Is it excessive to be working in 600 dpi for doing cartoon art in Photoshop?

March 18, 2009 4:13 PM

Is it excessive to be working in 600 dpi to do cartoon art in Photoshop?

I'm a cartoonist, and I've been drawing cartoons for quite some time using a mix of digital and traditional mediums. Basically, I draw my artwork on paper, then scan it in and color it with Photoshop. As time has progressed, I've bought better and better scanners, and faster and faster computers, and as a result I've steadily increased the DPI at which I both scan and color at.

I now always scan my toons at 600 dpi, and do all my coloring in un-resized 600 dpi files in Photoshop, too.

My question is, is this excessive? I've never really known a lot about the technological side of what I do, but lately I've been feeling a bit frustrated that my brushes are always too small and my file sizes are too big. I do like to print my work, and save it at high rez, and as a result I worry about loosing quality, but at some point I know there's quality, and then there's needless (and even counter-productive) excess.

What do other people do? And why?

posted by filibuster to media & arts (26 answers total)

If you want the option of being able to print your work large later on, have no trouble managing large files and have a computer that can handle the, then there's no reason not to work in 600 DPI.

I usually work in 300dpi or above just so when the day comes when I'm awesomely famous and they want to make an "Art of" book about me I have all the files there ready to go.
posted by AzzaMcKazza at 4:22 PM on March 18, 2009

I don't do too much drawing in GIMP (I use gimp, but same thing for this purpose), but I scan and edit at 300, though I might do 600 if I had my own scanner and wasn't using my school's computer lab where I'd rather not wait for like 5 minutes for a page to scan.

If space and speed aren't issues, then I wouldn't say that 600 is excessive. Overkill is good. But then, 300 is plenty, and pretty large if the brush sizes bother you.
posted by cmoj at 4:39 PM on March 18, 2009

600dpi sounds about right to me as long as you're continuing to 'paint' them, sure.

If you're interested in upgrading* your cartooning skills, though, you'll forget about dots completely and begin to work in vectors (Illustrator, not Photoshop) at which point resolution is infinite.

There would be a learning curve, and your "style" might have to change a bit to support it, but 10 years from now you'll be very happy to have all your work infinitely-scaleable and in perfect fidelity.

(* more like 'modernizing', I know.)
posted by rokusan at 4:43 PM on March 18, 2009

Actually, after a look at your website I rescind my hesitation: start learning to cartoon in Illustrator instead of Photoshop immediately. Your style is perfect for vector-based line art.
posted by rokusan at 4:44 PM on March 18, 2009

First of all, what you're probably dealing with is ppi (pixels per inch), rather than dpi (dots per inch). The difference is that one pixel is equivalent to several dots, because a pixel can be any colour, whereas to create an equivalent amount of graphic detail with dots requires several dots because the printer does this via halftoning with four inks.

Generally you work in pixels, and the printer converts it to dots, and these days you don't need to worry about the dots so much, just your pixels. Partly as a result, dpi is often confused for ppi (even in software interface - Photoshop had these mislabeled until a few years ago, making it mainstream) and so the terms have become hopelessly muddled.

Now, 300 ppi is close to double the graphic detail of what almost any printer can print on paper, which is about right - if the printer's print detail could match your image's resolution, jaggies would start to become visible. So using a image resolution 1.5 times greater than what the printer can print is about minimum to preclude jaggies, and 2 times greater (ie 300ppi) is about maximum (beyond which there is no advantage, unless of course you later wanted an enlargement print from the same file).

Which is a long-winded way of seconding AzzaMcKazza - 600 ppi sounds excessive for you (while 600 dpi is probably not enough and not what you're using), excessive unless you want the option of enlargements later on. (And even so, 300 ppi will give you a bit of wiggle room for enlargements, to a point).

300 ppi is a very commonly used resolution for print media. It's one of the biggest standards, and your image quality will be fine at that resolution.
posted by -harlequin- at 4:49 PM on March 18, 2009

When I was coloring alt comics back in the 90s, we did it at 600 dpi, just for the best resolution.
posted by Brandon Blatcher at 5:00 PM on March 18, 2009

One other thought is that you need to consider what is your purpose for enlargements.

