High Resolution for Print

[Joe]

Re: High Resolution for Print

> {quote:title=russpears wrote:}{quote}
> but how can line art be output at 6-8 time the resolution of the image when the line screen is the same. Are you refering to in the RIP half tone dot creation resolution. Or are you saying that you can increase the line screen in a way that only affects vector line art not raster images.
>
> I know everyone has give me much to think about, but getting info like this puts me back a bit where I "thought" I started to understand. I really cannot tell where the confusion comes from. Once I understand that stocastc dots are higher resolutions (2400-3000 and up) and not the halfton dots themselves, I could see how I confused the two as the same as line screen that are around 150 to 175). But this answer suggest to me that you can increase the final quality of line art in print work apart from raster images?
>
> Again I will need to do more research into this before post more on this, but if I can be given a direction towards possiable areas of my confusion I will go there first.


You are confusing line screen and resolution. You can make an image @ 72 ppi and output it at any linescreen your platesetter is capable of. But your RIP is going to rasterize and screen that file at the resolution the device needs. Like 2400 dpi. The dots in the one-bit file will be 2400 dpi. There is no longer a pixel based image involved. It's just dots. Either white area or black dots. That's all that is in the file at imaging time. And it's the dots that are 2400 dpi. The lower your linescreen the bigger the dots will be. Higher line screen means the dots will be smaller (and appear smoother to the human eye).

 

[Joe]

Re: High Resolution for Print

> Again I will need to do more research into this before post more on this, but if I can be given a direction towards possiable areas of my confusion I will go there first.


Here is a good exercise to try to grasp what is going on. Open a grayscale photo in Photoshop. Go up to the Image menu and click Mode:Bitmap to convert the image to a bitmap. Type in 2400 for your output resolution and under method select "Halftone Screen". Clcik OK. In the next box type in the frequency which is your LPI. Try 150. Put in any angle you wish and then select the "Shape" you want to use. Round, Diamond, Ellipse...whatever. Then click on okay. Your resulting image will be a screened one-bit image at the LPI you chose. Try it again with a different LPI number and compare the screen. Like from an 85 LPI and 175 LPI. You should see a noticeable difference when you zoom in at 100% view. The 175 LPI will be a finer screen.

Hope this helps some.

 

[russpears]

Re: High Resolution for Print

so what your saying is that my (300ppi raster image) and (1200ppi vector or line art) image will be processed into the 1 bit stocastic dots at around 2400dpi, but that resolution only goes to making the halftone dot that is used for what ever line screen the printer desides to run on the press be it 150lpi or 170lpi.

Now, if I get this right, both my low and high resolution-relitively speaking, will be at the same output resolution of the line screen and as such will should not appear different even at the level of the Halftone dots-although some will be a better defined dot than others. Now all line art should be cleaner with the benifit of the 2400 dpi stocastic dots, only because if it is a solid color on the plate-since (100%C). Moreover, any screens (less than 100%) will render as halftone dots. Here all solid 100% C, M, Y or K line art resolves at 2400dpi and all other screen will appear the same regardless if it is a continuous tone image at 300ppi or a vector line art without a solid 100% fill in plate C,M,Y or K.

Now that makes sence to me, but do I have it right?

 

[russpears]

Re: High Resolution for Print

a few key ides seem to be missing for me

 

[russpears]

Re: High Resolution for Print

I see now my confusion, but I hope that am not further mistaken on this. First off all the answers deserve my gratitude. You guys are great and I have really studied all of your replies.

As a designer, I want to get better at Prepress issues but I’m the type that needs to understand key ideas, rather than just talk about thinks I do not really understand to clients or printers.

Tell me if I am wrong, but vector art can resolve cleaner edges in any design, because it can make use of the 1200-2400 dpi the CTP or imagesetter will use to build the halftone dots- the line work will have a 2400 dpi edge and this is consistent with a 150 lpi halftone dot made with a 2400 dpi CTP). While the rest of the continuous tones in the design will only be able to be resolved at the halftone dot level where 300 ppi images will be relevant.

Did this get it right?

