted an 8 x 10 print edge to edge at 300dpi we
would need a camera with:

8 x 300 = 2400 pixels
10 x 300 = 3000 pixels

The just announced Canon D60 has a max. resolution of 3072 x 2048
thus even its images would have to be scaled/cropped to fit an 8 x
10 print edge to edge. It's dpi along the long side is:

3072/10 = 307.2 dpi

along the short side:

2048/8 = 256 dpi

--www.pbase.com/optionguru

Click to expand...

Yes, what you've written is spot on...except that it's only true for printers that produce true continuous tones like dye sublimation printers. For laser printers and inkjet printers that make "tones" or shades by grouping super-tiny spots of ink or toner in a tiny square, you need much more than 300dpi PRINTER RESOLUTION to reproduce 300dpi IMAGE RESOLUTION!!!

Remember, each pixel or dot in your image can have any value between 0 and 16.7 million. Laser printers and ink jet printers need more printer resolution than image resolution to reproduce the image properly simply because they cannot produce continuous tones. Each dot from the printer at the printer's maximum resolution is either Cyan, Magenta, Yellow or Black, and it's always full on Cyan or no Cyan at all, never 50% Cyan or anything like that. So to make up the continuous tone, you need a square of several printer dots to represent each image pixel. That means that if you used a 1440dpi printer, and you printed at 8x10, you would get the following results:

Start with e.g., 3mp image: 2048 x 1536. 8x10 photo will result in our image being cropped to 1920 x 1536. This means that our photo dpi in final printed form will have 192dots of image information/pixels per inch.

Printing at 1440dpi, means that each image pixel will be represented by 1440/192 = 7.5 dots each way. OK, let's say 7 dots each way. That means that the printer will be able to devote a 7 x 7 square grid of PRINTER RESOLUTION dots to represent each image pixel. 7*7 = 49. Cyan, Magenta, Yellow in simplistically speaking, 49 positions. The printer will combine Cyan, Magenta and Yellow in different amounts at each 7x7 grid to produce something that looks like tones at normal viewing distances (i.e., not under a 10x magnifying loupe) while adding black ink as well in specific cases to darken the grid as required. You won't get 3^49 combinations, because many patterns are equivalents of each other and give the same tone, so it is quite likely that the printer driver has some algorithm for generating the patterns of 7x7 printer dots to reproduce tones. The actual result might be that the printer cannot produce 16.7 million real tones at each dot at that resolution.

The other thing is that even with many dye sublimation printers that print true continuous tones, they only have a max resolution of 300dpi, so if you print 3mp, 4mp, 5mp or even a 12mp printer at 4x6, you won't see any improvement because they all have more information than the printer needs to reproduce a 4x6 photo at it's maximum resolution of 300dpi anyway.

Enjoy

Jason S Tay

David SL said:

Jimi said:

I think I had the same question...just to make sure let me rephrase
it: So if I take a 3MP camera even though it has more MP than a
2MP camera it doesn't take better 4x6/5X7 pictures?

Click to expand...

Let's do a little calculating to find out.

I'll use my CP990 as a reference 3mp camera it takes images with
maximum pixel dimensions of 2048 x 1536 pixels.

Now if we were to print this image out on a printer that has an
output resolution of 300dpi (Most photo laser printers have at
least 300 dpi res. for example the Fuji Frontier Laser Photo used
by some of the best online photo finishers.) we would map each
pixel to each dot in the output media, mathematically we want the
number of pixels that maps to a real world inch so we do a straight
division of the pixel dimension by the dpi:

2048 300 = 6.82 inches
1536/300 = 5.12 inches

So a CP990 image if printed out at 300dpi will yield a physical
print with dimensions of 6.82 x 5.12 inches without scaling. It
covers a 4 x 6 with room to spare, in fact the image would need to
be cropped or scaled down further to fit on a 4 x 6 print. Let's
see how a 2mp camera fairs.