I find that for rough architectural diagrams we often do the basic work at an 11x17 scale. If we use 300 ppi for this I feel more than covered for future enlargements. We have been known to print these files at up to 36x48 print size. In this case we are getting closer to 110 ppi. But, we also know that the average viewer will be standing farther away. There simply isn't enough detail to warrant approaching the board sized print and studying it from 10 inches away.

The largest print I ever did was an architectural rendering for a banner printed at about 8'x30' (if I remember correctly). I did that at 70 ppi. In retrospect I could have gone even lower resolution and saved some time, but folks did appreciate the 'count every brick' character that the image had when you got closer.

So, I think 600 is too much. One caveat is that the underlying line work in much of what I do is vector based. So I also strongly support looking into that avenue, which may both reduce file size and increase fidelity.
posted by meinvt at 5:13 PM on March 18, 2009

Long answer: It depends what you will do with it.

For web or video, it's overkill and you can get by with about 72dpi (although you lose the option to print in quality later.)
For four-color process, it's best practice, but you can make do with 300dpi.
For spot color printing (using inks of a solid color, usually Pantone colors) it's the minimum, and you should even up it to 1200.

The need to scale upwards (either physically or by jumping mediums) adds additional burdens, of course, since you have to consider what the final resolution will be.

As noted above, using vector art eliminates that problem, since it can scale indefinitely. You could make a billboard out of one of your cartoons. There's a definite learning curve, and you can easily end up with art that has a crazy amount of bezier points, which can make it difficult to print.

Short answer: It looks like you're doing fine.
posted by ChurchHatesTucker at 5:49 PM on March 18, 2009

If you're going to do vector-style art in a raster program, the higher the resolution the better.
posted by glider at 7:49 PM on March 18, 2009

I'll second rokusan - I had a look at your site too, and I never would have guessed that you were not working in vectors. You might enjoy getting ahold of Illustrator, the inexpensive Xara Xtreme, or even the free Inkscape (and maybe a Wacom tablet) and getting into vector drawing.

(Also, I got some chuckles from your cartoons & will visit again.)
posted by Tubes at 9:05 PM on March 18, 2009

You might find the Scan Tips website useful for demystifying your resolution questions. Wayne Fulton writes in an relaxed, informal manner and his site does a great job of explaining scanning basics.

Ultimately it would probably be more useful to you to learn to work in a vector-based program -- as stated previously -- because you'll be able to scale your work as large as you could ever want without a loss in quality, but learning a complicated new program when you're already good at what you do (especially if you're earning income from your work), probably isn't the best use of your time. In that case, you should consider Vector Magic to turn your art into scalable vectors and use Photoshop to touch up the color, if necessary, since you're already familiar with it.
posted by LuckySeven~ at 12:33 PM on March 19, 2009

Professional printer here, and has-been illustrator.
600 ppi is a good standard for any illustration that will be printed. 300 ppi is considered standard for photographs and other contone graphics that have a wide variety of tones - they generally don't have any crisp edges between areas of color, so the rosettes blur the features of the graphics enough that you won't see any stair-stepping or pixel edges. Your art, however, has large areas of flat color, so a 300 ppi resolution is more likely to show jagged edges, especially in the black plate.

But that only applies if you're looking to print your illustrations.

Now, 300 ppi is close to double the graphic detail of what almost any printer can print on paper
This only applies to low-end desktop printers. Laserprinters, even my goofy little HP 2100, print at 300 - 1200 ppi, and offset presses start at 2400 ppi.
posted by lekvar at 7:17 PM on March 19, 2009

This only applies to low-end desktop printers. Laserprinters, even my goofy little HP 2100, print at 300 - 1200 ppi, and offset presses start at 2400 ppi.

