Up Res or Down DPI?

[lobalobo]

Shooting with an inexpensive, 6mp camera (Fuji s6000fd), processing with free software (Picasa 3), and printing

small (4"x6") on an inexpensive printer (HP 7200) and decent but not expensive paper (HP Glossy Premium), I am

getting what is to me shockingly good images (by which I mean technically good given my limited skills as a

photographer). With these results, I'm planning to start printing 8"x10". When I began to consider how to do

this, I noticed that the default setting on the printer is 600 dpi, which I'm sure partially explains the

excellent quality of the 4x6 prints even viewed up close. It occurs to me that 8x10 photos are typically viewed

up close too, because they are too small to view from a significant distance so viewers walk right up to them

even if hung on a wall. This got me wondering how to make the 8x10 prints. Do I down the dpi (to roughly 275) or

up the resolution (a little more than doubling the size)? Of course I know that the obvious answer is to

experiment, and I will. But I haven't been able to so far as the low-budget work-flow described above is not

designed for anyone to tinker with and I'm still searching for the dpi adjustment. In the meantime, I thought

I'd ask what experience others have had as this must be a common issue. (And if it's addressed in a post or posts

that I missed, I apologize.) Thanks.

[Former P.N Member]

I've printed very good 8x10s from a 3MP Canon D30. I'd say that either way you decide to go you should have more

than enough data to get a decent print.

[ronald_moravec1]

Not an expert inkjet printer, but from what I have read Epsons like 360 dpi and all others like 300.

 

600 is simply not used by any inkjet printer and the image is downrezed with the printer software.

 

I would change the rez to 300 and then resize as necessary. Some say a decent print can be made with 240 or 180, not great but decent.

 

Try various combinations to make an 8x10 and crop the center or important part to 4x6 and make a print and see what the differences are. Make sure you sharpen the prints at final size. The Bayer pattern in a digi cam capture produces a soft image and camera sharpening can not be correct for every size because it does not know what size the print will be. Now I opened another can of worms for you. Sorry.

[lobalobo]

Thanks for the responses. As for 600 dpi being down rezed, I wonder then why the printer readout is 600 dpi, which is nothing but confusing. What's the point? As for sharpening, I agree with the prior post, and almost always shoot RAW anyway (the Fuji s6000fd is a digicam, but a sophisticated one), so I'll be sure to resize first, then sharpen.

 

One last question. I imagine that the printer will up rez on its own if commanded to print more dots than it has pixels. I wonder how consumer printers such as my HP 7280 do at this as compared, say, to Irfanview or (Bearded Frog) Enlarger Pro. Anyway, I will experiment, and will report back for others, but remain curious in the meantime.

 

Thanks again.

[howard_owen]

I've gotten good results by simply changing the size and letting the dpi fall where it will. Printers are very good at doing their own resampling, or at least Epsons seem to be.

[Ed_Ingold]

The printer will adjust the resolution to suit its own needs. However it will not change the appearance of the image. If pixelation is visible in the image (using appropriate magnification considering the screen size) at the size to be printed, it will be visible in the print as well - the pixels will be faithfully delineated. Conversely, if the pixels are too small to be seen in the image (again at an appropriate magnification) then they won't be seen in the print.

 

The practical effect is simple. If the image resolution is too low, upsample before printing. If it is too high (significantly greater than 300 ppi) you can leave it alone. However downsampling can greatly reduce the file size, which takes less time to transmit and print.

[Former P.N Member]

I should also mention one of the best pieces of software you can find for printing. <a

href="http://www.ddisoftware.com/qimage/">QImage</a> is great as you just give it the image you want to print,

tell it the size you want and it does everything else. All you have to do is crop the image if the aspect ratio

is different than what you're printing. (Keep in mind that you still have to manage monitor calibration and

profiling as well as color spaces but you have to do that no matter what program you're printing with.)

<p>

The interface is a bit different than 'standard' windows programs so there is a learning curve but the results

are worth it - IMO. I've been using it for years and it's the best photography related money I've spent.