Let's use the 2mp CP900, with pixel dimensions of 1600 x 1200:

1600/300 = 5.3 inches
1200/300 = 4 inches

So its revealed that a 2mp image without scaling is printed at
300dpi at 4 x 5.3 inches which is less than the standard 4 x 6
print size. Most photofinishers automatically scale up the image so
that there is no border unless you ask for a print "to fit". If it
were printed at 5 x 7 there would be even more scaling, remember
that scaling increases the size of the pixels on the output media
which results in blurring of detail and increased pixelization. The
3mp image in comparison has its native print size very close to 5 x
7 : 6.82 x 5.12 as calculated previously, the scaling that would be
performed to have it fill the print edge to edge would be minimal
compared to the 2mp camera which was already being scaled at 4 x 6
size. You can see why more MP translates to a better image, since
more pixels are available to map to real world inches under the
output mediums print dpi rating.

300 dpi is considered the standard high resolution print dpi by
many graphic designers, most designs that are meant for high res.
display are proofed at that dpi or highier. We can then look at the
resolution of our camera in terms of the minimum dpi we would like
to see at a particular print size.

For example, if we wanted an 8 x 10 print edge to edge at 300dpi we
would need a camera with:

Click to expand...

...snipped

You are correct. A 2mp, 3mp, 4mp, 5mp, 6mp and 12mp camera will all produce similar quality 4x6 and 5x7 prints ... ASSUMING THEY USE THE SAME LENS!!! Also, our hypothetical situation also assumes that they all use sensors with identical size, otherwise you could not do the comparison using the same lens, now could we?

Remember that 2mp cameras tend to be cheaper price bracket cameras. To save cost further in the already lower margin entry level segment, the manufacturer may be tempted to slap on a cheaper lens. Cheaper lens usually means lesser ability to resolve detail well. A 3mp Canon D30 D-SLR with a high quality expensive add-on prime lens will always take better sharper shots than my Nikon CoolPix 995 3mp camera with tiny slow lens. Will there be a difference in the prints? Yes, there will be, but in this case we're not comparing apples with apples, are we? Then move on up to higher megapixel cameras - to register more detail, the optics has to be better. They charge a premium for these cameras as well.

People ask me things like, "Am I correct in saying that I cannot get sharp photos with a 1.2mp camera?" which is of course incorrect. Sharpness is a function of the quality of the lens and the auto-focus electronics. The megapixels will determine the amount of detail that can be registered, which is of course, also tempered by the quality of the lens in resolving detail without too much optical distortion.

I would stick with the recommendation that you buy the largest megapixels you can afford while balancing that with your budget as you should not skimp on features as well if you're serious about photography.

Jason S Tay

Jimi said:

I think I had the same question...just to make sure let me rephrase
it: So if I take a 3MP camera even though it has more MP than a
2MP camera it doesn't take better 4x6/5X7 pictures?

I was holding out for a 4MP camera because I thought that since it
had more MP it would look better...but if that's not the case I'll
go out an buy a 3MP today!!!

Thanks

Click to expand...

Jason S Tay said:

Megapixel cound that is quoted comes from the product of the height
and width in pixels that you shoot the image at. So if you shoot at
2048 x 1536, the megapixel count is 2048 x 1536 = 3145728 or 3.15
megapixels.

dot per inch is usually used within the context of how you will
present the pixels on screen or in print. Higher dpi will mean a
smaller overall image size with the same number of megapixels.

2mp and 3mp will not normally reveal any difference at 4x6 or 5x7.
There will be a slight difference at 8x10 (depends also on the
subject matter and difference in quality of lens used on the 2mp
and 3mp camera), and the difference in quality will become larger
as you go to larger print sizes, until the 3mp maxes out. According
to Nikon, 2048 x 1536 at FINE quality will produce acceptable 11x14
prints.

Click to expand...

It depends on camera. More megapixel does'nt always mean better print. For example 3 megapixel for Canon D30 is much better, that 4 megapixel image from crapy S40, there are many other factors like noise to data ratio, lens quality, "smart" byer interpollation algorythm (compare Nikon D1 and D1h images, the same sensor, different electronics and firmware, difference is very visable, lok at red jaggiez of D1) and one of other important thing is dit depth of A/D convertor and ability to write RAW files. 8 bit per channel is good enough for storing images, but ain't enough for capturing 'em. Professional D-SLRs and video cameras has 12 or 14 bit analot-to-digital convertors. The wider dynamic range is, the better, always!