No, 2400 ppi is close to the resolution of 35mm film printers (and scanners) - a 6 megapixel image to/from 35mm film. You're suggesting that an entry-level offset press can print a cold-war-spy style microdot! On paper! Offset presses are 2400 dpi. This is part of the dpi/ppi mix-up I was talking about.
posted by -harlequin- at 5:06 PM on March 23, 2009

-harlequin-, this is what I do, day in and day out. It's my bread and butter. I've been doing this for a decade now. I have a full, working knowledge of the difference between dpi, lpi and ppi.

I stand by my previous statement.
posted by lekvar at 6:02 PM on March 23, 2009

Actually, having said that, I wonder if the points you're using for basis aren't different from the points I'm using for basis.
posted by lekvar at 6:26 PM on March 23, 2009

The claim that an offset press prints at 2400 ppi is erroneous. Let's say it doesn't - let's say it "only" prints at 1200 lpi, thus with a 2x spi you would want to print images at 2400 ppi to take maximal advantage of the print quality. This is not even close to a printer that can print 2400 ppi, and the lpi is... unusually high.

That said, yeah, I think I'm with you. I assume part of what you're getting at is that a
press will be able to cut the halftone dots along a vector, and so extra crispness can be made available because of the line-art style of these comics. And you contend that my suggestion (that 300 ppi is almost certainly beyond the printer's resolution) is not a sure thing. This might be true, but my thinking is that more than 200 lpi seems unlikely, and even 200 lpi seems more likely than 150 lpi (or less), all of which would make 300 ppi ideal, possibly even overkill, and 600 ppi a waste.

For filibuster: If your comics are being (or will be) printed as top-end print jobs on glossy paper, then 600 ppi may hold benefit. But those people suggesting moving to vector have the best printing solution from a technical standpoint, providing it's a transition with little disruption - because as the saying goes; if it isn't broken, don't fix it.

If you want to spur a complete consensus, you need to indicate what kind of print quality your work is being printed at (probably somewhere in the spectrum from newspaper to glossy magazine), and/or what kind of quality it might someday end up on, if you wanted a enlarged commemorative print or something. If your computer is being a pain at 600 ppi, you'd want a reason to be working at 600 ppi.
posted by -harlequin- at 9:53 PM on March 23, 2009

"and even 200 lpi seems more less likely than 150 lpi (or less), "
posted by -harlequin- at 9:54 PM on March 23, 2009

-harlequin-, what is your definition of a point? You seem to be using ppi and lpi interchangeably.
posted by lekvar at 12:16 AM on March 24, 2009

ppi = pixels per inch. In a bitmap image, a pixel is a coloured rectangle in the grid, and can be any colour of millions.

One line is the smallest dimension along which a printer creates enough half-tone information to create any arbitrary colour that is discrete and distinct from other arbitrary colours surrounding it. (I realize that that's a simplification, and that there are special cases, but I don't think those are relevant here)

If you can print with a resolution of significantly more lines per inch than there are pixels per inch being printed, the pixels will be distinct in the print. If your pixel resolution significantly exceeds your lpi - ie the traditional 1.5 to 2 times - then more information is being presented than can be discretely printed, the printer cannot render individual pixels, and the print quality will thus be limited by the lpi instead.
posted by -harlequin- at 2:10 AM on March 24, 2009

The claim that an offset press prints at 2400 ppi is erroneous.
Only if under certain circumstances, such as taking dot spread into account, which can be mitigated with coated paper stock, or by using a DTP imaging system rather than imaging plates via film. My direct-to-plate system can create up to 4000 discrete "dots*" per inch, and thus could, theoretically, image a 4000 pixel-per-inch image. Whatever metric we're using, pixels, dots, points, or lines, most modern laser-based printers have an operating resolution of 600 or higher "dots" per inch, per color; the screening may space these "dots" closer or further, depending on the halftone value, but that has bearing on the absolute resolution of the output device.

The higher the resolution of the source file (or better yet, as many have mentioned, vector-based art) the crisper the border between two distinct areas of color. This is especially important in the black channel.

*non-technical definition in this case; an area of plate that has been exposed by the imaging system's laser.