[lobalobo]

Does Qimage produce an up-rez output file? That is, imagine I want to up rez an image for a print larger than my printer can handle and then have the file printed for me at a commercial lab. Can I tell Q-image to produce the suitable file even though it won't be printing? Thanks to all; I will experiment and report back.

[Brad_]

The interpolating resampling filter within the Epson firmware works in concert with their proprietary screening process;

yielding superior image quality. Decoupling that synergy by resampling in your image editing software still works fine;

but better results are had by letting the printer handle it.

[beaglefur]

The term DPI (dots per inch) and PPI (pixels per inch) are often used interchangably, but they're not the same. Printers need several droplets of ink per pixel to create smooth tones which is why images at 300 pixels per inch are printed at 600-2800 dots per inch. Generally speaking if you are between 200 and 300 pixels per inch you are in good shape.

[bruce_watson1]

<i>As for 600 dpi being down rezed, I wonder then why the printer readout is 600 dpi, which is nothing but

confusing. What's the point?</i>

<p>

Perhaps you are confusing image resolution (measured in pixels per inch, or ppi) with the printer's internal use

of ink which basically describes how many ink dots the printer will use to represent each pixel (measured in dots

per inch, or dpi). These are two completely independent settings - one you make in your image editor, one you

make in your printer driver.

<p>

HP printers seem to have a "native" image resolution of 300 ppi. What that means will of course vary somewhat

from manufacturer to manufacturer, but it usually means that the driver will take whatever image resolution you

give it and up or down sample it as required to get to its native resolution. Why? Because it makes the math

much easier -- gives the driver a known base from which to build it's dither pattern, which it uses to decide

where the ink dots go to describe the pixels of your image.

<p>

That much of an explanation should get you started. It's certainly not very complete, but about all I can do

within the limits of a posting like this.

<p>

What to do then? The easiest thing to do is to send your file to the printer sized to the image size you want, at

whatever ppi that turns out to be. The driver itself will do the up sizing, and most printer drivers do a very

good job of this. Then, tell the driver to print at whatever its high quality photo setting is -- this will

result in something like 1200 dpi for an HP printer.

<p>

That will get you good results and is a good place to start. From there you can do some searching on the web and

learn more at your leisure.

[Former P.N Member]

<i>Does Qimage produce an up-rez output file? That is, imagine I want to up rez an image for a print larger than

my printer can handle and then have the file printed for me at a commercial lab. Can I tell Q-image to produce

the suitable file even though it won't be printing?</i>

<p>

QImage has a 'print to file' option that will do exactly what you describe. I've prepared 16x20 prints using

that method.

<p>

<i>The interpolating resampling filter within the Epson firmware works in concert with their proprietary

screening process; yielding superior image quality. Decoupling that synergy by resampling in your image editing

software still works fine; but better results are had by letting the printer handle it.</i>

<p>

The best quality prints I've produced have been by taking the original file (be it JPG, TIFF, or RAW) and

printing it through QImage. I do the cropping in QImage after doing image editing in PS CS3 that way I only have

one 'master' file to keep track of. QImage will do whatever resample processing is necessary as well as the

final USM prior to printing. It consistently does a much better job than I can via PS.

[lobalobo]

Thanks to all. In response to David and Bruce, I had read a number of tutorials on the subject of DPI. These tutorials typically start with the statement that a good print requires something in the neighborhood of 300 DPI, and I had assumed they meant by this the number of ink dots on the page; they then offer a translation to pixels, offering as an example, say, that a 1,200 pixel side of an image to be printed at 300DPI will be 4 inches long. (Some online services have the same explanation, MyPublisher, e.g., when it gives guidelines for pixel dimensions and print size.) Do I take it from your responses that what these tutorials call DPI is really PPI after all? If this is the case, and one wanted to adjust the printer's use of pixels, what would be the adjustment, to the number of pixels per inch? If that's true, then I do understand why the only thing anyone would ever have to tell a printer is the size of the print, as the number of pixels per inch would be determined by the dimensions of the print. Do I now have this right? Thanks.

