Question about "scale to fit media"

[dtrayers]

I have a question about the "scale to fit media" check box in Print with Preview screen in Photoshop.

For example, I have a 2400x3000 file which will print at 300dpi for an 8x10.  If I want a smaller size print, say 4x5, and print with preview, it will show the image beyond the size of the paper.  I can check "scale to fit media" and then the bounding box will fall inside the print area.  Effectively, I think that this is printing at 600 dpi, which is way more than necessary.

The question is, would I get better quality down-sampling using bicubic sharper to a 1200x1500 file, or will I get the same quality as leaving it at 2400x3000 and selecting "scale to fit"?

I realize that upsizing should always be done instead of the 'scale to fit', but since downsizing is throwing out information, is 'scale to fit' ok?

Thanks.

[Stephen Best]

The question is, would I get better quality down-sampling using bicubic sharper to a 1200x1500 file, or will I get the same quality as leaving it at 2400x3000 and selecting "scale to fit"?
[a href=\"index.php?act=findpost&pid=77208\"][{POST_SNAPBACK}][/a]

You'll get optimum results by down-sampling to an integral factor of the printer resolution (say 360ppi for Epson), output sharpening at that resolution and printing with *no* resizing. Resampling (which is what "Scale to Fit" in Print with Preview does) after output sharpening should always be avoided. Whether the results are visibly better though will depend on the overall image quality, degree of down-sampling, paper surface, how critical your eye is etc.

[bjanes]

You'll get optimum results by down-sampling to an integral factor of the printer resolution (say 360ppi for Epson), output sharpening at that resolution and printing with *no* resizing. Resampling (which is what "Scale to Fit" in Print with Preview does) after output sharpening should always be avoided. Whether the results are visibly better though will depend on the overall image quality, degree of down-sampling, paper surface, how critical your eye is etc.
[a href=\"index.php?act=findpost&pid=77210\"][{POST_SNAPBACK}][/a]

Steven's answer is correct, but in the case of the Epson printers, as Bruce Fraser explains in his recent book Real World Sharpening, the native resolution of the printer (multiples of 360) refers to the addressable resolution of the printer, not placement of dots, since dot matrix printers use error diffusion dither.

The advantage of printing at multiples of 360 is apparent with line pair targets, but is usually lost with real world images. Bruce usually sends the image to the printer at the native capture resolution without resampling, but does state that resolutions of > 720 ppi may actually degrade the results. He finds that the printer can usefully make use of resolution up to 480 ppi.

It is important to sharpen at the final output resolution or else your sharpening halos may be downsampled out of existance.

Bill

[Lisa Nikodym]

The advantage of printing at multiples of 360 is apparent with line pair targets, but is usually lost with real world images.

Not entirely true.  I did this experiment with my Epson 2200 and a piece of a typical landscape photo several years ago (and reported the results here), and found that 240 ppi and 360 ppi looked identical to me, while 180 ppi was slightly worse, but even worse than 180 ppi was the original resolution of the image, which was about 220 ppi.  You had to look very closely to see the differences between them all, but it was there.  A hobbyist probably wouldn't notice, but if you're looking for gallery-grade quality there difference is there.

The comment is never resizing after sharpening is right on.

Lisa

[Stephen Best]

Bruce usually sends the image to the printer at the native capture resolution without resampling, but does state that resolutions of > 720 ppi may actually degrade the results. He finds that the printer can usefully make use of resolution up to 480 ppi.
[a href=\"index.php?act=findpost&pid=77291\"][{POST_SNAPBACK}][/a]

I'll have to see what Bruce says in his latest book (which I have but haven't got around to yet) but my understanding is that resolutions over 360ppi would only be usable on the low end Epson's (2400 etc) or when "Finest Detail" is set on the x800 series. It also helps to use a resolution for which you have an output sharpening workflow. Certainly, the results at 360ppi are better to my eye than the oft recommended 240ppi. Most people print big it seems because that's the size their printer prints, rather than because the image would benefit from it ... IMHO.