-harlequin-, are you a paper-pusher too?
posted by lekvar at 12:18 PM on March 24, 2009

My direct-to-plate system can create up to 4000 discrete "dots*" per inch, and thus could, theoretically, image a 4000 pixel-per-inch image.

No, a 4000 dpi offset printer absolutely cannot image a full-colour 4000 ppi image (to do that you would want something like a 30,000+ dpi printer in the real world, probably around 12,000 dpi in a perfect mathematical universe). I think you're assuming that a dot can represent a pixel even when printing colour, probably because in special other cases (see below) sometimes a dot could represent a pixel, but not here. Not by a long shot. Pixels are not equivalent to dots. You need a whole LOT of dots to match the visual information given by a pixel. This is because a printed dot is one blob of just one of the (CMYK) ink colours (or a spot tone), while a pixel is any of millions of colours, and that discrepancy puts many orders of magnitude more visual information in a pixel than in a dot.

In order to cross that divide and be able to print something that is "any of millions of colours" (ie a pixel) out of just four colours of ink, you need an area of paper large enough to mix your inks in the exact amounts needed to produce the precise colour desired. And to mix inks in exact amount you need an area large enough for you to print a whole bunch of dots, to create varying ink coverage densities, in each ink, such that from a distance, they appear blended and form the desired colour. The typical method for doing this is the halftone screens. Thus the resolution of the halftone screens marks the upper resolution at which pixels - elements that can depict any colour out of millions, not just the present ink colours - can be rendered.

If you are printing a one-bit image (ie all pixels are either full black, or full white. No grayscale, no colours), then you could use printed ink dots to represent pixels, and get the kind of dot-pixel equivalency and higher resolutions that you are talking about. I'm assuming that this sort of scenario is what you're thinking of, but it does not apply in cases like this when printing full-colour images - the screen resolution lpi is the limit on your print resolution.

In a colour raster image, a ppi more than twice the lpi is just wasted data, regardless of dpi.

-harlequin-, are you a paper-pusher too?

Not really. I'm a designer, and while I dabble on the side, and have a 4-year degree that included this sort of stuff, these days professionally I'm in the computer sphere.
posted by -harlequin- at 3:03 PM on March 24, 2009

Offset printing is achieved by translating the color data into 1-bit images; those 1-bit images are then imaged onto film or directly onto the plate. In CMYK printing, the four colors are then combined, on press, with a "fifth" color: the substrate. No, CMYK printing can't entirely replicate the color gamut of an RGB image, but that's not what's at issue here.

Yes, the linescreen breaks the 1-bit channel information up when there's screening into values commonly between 100 and 300 lpi, but the laser is capable of plotting up to 4000 points on a one-inch-by-one-inch grid. That is the basic definition of the printer's resolution.

If you are printing a one-bit image ... then you could use printed ink dots to represent pixels...

This is how all printing is done, imaging 1-bit images.

The artist's file is created with arbitrary values. The file is sent to a RIP server which takes the color information and converts it into 1-bit data sets; this can be CMYK or spot color, depending on how the file is created and processed, but the end result is still a 1-bit image. The RIP server determines the output resolution at this stage; I commonly set the output resolution to 2400 dpi because I work for a commercial printer. If I were printing fine art prints I would set the output resolution much higher.

The 2400 dpi 1-bit image is then screened, and this is where linescreen values become important. The linescreen value creates the halftone rosette that you see under a loupe when you look at a printed piece. The higer the linescreen, the smaller the dots that make up the halftone. Newspapers and web presses commonly use a linescreen of 85, promotional materials use 150-175 linescreen. But this is not the same as the imaging laser's output resolution, this is the way in which the rosette is formed, which is how a color gradation is created. If there are more than one colors used in a printed piece, the screens are angled in 15-degree increments in the RIP so that no two dots are imaged directly on top of each other.

In images where there is high variation in color values from one are to the next, say a photograph, 300 pixels per inch is perfectly acceptable because the screening that's necessary "breaks up" the image, and any pixels-per-inch beyond this, as you say, is more or less wasted data. But there's a significant difference between a photograph and an illustration, as the illustration, especially in the case of filibuster's illustrations, contain large areas of solid color. Those large areas don't have the same color variation from point to point, so the screening of the output device can't be relied upon to "break up" the pixel information.