[roger_smith4]

Yes, just tell the printer the size you want it printed at, feed it some pixels and so long as the dpi/ppi is within the realm of reasonableness for the viewing distance you'll get a decent print.

[lobalobo]

Right; makes perfect sense now. DPI is a printer characteristic; PPI is a print characteristic. So my original post should have asked "Up Res or Down PPI?" not "DPI". In an case, as I mentioned, I'm going to try each but I bet I won't be able to tell the difference on my inexpensive equipment. Thanks to all.

[steve_swinehart]

"The interpolating resampling filter within the Epson firmware works in concert with their proprietary screening process; yielding superior image quality. Decoupling that synergy by resampling in your image editing software still works fine; but better results are had by letting the printer handle it."

 

There is no interpolating "filter" in the Epson software. The software is a conversion process in which it takes the incoming file and formats the data at 720dpi for output. This is not exactly the same as a smart interpolation program - yes, it will upsize the image to fit the print but it does not do it with the same finesse as other programs. You can get better output from a RIP specifically because the RIP totally bypasses the printer driver.

 

Qimage is like the printing portion of a RIP. If you're doing large scale prints that require a great deal of upsizing - Qimage will work better than sending the data through the Epson driver. The driver is made to position the head, output the ink, and advance the paper. It is not good at upsizing a file.

 

Qimage formats the data at 720 dpi and sends it to the Epson printer driver. This gets the driver totally out of image resampling - and with a large image you will get far better results.

[steve_swinehart]

"That is, imagine I want to up rez an image for a print larger than my printer can handle and then have the file printed for me at a commercial lab."

 

This is NOT how you want to handle the workflow. You need to contact the lab you're going to use. If you want the image printed using a LightJet or Durst (Lambda or Theta) you do not upsize the image. They should have purchased the interpolation software as part of the machine. What they will do is set the machine for the required print size and send it your file. The machine will interpolate the image to its native format for the output size.

 

If they're using an inkjet printer, then they're probably are using a RIP. The RIP will interpolate the image to the final format as it is bypassing the printer driver and directly controlling the printer. Before you assume you need to upsize the file - you need to contact the service you anticipate using to see what they want. My bet is they will just ask for a file at its native resolution.

[lobalobo]

Thanks, Steve. Your second post makes perfect sense. I do have a question about your first one, though.

 

You say: "There is no interpolating 'filter' in the Epson software. The software is a conversion process in which it takes the incoming file and formats the data at 720dpi for output."

 

I thought what had learned from this thread is that DPI (as opposed to PPI) is a characteristic of the printer, and independent of the number of pixels in the image. So, if I have a native resolution that will produce only, say, 200 PPI and I prefer 300 PPI, then I have to up rez the image (or have software do it for me automatically) but whether I up rez or not, if the DPI of the printer is, say 720, then it will be 720 regardless of the PPI. Do I have this wrong? If not, what do you mean by "formats the data at 720dpi for output"? Thanks.

[steve_swinehart]

Thanks, Steve. Your second post makes perfect sense. I do have a question about your first one, though.

 

You say: "There is no interpolating 'filter' in the Epson software. The software is a conversion process in which it takes the incoming file and formats the data at 720dpi for output."

 

"I thought what had learned from this thread is that DPI (as opposed to PPI) is a characteristic of the printer, and independent of the number of pixels in the image."

 

That is correct. The current Epson printers have "resolutions" of 1440 or 2880 - or 2x720 and 4x720. The printer really works at 720 dpi - but uses a dithering routine in laying down the ink that increases the print resolution (not image resolution) to either 1440 or 2880 through doubling or quadrupling the ink dots in an area. But, it is still working from its native resolution base of 720 dpi. So, no matter what file you send to the printer - 100 ppi or 360ppi or 250.5 ppi - it will get translated to 720 dpi that is then dithered upon output to either 1440 or 2880 dpi.

 

That does not equate to higher image resolution. You have to think of this in two ways. In the image file you have ppi - and if you were to spread 100 ppi over a 20x30 image - you would see large boxes being printed on the image - those are the pixels - and it would be a very low resolution image.

