Can someone explain DPI?

[maynardsayswhat]

We're printing and design professionals. We should use accurate information and have a very clear understanding of how things work, why they work that way, and how we can fix projects when they ain't working.

Or am I crazy?

 

[UberTech]

This is a demostration of laser power rather than the actual effect it has on the paper/toner. Some grayscales are crude for demonstration purposes.

 

[gordo]

This is a demostration of laser power rather than the actual effect it has on the paper/toner. Some grayscales are crude for demonstration purposes.

OK, You didn't provide any details with the graph so I'll give my interpretation - please let me know if it is correct.

The background grid is the addressability grid - it is 2400 dpi.

The circles represent splats of laser energy, ink, toner or whatever.

The left most 600x600x8 graphic represents splats of energy about 1/600ths of an inch in diameter. Each splat of energy can be modulated to represent one of 8 levels of grey - although only 4 levels are shown.

The middle 2400x2400x1 graphic represents splats of energy about 1/2400ths of an inch in diameter. Each splat of energy can either be 100% on or off. This is how CtP and imagesetters work. In this case grey levels are created by building halftones.

The right most 1200x1200x4 graphic represents splats of energy about 1/1200ths of an inch in diameter. Each splat of energy can be modulated to represent one of 4 levels of grey - and 4 levels are shown.

The hair diameter is for scale - it is about 100 microns in diameter - about 1/24 ths of an inch. (Actually that's a bit thick - most human hair is about 30-50 microns wide).

Is the above interpretation correct?

thx, gordon p

 

[UberTech]

Sorry, I should have added a bit more of an explanation.

The left most is 600DPI at 8 bit, the grey scale to the extreme left is the representative of the 255 grey steps possible with 8bit. The 4 circles are 3 different 'power levels' 100%, 75%, 50% and 25% I have done this for each method for demostration how each deals with different levels. Keep in mind the 2400 example is just that, an example.

The 1200DPI at 4bit greyscale is coarser than the 600 DPI version as it is 4bit and only has 15 grey levels.

Laser power is adjusted using pulse width modulation 8-bit and 4-bit terms are computer term and translates to binary. Go to this website and enter 8 '1's (11111111 as in 8-bits) in the top box and see what the decimal value is, do the same for 4 and 1.

At this point the xeroids will pipe in and say that 2400x2400 is more like offset which is true but unfortunatly it is a technology that does not support anti-aliasing and lacks the colour fidelity that multi bit processing can offer. I think when then 1200x8 engines come out from KM everyone will stand back and go, oh.

 

[Keith]

This is how CtP and imagesetters work. In this case grey levels are created by building halftones.

thx, gordon p

This thread has been great! I have earned a lot. But, now I want to know, what exactly is a halftone? I'm curious as I have a Xerox that addresses up to 2400x2400 at 1 bit depth.

 

[gordo]

This thread has been great! I have earned a lot. But, now I want to know, what exactly is a halftone? I'm curious as I have a Xerox that addresses up to 2400x2400 at 1 bit depth.

Try: Quality In Print: Halftones and grey levels - part 1 of 2

gordon p

 

[Eyal Gragir]

the images are misleading. the human eye will not see all those details but an integral of the very small details so the difference will be less appartent if at all. in addition laser dots are not separated so that the surfaces and dots look much smoother.
EG

 

[CoQ85]

i know 8 bit means 256 values. so a 600x600 8 bit printer can really create 256 different steps per channel? how does this work? does it vary the voltage on the laser or something to control how dense the toner covers the paper?

i thought the whole point of halftone screens was to make it possible to simulate different tint percentages assuming the limitation either "putting a spot of ink/toner or not".

i've been thinking about this, and i always kinda assumed that the 600x600 8 bit just meant that the rip flattens everything to a 600x600 8 bit raster image and anti-aliases etc, but then the imaging device halftone screens it at 1bit at some unmentioned "device resolution".

i assumed this was the case because i didn't realize there was a way to create different shades without halftoning.

can anybody clarify on how a printer creates 256 distinct levels of cyan without simulating it using a screen?

thanks

 

[Eyal Gragir]

the 8 in this discussion is 8 levels not 8 bit.
it can be achieved by controling laser start stop in between the pixels on the scan axis, thus creating more than 600x600 combinations of area coverage.
or it can have 8 drop sizes in the inkjet case.
any of the above is done in the imaging device. in some inkjet cases the RIP is translating the 256 gray levels of a separation to 8 gray levels.
I hope this clarifies it.
EG

 

[David Ingram]

Also

Also

Printers like dye sub use a thermal head so 256 diff tempratures result in the "Screen" of the color - screen is kinda diff w dye sub. but the person above me got it 1st, it varies the dot size. But, as soon as you are "tinting", the imager is laying down droplets of ink in smalle rand smaller fashion as you get to a 1% or 0% dot on paper.