This is especially critical when there's heavy use of black ink. Black ink on a white substrate shows the pixel edges more clearly than the other colors because of the high level of contrast. Most photographs, when printed, don't contain much data in the black channel because it tends to make the image muddy and indistinct; most of the contrast is handled by the other color channels. But illustrations tend to have a much higher use of black ink, so it become important to increase the resolution in order to keep the individual pixels from becoming apparent when the art is printed.

To see this in action, type something using a 12-point font in Photoshop at 300 pixels per inch, then do the same at 600 and finally 1200 pixels per inch. Save the resulting files as 1-bit tifs and print them using a decent-quality laser printer. Viewed though a loupe, the edges of the type in the 300 ppi example will be visibly jagged. The 600 ppi example will be less so, but there will still be rough edges. At 1200 ppi, the discrete points are small enough to form a curve along the outside of the letters, closely approximating the look of a vector. This goes back to elementary Algebra; the more points you have defining a curve, the soother the curve is.

I assure you that the printers I use, from my goofy little HP 2100 to my platesetter, are perfectly capable of imaging 1200 separate points on a curve within a square inch, and my devices are by no means unusual in this.
posted by lekvar at 5:31 PM on March 24, 2009

You think I'm denying that the printer can print thousands of separate dots along a curve within a square inch? You're still not following. It seems you're still not getting quite what is going on under the hood in your RIP server and exactly how that relates to printing colour pixels.

What your RIP does is look at a pixel in the image to be printed, and then represent the massive colour information in that pixel as a cluster of 1-bit pixels that individually do not have nearly enough states to depict that massive colour information (only two, instead of millions), and those 1-bit pixels are what what can be represented 1-to-1 by the dpi dots. Not the colour pixels in the image to be printed!

Thus, to print a image that is made up of 300x300 pixels, your RIP looks at that bitmap and from it creates CMYK 1-bit-bitmaps which have a resolution 1200x1200 pixels. Hence dpi in the thousands is absolutely required merely to be barely capable of printing ppi in the hundreds.

300x300 just inflated to 4x1200x1200, because the RIP only has 1-bit pixels with which to convey information, yet is being asked to convey the information contained in 24bit or 32bit pixels, so it needs a patch of dithered 1-bit pixels to produce the information that previously was being contained in just one, single, colour pixel.

So let it all fall into place.
- you know that full-colour bitmap images are printed with halftone screens.
- you know that lpi is a fraction of dpi (also, I've explained why)
- you know that lpi is the upper limit on print resolution when printing a colour bitmap image
- you know that best practice is a ppi not more than twice the lpi
- you know that this means that a 300ppi image resolution is the useful max for a 200 lpi printer
- you know that any 200 lpi printer has a colossal dpi in the thousands, and can print thousands of points along a curve inside a square inch.
- therefore you also know that being able to print thousands of dots per inch does not mean that you can print thousands of colour pixels per inch, or full print a colour bitmap at a ppi even remotely close to the dpi.

That is the point I'm making - a 2400 dpi printer CANNOT print a colour bitmap image at anything close 2400 ppi. It will max out at something like 300 ppi, nowhere near close to 2400 ppi. It's maths. It's physics. You either need to print a non-colour image, or use a spot tone per colour, or need to move to an entirely different printing technology (such as film, or dye-sublimation) to get dpi output to resemble ppi output.

With CMYK (like offset), being able to print thousands of dots (or 1-bit pixels) per inch is necessary to be able to print hundreds of pixels per inch.

I've explained the mechanics of why this is the case, and I really think that there is enough explanation and information in my posts to clarify the matter and draw a solid line along the massive difference between pixels and dots. I think you have an established understanding that works quite well but does have a hole that has been patched over, and that this understanding is too entrenched to easily accept any correction without an awful lot more convincing than is worth my time.