 

Regardless of the fact that it's a low resolution image and you can see the pixels - it is still being printed by the printer at 720dpi (dithered up to 1440 or 2880 - see below). So, you would have an extremely low resolution image because the small file size of only 100 ppi is being spread over a very large area being printed at either 1440dpi or 2880dpi.

 

 

"So, if I have a native resolution that will produce only, say, 200 PPI and I prefer 300 PPI, then I have to up rez the image (or have software do it for me automatically) but whether I up rez or not, if the DPI of the printer is, say 720, then it will be 720 regardless of the PPI. Do I have this wrong? If not, what do you mean by "formats the data at 720dpi for output"? Thanks."

 

You're confusing the PPI of the file with what is actually being printed. 720 dpi is the base resolution of the printer. The printer driver will take the data (ppi) that is sent to it and change the data to 720 dpi to match the print head output. It's not increasing the resolution, it's only making the data compatible with how many little droplets of ink per square inch it can make (720 is the base rate). However, it then applies dithering algorithms to the 720 to increase the resolution (1440 or 2880) in effect the software is taking the 720 and putting 2 drops of ink in an area instead of 1 (1440) or 4 drops of ink (2880) instead of 1. But, it's a little more complicated than that - if it just layed down 2 or 4 drops on the space where there should be 1 - the area would just be more ink saturated - not higher resolution.

 

So, what the software is doing is figuring out the size of the droplets (how many pico-liters per drop) and the sequence it needs to drop them in an area. The head moves horizontally across the paper. As it moves across the paper it will make multiple smaller droplets of a color in an area rather than one large drop - you now have finer resolution as the drops are smaller - but in reality it is just multiplying the 720. Edges become sharper because they are more finely defined. The color saturation stays the same because 2 or 4 smaller droplets equal the same amount of ink as 1 larger droplet.

 

Now, the printer driver does have the ability to upsize an incoming file to match the size of the image being printed. However, the methods used by the printer driver are not as sophisticated as other types of interpolation software. On small prints (13x19 or smaller) - you won't really see a difference if you upsize in PS, a secondary interpolation program or just send the print to the printer.

 

When you get into large scale prints (I print 20x30) - there is a real difference in just sending the file to the printer and using software that uses more sophisticated methods of interpolation. IF the printer driver is so great at interpolation - one would have to question why Epson bundles the Colorburst RIP as part of the 7800 - 9800 series printers. Could it be that a specialized piece of software is better at controlling the printer and interpolation than the free printer driver?

 

Does this make sense to you?

 

[steve_swinehart]

Let me see if I can do this a little better...

 

I really meant to say a 100 pixel image NOT a 100ppi image. What I was trying to get at is that you could take 100 pixels and distribute them over a 20x30 inch print. You would have very large pixels that would show up as a color boxes. The printer will take the 100 pixel image and print it at 360, 720, 1440 or 2880 depending upon what you select in the printer control for the "Print Quality" - and the type of print media selected.

 

If you look at the Epson printer control, and select "Advanced" - it will take to a control menu that has a "Print Quality" control. From that control you can select: Normal - 360dpi; Fine - 720 dpi; SuperFine - 1440 dpi; and SuperPhoto 2880 dpi (the choice of Enhanced Matte under "Media Type" will allow you to see all of the selections).

 

With some paper choices you will be restricted to 360 (e.g. plain paper) and with other paper choices you will be restricted to 1440 or 2880 (e.g. Ultrasmooth Fine Art). These are built-in working parameters so you can't over ink or under ink specific paper types.

 

The "Print Quality" selected tells the printer driver how to dither the ink droplets - that is, the number and size of droplets per square inch. At 360 it is dithering down (dividing 720 by 2), at 720 working directly at the native resolution, and at 1440 and 2880 multiplying the 720 and making the droplets smaller.

 

I hope that makes the explanation a bit simpler and clearer that the 720dpi is totally independent of the image pixel count.

[lobalobo]

This is all perfectly clear, Steve. Thanks so much for taking the time.