The lust thing that is important in this way is sensor, the bigger is is, the better. Nikon D1x is more noisy than D1h, why? Cuz it has the same size of the sensor, while D1x's one has more sensitive transistors, thus they are smaller, thus more niose. Manufacturers are trying to sole this problem, because to fight the film they seposed to create atleast 16 megapixel sensor 24*36milimeters size, with low noise. In the special multimillion dollars satellite digital video cameras, that has resolution of 100+ megapixel, they using air conditioneer to freeze sensor to the lowest possable temperature, to minimize noise level, because the colder sensor is, the less noise it produce. We don't yet have ability to have small conditioneer in our digicams, so there is no wonder that we are experiencing noise, ecpecially at high ISO levlers. And don't have ability to use F1:1 zooms in everyday life, like those on satellites.

David SL said:

300 dpi is considered the standard high resolution print dpi by
many graphic designers, most designs that are meant for high res.
display are proofed at that dpi or highier. We can then look at the
resolution of our camera in terms of the minimum dpi we would like
to see at a particular print size.

For example, if we wanted an 8 x 10 print edge to edge at 300dpi we
would need a camera with:

8 x 300 = 2400 pixels
10 x 300 = 3000 pixels

The just announced Canon D60 has a max. resolution of 3072 x 2048
thus even its images would have to be scaled/cropped to fit an 8 x
10 print edge to edge. It's dpi along the long side is:

3072/10 = 307.2 dpi

along the short side:

2048/8 = 256 dpi

You'd crop along the long side and scale up along the short side to
fit the page. Since the pixels are square we can find the
normalized print resolution at the scaled size by:

307.2 - 256 = 51.2/2 = 25.6 dpi normalized.

So the final print will have 25.6 dpi cropped off along the long
side and 25.6 dpi added (scaled) to the short side. Final effective
print resolution of:

281.6dpi

very close but not quite maxing out the print mediums 300dpi
capability since the short side only had 256dpi of res. to begin
with.

Since the long side is cropped we can find the number of pixels
trimmed to correspond to the scaling done on the short side.

10 x 281.6 = 2816 pixels along 8 inch side. (crop of 3072 - 2816=
256pixels)

I've printed out D1x images (3008 x 1960) at 8 x 10 without
scaling, because there is more than enough pixels to achieve the
300dpi print output on the long side(because its available dpi is
300.8 but since the printer can only output 300, the extra .8 can't
show up in the print), side bars are visible on the print since the
number of short side pixels is insufficient to cover the full 8
inches at 300dpi. It's easy to figure out how big those side bars
are:

1960/300 = 6.53 > > 8 - 6.53 = 1.46/2 = .73 inches

I just measured the print and sure enough, the side bars are .73
inches. The D60 and the newly announced D100 (3008 x 2000) would
reduce the side bars for an 8 x 10 "to fit" print since they both
have more pixels along the short side.

D100:
2000/300 = 6.66 > > 8 - 6.66 = 1.33/2 = .66 inches

D60:
2048/300 = 6.82 > > 8 - 6.82 = 1.17/2 = .58 inches

As illustrated earlier if they had short sides of 2400 pixels there
wouldn't be any scaling, it would be a perfect mapping to 8 inches
2400/300dpi.

Click to expand...

Thanx for smart and easy to under stand explanation, I knew it, but I hope it will help many other people to understand. I think, you gotta repost it to the all main forums, like General Talk. It is gonna be like FAQ.