Tint = screen for the most part, now some inkjets also use Light Cyan, and Light Magenta and in some cases 3 diff levels of Black, K, LK, and LLK.

So depending on the device, and whats driving it, 50% cyan may be represented by
LIGHT Cyan, instead of cyan itself... till 2005 or so the shops around her used negatives.

Imagine burning a plate for press that was cyan, the solid areas of ink would be clear on the negative, the tited ares would be halftoned, at what ever LPI / frequency the device marked it. MOST commonly 2400 DPI (device Res) and 150 LPI (Halftone Screen frequency)

There are new screening technologies, such as FM, Crystal Raster, etc. that use a "Stochastic" screening method, wherein the screening does not conform to the conventional angles. but the master imager is still using DOTS.

in a Linear system, 50% screen of say cyan, is 50% on PLATE say, then gains however much for the type of paper etc...

 

[UberTech]

the 8 in this discussion is 8 levels not 8 bit.
it can be achieved by controling laser start stop in between the pixels on the scan axis, thus creating more than 600x600 combinations of area coverage.
or it can have 8 drop sizes in the inkjet case.
EG

Ah, yes it does. Check the specs. 256 grey levels per colour. Infact your entire post is completely bizarre. The power level is a adjusted using Pulse Width Modulation. Effectively converting digital to analogue.

 

[Bo3b]

with 600x600x8 each square inch has 600x600 dot (360000 dots) in xerography the 8bits means that each dot can be a different size, 8 bits means 256 different sizes for each dot. In CMYK data each plate is 600x600x8 - varying the dot size gives the impression of contone images. Varying the dot size means varying the laser intensity on every dot which in turn can lead to mottle in solids and can also produce unstable patterns. Blends (vignettes) is also more difficult to do and can produce banding.
in the 2400x2400x1 printers the laser intensity is constant as it the dot size. Now each square inch can contain 5760000 dots. However to create halftones we now use a pattern of dots (screen if you like) what is different is that with 600x600x8 each halftone effect was created with 4 dots of varying size (1 for each of CMYK) we now are using 64 dots (16 for each of CMYK).
The 2400x2400 printer will have smoother/sharper text and line art, more stable solids and less steps and color variations in blends.
All the RIPs, however, are still ripping the files at 600x600 - the print engine interpolates this data to the 2400x2400dpi for output. Ripping time would take considerably longer, and the data sent to the printer significantly larger if the RIPs ripped at 2400x2400dpi. This is done to maintain the production printing speeds - shifting the amount of data that would be produced in a 2400x2400 CMYK file would take too long for a printer that prints faster than ~50 pages per minute so to get the engine printing at rated speed you have to send smaller chunks of data, hence the 600x600 ripping

 

[UberTech]

Nope, hah hah wrong again! Is anyone actually looking at the picture I posted? You are using 16 dots to make the equivalent of one dot. Put simply 16 dots to simulate 1 8 bit dot.

 

[Bo3b]

No Not wrong. 1x8 bit dot does equals 16x1 bit dots. If we're printing colour then there are 4 plates thus for some colours you need 4x8 bits dots or 64x1 bit dots.
From experience I've seen very little difference when looking at images between 600x8 or 2400x1 apart from the obvious halftone vs contone differences. The big difference is with text and line art - the 2400x1 output is much sharper than the 600x8 in this area.
The 2400x1 bit output is certainly more stable in solids and the blends are certainly better than the output from 600x8 bits
Having said all that, some of the best digital output I've seen lately has been from an older Xerox 2060 (which is 600x8).

 

[angryfish]

canon do not surport over 120lpi ,so 1200dpi is enough