So I'm done here. Sorry.
posted by -harlequin- at 7:27 PM on March 24, 2009

- you know that full-colour bitmap images are printed with halftone screens.

True.

- you know that lpi is a fraction of dpi (also, I've explained why)

True.

- you know that lpi is the upper limit on print resolution when printing a colour bitmap image

Only partially true, and that conditional truth is largely dependent on the nature of the image to be output.

- you know that best practice is a ppi not more than twice the lpi

Once again, only partially, and conditionally true. Best practice for photographs, not for illustrations.

- you know that this means that a 300ppi image resolution is the useful max for a 200 lpi printer

Again, only for contone images with high variation in color. Not applicable for illustrations.

- you know that any 200 lpi printer has a colossal dpi in the thousands, and can print thousands of points along a curve inside a square inch.

Lines per inch is just a screening method, not an output resolution. The two are linked, but functionally separate.

- therefore you also know that being able to print thousands of dots per inch does not mean that you can print thousands of colour pixels per inch, or full print a colour bitmap at a ppi even remotely close to the dpi.

I'm sorry, but this is untrue, and I think it stems from a misunderstanding about the difference between lpi and resolution.

Thanks for the interaction though, it's forced me to go back and re-examine some assumptions I haven't thought about for some time.
posted by lekvar at 8:48 PM on March 24, 2009

You wrote me in email, so I've revisited. I do not feel you have been adversarial, more like a locomotive - powerful, but running on rails. I am frustrated by this failing of text to convey to you what you are glossing over, and extremely frustrated by how much of my time I have wasted in this thread.

If I sat you down with coloured pens, make up some source-image pixels, and we walk manually through each step of the RIP and then on to printing, making you into a human computer and a printer, we would be in complete accord - you would demonstrably fail to accurately reproduce the image if it were not so. I keep thinking that we must already be in accord, and perhaps we just have different visualisations, but then you say things that seem to directly preclude it, so I go back and forth. I know you will not accept it, but the conclusion I keep coming back to is that there is a hole in your understanding that has been papered over and built on top of for so long that I lack the time to make it visible to you by mere text, and there is no point in me trying to unravel this.

I'm well aware of various the methods you keep bringing up by which under various constraints you can print some graphic details at the resolution of the dpi, I even briefly referred to them before you brought them up, to indicate that they're not relevant to what I've been trying to get at, attempting to pre-empt all these tangents.

So at this point, I'd be happy for us to agree simply on the utter basics - that an offset press attempting to reproduce a full colour bitmap, such as a photograph (ie. not line art, not limited use of colour, etc etc) requires a substantially higher dpi than the ppi to have any chance of rendering the detail of the source image. Seen here is a super simplified example where line art, spot tones, and other methods are either not helpful or else way more trouble than they're worth, so for every pixel of the source image, the printer needs to take up enough area on the paper for 36 printed elements, meaning that it prints pixels at 1/6th its dpi, precluding the ability to deliver ppi at anything approaching the resolution of the dpi without greatly degrading the image detail. We're together on this, correct?

So to further state the obvious, a voltage converter allows you to turn amps in voltage and vice versa - the trade-off is that to raise your amp output, you must lower the voltage output, and vice-versa.

A stick on a fulcrum allows you to create leverage - you can turn a weak but large motion into a strong but short motion. The trade-off is that to raise your strength, you must sacrifice some distance of travel.

An offset printer allows you to trade resolution for colour. You can turn high-resolution-but-few-colours (high elements-per-inch, but only a few ink colours), into many-colours, but you must sacrifice resolution to do so (elements any colour, but low elements-per-inch). dpi > ppi

I'm well aware (and have mentioned) that the comic art is well suited to various other means of taking better advantage of CMYK printer resolution.
posted by -harlequin- at 3:02 AM on March 25, 2009

I thought you were done here :)

Seriously though, if you'd like to continue this conversation, we should probably take it it MeMail rather than longboating this thread. If filibuster hasn't made up his mind on the subject by now, it's likely because we've bored him into a coma.
posted by lekvar at 11:59 AM on March 25, 2009