Just want to add, that dots per inch, and lines per inch are equal only for dye sub and to-chemical printers. For ink jet and toner laser printers these numbers are different. I will not tell you exact resolution of them, and it depends on a few factors, just will tell that resolution of CMYK(4 color) 720*1440 dpi Epson is about 140-160 lpi, CcMmYK(6 color) Canon S900 1200*2400 dpi is about 260-300 lpi (like offset), for CMYK 1200 dpi laser printer, it is gonna be about 160 lpi. Lines Per Inch is a measure, that used to describe, how many different anycolor lines printer could print on the one inch of paper. As I told before, for dye sub printers dpi equals lpi, for other printers it could be measures by printing special target images.

David SL said:

I think that my hunger for increasing megapixels will end once I
am able to print an 11 x 14 inch print at 300 dpi. We can use the
same principles applied above to find out the minimum sensor
dimensions needed for this.
11 x 300 = 3300
14 x 300 = 4500

Click to expand...

Well, 35mm film could satisfy your hunger today, until 16 megapixel mobile digicam is not yet available. Resolution of Nikon coolscan 4000 ED output file is about 24 megapixel, even more than you need, so you've got a big crop ability!

David SL said:

3300 x 4500 pixels or about 15 MP. (14,850,000) I figure that's
about 2-3 years away. If the sensor dimensions are larger we can
crop away the excess and still print at the output mediums max.
assuming its 300dpi.

David SL said:

I was holding out for a 4MP camera because I thought that since it
had more MP it would look better...but if that's not the case I'll
go out an buy a 3MP today!!!

Click to expand...

As illustrated above it may look better depending on the resolution
of the output medium. For Fuji Frontier prints its 300dpi, but that
would vary depending on what you use, for example the numbers on a
home inkjet would be different. The lower the output resolution the
larger size print you can get before seeing a difference in quality
between the two sizes. However, given the same output dpi you can
always calculate the print size and effective print resolution of
an image of a given MP size. For 3 and 4 mp cameras, 4 x 6 and 5 x
7 will roughly look the same but 8 x 10 and highier will look
better for a 4mp camera than a 3mp at 300dpi.

Click to expand...

Peter B. said:

Wow,
I think you should get an award for putting the most time into a
response. I must say that this site is great, I ask a question and
now I have the answers.
Just like my saltwater fish tanks, half of the fun in a good hobby
is the learning.

Click to expand...

Ye, thats why I like forums like this, people are always happy ot help newbie, explain different questions. I think Phill have to create a FAQ, for the questions like What is dpi, lpi, what factors are important for the digicams and so on.

Jason S Tay said:

Yes, what you've written is spot on...except that it's only true
for printers that produce true continuous tones like dye
sublimation printers. For laser printers and inkjet printers that
make "tones" or shades by grouping super-tiny spots of ink or toner
in a tiny square, you need much more than 300dpi PRINTER RESOLUTION
to reproduce 300dpi IMAGE RESOLUTION!!!

Remember, each pixel or dot in your image can have any value
between 0 and 16.7 million. Laser printers and ink jet printers
need more printer resolution than image resolution to reproduce the
image properly simply because they cannot produce continuous tones.
Each dot from the printer at the printer's maximum resolution is
either Cyan, Magenta, Yellow or Black, and it's always full on Cyan
or no Cyan at all, never 50% Cyan or anything like that. So to make
up the continuous tone, you need a square of several printer dots
to represent each image pixel. That means that if you used a
1440dpi printer, and you printed at 8x10, you would get the
following results:

Start with e.g., 3mp image: 2048 x 1536. 8x10 photo will result in
our image being cropped to 1920 x 1536. This means that our photo
dpi in final printed form will have 192dots of image
information/pixels per inch.

Click to expand...

It is called Lineature. If you're printing with regular raster, like in offset and laser printers (not always), it has fixed lineature that seposed to overrided atleast by 150%. If you're printing with ink jet printer, that uses stochastic raster, it doesn't have a files lineature value, it just has an approximidiate resolution that is minimal, otherwise you'll see small cubes of image data.

BTW, don't forget, if you want to print a file on the big paper, so youe image's dpi is less than printers lpi on this size, you gotta blow it up (upsample). You'll not see more details, but transition between pixels will be invisable. Like they don't when printing A3 size photos from D30.