[Stephen Best]

Not entirely true.  I did this experiment with my Epson 2200 and a piece of a typical landscape photo several years ago (and reported the results here), and found that 240 ppi and 360 ppi looked identical to me, while 180 ppi was slightly worse, but even worse than 180 ppi was the original resolution of the image, which was about 220 ppi.  You had to look very closely to see the differences between them all, but it was there.  A hobbyist probably wouldn't notice, but if you're looking for gallery-grade quality there difference is there.
[a href=\"index.php?act=findpost&pid=77305\"][{POST_SNAPBACK}][/a]

An A4 or 8x10 print requires 10 megapixels at 360ppi. If you're starting with less data than this, you won't see a difference.

[Stephen Best]

In the interests of science, I just did my own tests. I started with a 400MB+ scanned image which would have been over 1000ppi at the print size (7.5x9.5). I then chopped it in half and used this to downsample (with Bicubic Sharper) to 240, 360 and 480ppi respectively. I normally use my own sharpening workflow but for this test used Bruce's Photokit Sharpener (which I presume he would be using) and the Inkjet 240 Glossy, 360 Glossy, 480 Glossy and 480 Glossy Fine presets. I printed all with my Epson 4800 on Ilford Smooth Pearl at 2880dpi High Speed (I'd done an alignment procedure on the printer fairly recently). For the 480ppi files, I also turned on Finest Detail in the driver.

Examining the output, there's a decrease in local contrast and a smoothness to the flatter areas as the resolution increases. At 240ppi, the output just looks coarse and details disappear under a loupe. At 360ppi (and above) it's harder to pick differences with the naked eye, other than an impression of smoothness. The 480 Glossy Fine output is distinctly smoother (and maybe softer). How much all this has to do with the sharpening parameters is anyone's guess. Since it makes no difference to me, I'll give the 480ppi workflow a try with a range of images to see if it gives me better results than I've been getting.

Note that all of the above applies to captured data at optical resolution with no interpolation (Bayer or otherwise) or upsampling.

[Hermie]

> The advantage of printing at multiples of 360 is apparent with line pair targets, but is usually lost with real world images.

Most of the time you won't be able to see a difference between 300ppi and 360ppi.

Harald Johnson writes in his "Mastering Digital Printing" that Epson recommend 300-360ppi as their current magic number.

At 300ppi there were artifacts in the very fine fabric in the image below. Upsampled to 360ppi, the print was perfect.



Herman

[bjanes]

Examining the output, there's a decrease in local contrast and a smoothness to the flatter areas as the resolution increases. At 240ppi, the output just looks coarse and details disappear under a loupe. At 360ppi (and above) it's harder to pick differences with the naked eye, other than an impression of smoothness. The 480 Glossy Fine output is distinctly smoother (and maybe softer). How much all this has to do with the sharpening parameters is anyone's guess. Since it makes no difference to me, I'll give the 480ppi workflow a try with a range of images to see if it gives me better results than I've been getting.

[{POST_SNAPBACK}][/a]

In interesting post, Stephen. Rags Gardner has published some interesting results on his web site. According to his experiments, the native resolution of the Epson 2200 is 288 ppi; he was using the 1440 ppi setting in the driver. For an integral multiple of native resolution, 576 ppi could be considered. Look in the technology section for printing.

[a href=\"http://www.rags-int-inc.com/]http://www.rags-int-inc.com/[/url]

It would be interesting to do similar tests with your higher end printer.

Bill

[TimothyFarrar]

Wow, what an amazing topic.

Bill, thanks for the link!

This will definatly make me rethink printing on my Epson 7800.

Since the 7800 has a maximum rated dpi of 2880 x 1440, I would guess that 288 ppi probably is the correct setting for it also (will have to test to be sure).

[Stephen Best]

Since the 7800 has a maximum rated dpi of 2880 x 1440, I would guess that 288 ppi probably is the correct setting for it also (will have to test to be sure).
[a href=\"index.php?act=findpost&pid=77403\"][{POST_SNAPBACK}][/a]

This is really a function of the driver, but the Programming Guide for the 4800 (and presumably 7800/9800) only talks about 360 and 720ppi. I would guess that 288ppi is a better choice than 300ppi (for any Epson printer). The reason I've standardized on 360ppi is because, as I understand it, this is a straight path through the driver. But you really need to do eye-ball tests to see what works best for you and your images. The ideal is to use something like Qimage which tightly integrates the sharpening and output stream generation (but I'm on a Mac).