 

[Blair Wolfe]

Re: High Resolution for Print

Do you understand what vector art really means? This is not generaly used in a discusion about dot resolution, because a computer doesn't use a bitmap or series of pixels to create vector art. As the term implies, it marks the starting point for the line or shape that is vector art, and calculates a staight line to the next point in the figure. This is why vector art can be scaled to any size with no lose of clarity. The distance between points is just increased, and a staight line is sent from point A to Point B. With a bitmap or a dot based input image, enlargement expands the dots or pixels until they become so big that they look rough & chunky. Unfortunately, haftones cannot be vector art because they originate as a serries of dots. The other crazy part to this, of course is that the imagesetter is only able to create an image from either type by means of very fine laser dots. At least with vector art, there is no patern getting enlarged by the computer sending the file.

 

[russpears]

Re: High Resolution for Print

Yea I do know what vector art is and I do think that if a peice of art has a gradient in it with a percentage of one or more of the process colors it will be rendered out as halftone dots in the CTP. But my point is that it seems to me if you have a defined box that is vector and 100% K(resolution independent) it will be produced as a box at the resolution the CTP used to produce the halftone dots 2500ppi. Now if I am wrong about what happens at press or the CTP, fine I will try to understand better. But I do know what vector art is.

 

[John Clifford]

Re: High Resolution for Print

Vector data is always RIPped (a Rip is a Raster Image Processor) to the resolution of the device the data is being sent to. If you have a vectored item and it's going to a 300dpi laser printer, the resolution of the finished printed piece will be 300 dots per inch. If you image the same item to a 2450dpi platesetter, the resulting image will be 2450 dots per inch. Resolution is only critical if you're screening an image for press. Thus photos, or raster images, are screened at an lpi (this determines the size of the halftone dots -- which are not the same as resolution dots). The RIP will create the dots using the resolution of the output device. Of course, if there is more data than necessary to create the halftone dots, that data is "thrown away." That is why you won't see any increase in print quality from higher resolution images. I had this argument once with a publisher who insisted that if they had their vendors scan at 600dpi, that it would increase the quality of their image 100%. The problem was that they believed it, despite the fact that there was, in reality no difference as they were printing at 150lpi, so anything over 300dpi was lost in the conversion to halftone dots.

I hope I didn't confuse the issue even more. I think that we might be better served if we look at our terminology and come up with a different term for lpi dots, maybe lpi specs, or lpi units. I don't think we'll get the computer people to change the term for dots in terms of resolution.

 

[John W]

Re: High Resolution for Print

<Of course, if there is more data than necessary to create the halftone dots, that data is "thrown away." That is why you won't see any increase in print quality from higher resolution images. I had this argument once with a publisher who insisted that if they had their vendors scan at 600dpi, that it would increase the quality of their image 100%. The problem was that they believed it, despite the fact that there was, in reality no difference as they were printing at 150lpi, so anything over 300dpi was lost in the conversion to halftone dots.>

You are so dead wrong here. If you are using a postscript rip, placing a 300dpi image @ 50% or a 600dpi image @100% (the equivalent), you do indeed get higher quality. As an example, the lines along the side of a car image will be smoother at the higher resolution while steppier (sorry for the lingo) at the regular resolution. Do the test and see! That is why some car brochures images look crappy while others look fabulous.
If you use a pdf workflow and do not alter the job options of some default settings to prevent downsampling when above 300dpi, you will see the inferior steppy lines along the side of the car as above. If you keep the extra resolution, you will see the extra quality. Problem is few do the proper test to prove it. We have and it shows in a proper proof (not a bloody inkjet). We teach about this at our union school and show hard copy examples.
Also, blends (gradients) will have a richer (more data colected) result if ripped at higher resolution even if the screening is not that much higher. We now see variable resolution plate setters (Screen has always had them while others were fixed resolution such as 2400) which can produce more quality because more initial data is collected at the Rip. We've done numerous tests on press to prove this and got a Screen (the brand) platesetter to utilize it. Another example is fine small script text which is superior at finer resolution (say 3600 or 4000dpi) even when using 150-175lpi.
Anyone who says this is always overkill is cheating you and themselves (granted sometimes it is overkill). I do not want to allow watered down quality as the norm so I will always speak up about this.