Jason S Tay said:

Printing at 1440dpi, means that each image pixel will be
represented by 1440/192 = 7.5 dots each way. OK, let's say 7 dots
each way. That means that the printer will be able to devote a 7 x
7 square grid of PRINTER RESOLUTION dots to represent each image
pixel. 7*7 = 49. Cyan, Magenta, Yellow in simplistically speaking,
49 positions. The printer will combine Cyan, Magenta and Yellow in
different amounts at each 7x7 grid to produce something that looks
like tones at normal viewing distances (i.e., not under a 10x
magnifying loupe) while adding black ink as well in specific cases
to darken the grid as required. You won't get 3^49 combinations,
because many patterns are equivalents of each other and give the
same tone, so it is quite likely that the printer driver has some
algorithm for generating the patterns of 7x7 printer dots to
reproduce tones. The actual result might be that the printer cannot
produce 16.7 million real tones at each dot at that resolution.

The other thing is that even with many dye sublimation printers
that print true continuous tones, they only have a max resolution
of 300dpi, so if you print 3mp, 4mp, 5mp or even a 12mp printer at
4x6, you won't see any improvement because they all have more
information than the printer needs to reproduce a 4x6 photo at it's
maximum resolution of 300dpi anyway.

Click to expand...

Sorry, maybe my explanation was little raw, and there are some grammatical problems. Mybe I'm not a good teacher, cuz it doesn't always mean that if you know something, you can easly explain it. Anyway, I tryed my best.

Igor Radionoff said:

It depends on camera. More megapixel does'nt always mean better
print. For example 3 megapixel for Canon D30 is much better, that 4
megapixel image from crapy S40, there are many other factors like
noise to data ratio, lens quality, "smart" byer interpollation
algorythm (compare Nikon D1 and D1h images, the same sensor,
different electronics and firmware, difference is very visable, lok
at red jaggiez of D1) and one of other important thing is dit depth
of A/D convertor and ability to write RAW files. 8 bit per channel
is good enough for storing images, but ain't enough for capturing
'em. Professional D-SLRs and video cameras has 12 or 14 bit
analot-to-digital convertors. The wider dynamic range is, the
better, always!

Click to expand...

Yes, that's right, which is why I alluded to this in one of my posts in this thread that comparing prints of different megapixel cameras at 4x6, a Canon D30 will definitely be much better than my Nikon 995 because of the opportunity to fit a better lens to the D30 eventhough both are 3megapixel. Also, the D30's algorithms may also be more refined/sophisticated.

That's why I always try to explain to people that megapixels will determine the amount of detail that the camera can register out of what the lens delivers. Sharpness is dependant on the quality of the lens and the Auto Focus system. Many people mix and confuse these two very different things.

Igor R. said:

The lust thing that is important in this way is sensor, the bigger
is is, the better. Nikon D1x is more noisy than D1h, why? Cuz it
...
satellites.

Click to expand...

Uh, not "lust", you better make that "last"! Lust is something that a guy might feel for a woman or a person might feel for a desirable camera!

I hear actually that to make full frame sensors more affordable, there are projects underway to try to make full frame sensors out of thin film transistor (TFT) technology. Apparently it's a much cheaper way to make such a large sensor. Also apparently the noise level is lower. But that could be largely hearsay and could still be years and years away.

Jason S Tay

Igor Radionoff said:

It is called Lineature. If you're printing with regular raster,
like in offset and laser printers (not always), it has fixed
lineature that seposed to overrided atleast by 150%. If you're
printing with ink jet printer, that uses stochastic raster, it
doesn't have a files lineature value, it just has an approximidiate
resolution that is minimal, otherwise you'll see small cubes of
image data.

Click to expand...

Hey Igor, thanks for the info. It's always better to explain when one also knows the correct name for it!

Igor R. said:

Sorry, maybe my explanation was little raw, and there are some
grammatical problems. Mybe I'm not a good teacher, cuz it doesn't
always mean that if you know something, you can easly explain it.
Anyway, I tryed my best.

Click to expand...

Not at all, you're doing just great. Carry on the great contributions you are making to the forum!

Jason S Tay