John W

 

[russpears]

Re: High Resolution for Print

I think we all know what vector art is and that everything gets rastered at the RIP. The finer point of this is that with anything 100% on any plate, the art will render at the CTP resolution say 2500 dpi. Anything with 99% or less will start to reveil the halftone dots the CTP is actually outputting for all art at the specified lines per inch (75 - 150)- my gradient example is such a case.

However and if I understant this right, type can be a different case: here 100% K type looked like it lacked the halftone dots, no doubt because the type characters were 100% filled. Thus what I just described should mean that any vector object with 100% fills in the original file should, on any plate, be resolved at the higher resolution of the RIP/CTP of 2500 dpi or more with no visiable trace of the halftone dots and at a resolution to that of the CTP, if I understand the results of the RIP correctly?

Does this sound right to you printers?

 

[John W]

Re: High Resolution for Print

Russell,
No platesetter nor printing press is a vector machine as you know, the're all bitmap machines which means pixels become bitmaps such as a one bit tif. The resolution of the RIP which is matched to that of the platesetter (or variable with one of the resolutions selected to match the platesetter) creates the tiny bitmaps. For linework (LW) which involves vector data such as curves, masks, clipping paths and type, the higher resolution, perhaps 2540 is used. For the images and vector blends (which from a vector program are simply hundreds of flat tints that change sequentially) or image blends (blends made in an image program with a little 'noise' added) the rip generally likes twice the screen ruling so 300 dpi for 150lpi. But the RIP is interpolating the data and modifying it to suit the screening of the device. It is that interpolation that can be altered and improved by using a higher effective resolution for the images. In that way, a higher resolution image or one placed smaller than 100% of size sends more collected data to the RIP. Analogous to a chef using a 10,000 page recipe book instead of a hundred page one. He takes longer to cook the food but produces a finer and nicer (nice dictionary meaning is precise - this thanks to Tony Randall) product.
Then the postscript rip is switching its language using the interpolation to produce the screening. Each of the halftone dots of the 150 lpiscreen is made up of the 2540 bits. If you view this with a high powered glass, you can actually see the lumpy nature of the halftone dots. In layman's terms it is these lumps and the averaging (dithering) that allow more smooth transition within the halftone dot field for the smooth detail of the image. Dot differential means detail but interpolation means coarser or finer detail.
John W

 

[russpears]

Re: High Resolution for Print

So what your saying is that the higher the ppi of the image sent to the rip the more precise it can be when determining, not just the size of the halftone dots, but also the subtle variations in the shapes of the halftone dots themselves that lends it’s self to better detail. Now I this is a different-and interesting issue, and I still have not gotten any verification about the effective resolution of the line art from the RIP. I know the rip will look to continuous tone images and interpolate them as varied halftone dots, but am I right about the resulting resolution of the type (line art ) that are 100% on any plate. Again I’m not thinking the RIP is outputting any vector art here, but that the rastered file from the CTP example should render solid 100% K type at the 2400ppi even if the Line ruling is set at 175 lpi.

 

[John W]

Re: High Resolution for Print

You should have had your answer. All data must be ripped and rastered at the set resolution, including line art and text. Text after all is a quasi vector item; draw around the letter adding or subtracting for welling (where the sharp crevices are) and then rastering at say 2540. As I stated, if you look at the lineart you will see the rastering edges. They are indeed fine.
So yes 175lpi at 2400 resolution is an example. We feel that 175lpi requires a higher resolution perhaps 2438 or 2540 or even 4000. Then the lineart edges are less noticed than at 2400. But there is a compromise as such within the device handling including some smoothing functions etc depending on whose device we're talking about plus one has to factor speed, memory etc.
Back in the 1980's, we scanned at 350 dpi on a Screen 608 scanner for 150lpi images. That was because the scanner ran from -14% to 113.5% dot allowing us to bring back detail from less than 0% (highlight) and more than 100% (shadow) should we so choose. We used this technique to gather and show more detail in a white wedding dress or dig more detail from the shadows of a furniture setting. We scanned at 400dpi for 175lpi as well. The LW data was converted as L-tech data at five times the image resolution, so 350 was at 1750 and 400 was at 2000. Now the devices were different and screening did not have all the modern calculations of today but in a simplified way, it does apply.
John

 

[russpears]

Re: High Resolution for Print

Sorry, I cannot follow your explanation too much. But in more of a laymans terms, is the resulting edges of 100% K filled type from the example RIP resolved at 2500 dpi given that the halftone dots are completely filling the whole text shape. In short, if I sent two images one a 300ppi rastered text and one as outlined (vector) type both 100% K. The rastered type will not be able to make use of the 2500 dpi, like the vector type can given its resolution is fixed at 300ppi.

 

[John W]

Re: High Resolution for Print

Strictly speaking, that is correct. However, an application such as Photoshop has the ability to tag the text as vector when going to the rip even from within the image. If the Rip is smart enough it can generate nearer text quality from within the 300dpi image at rip resolution. So it wholly depends upon how that text was made. what rip sees it etc.

Generally, we do not want any text within the image (mostly for trapping reasons). That is not to say it cannot be done. Some rips can improve the 300dpi result and some rips can enlarge images up to twice size without image degredation.

In fact, you are posing a difficult question that does not have a straight answer because of a myriad of variables. Sorry bout that!

John W

 

[russpears]

Re: High Resolution for Print

So in general the notion is correct, right?

As far as the vector info from PSD files being used in RIPS thats great. I know we can pull that same information into Illustrator and its good to hear the RIPs are makign use of it too.

 

[michaelejahn]

Re: High Resolution for Print

@ Russel Spears,

There are many different production environments where different input scanning resolutions are required - but if your final product is a printed sheet, and the halftone dot patteren they will use are dots, and the 'line screen' they will use is 175 - then contone images (like a picture of a face or a product "Continuous tone" should be around 300 pixels to the inch - and high res graphics - like a black and white logo that was 'scanned' - bitmap images (1 bit, Black OR white, no grey) should be scanned at 900 to 1200 ppi.

Should these two types of images overlap - it reduces the unsightly problem of having a visulayy noticeable jagged edge where the two pixels intersect.

These are rules of thumb, and not rules of law. For example - I just finished up a software manual, full of screen captures from an application - not one single image in this manual is higher than 96 pixels to the inch - in a few cases, lower where i 'zoom in' to point out an imaging effect.

Another thing you may consider is that some people takes fine detailed images like a bicycle wheel with spokes - and while the digital photograher shot it so, at a ceryain size, there would be an "effective resolution" of 300 ppi, a designer may have enlarged it in an application like Quark or InDesign, and then the effective resolution is now lower - so now the image looks jaggie - the spoke lines look like a stairway - this is what they mean of 'effective resolution. In many images, this matters little - faces, clouds, grass and often clothes - but when you have pots and pans, jewlery and other images with both detail or high contrast - the effective resolution issue is more important.

This becomes more complex when you are in the poster or billboard world, as one needs to factor in the viewing distance...

hope this helps !

 

[John Clifford]

Re: High Resolution for Print

John Willis,

Sorry it took me so long to respond. Yes, data IS thrown away in the scenario I've presented. 150lpi is 150lpi. The car edge will be imaged at 150lpi regardless of whether you start with a 300dpi or 600dpi image. The fact that printing has limitations is at work here. You can't include more data than can be put down on the plate. The dot size (made up of the resolution dots) is finite. You can only fit so much data into that dot. The additional data in dpi, is outside the lpi dot.

I'm sure that I should be able to make this clearer, but the reality is that you can't fit 10lbs of s**t into a 5lb bag.

 

[russpears]

Re: High Resolution for Print

Are you saying that the finer image detail represented in the various halftone dot’s shape itself, offer negligible detail to the overall printed work? Like wakening up to an atomic clock over a typical clock?

Edited by: Russell Spears on May 28, 2008 11:51 AM

 

[russpears]

Re: High Resolution for Print

I can see how 1 bit line art can benefit as well from higher resolutions, but vector art that is any solid process color should benefit from the maximum resolution of the RIP. Even your 600 or 1200 ppi line work would not render as smooth, I would think as the vector art would once it was rastered at 2500dpi? Right? I just wanted to have this notion confirmed or disconfirmed for me, because it will help me set up documents better. Does this make since from a printer’s perspective?