Print size vs resolution

[Biomech]

It would be more accurate to say it has a "maximum resolution" just like your monitor.

Ah, indeed, that makes more sense

[ravncat]

Originally Posted by sk66

Don't confuse the "printer resolution" with what you "should use"...it's the minimum resolution you should use to get the best image quality from the printer.

But if you are going above it, you should use a multiple to get rid of nearest neighbor artifacts (At least with high frequency detail). I'm finding with my prints that I get better results especially in areas with lots of fine detail (not so much with images that are mostly low frequency detail)

[sk66]

Originally Posted by ravncat

But if you are going above it, you should use a multiple to get rid of nearest neighbor artifacts (At least with high frequency detail). I'm finding with my prints that I get better results especially in areas with lots of fine detail (not so much with images that are mostly low frequency detail)

Depends I guess. Are you talking about increasing the DPI to reduce the print size while leaving image size alone? (I was).

The only way to change print size is to "resample". Usually the printer driver will do a better job on the downres than you will. Nearest neighbor is the most basic algorithm and I don't think current drivers/software use it for downsizing. You could use nearest neighbor to increase "sharpness" at the cost of tonality I suppose.

For continuous tone and photo sensitive printers I don't think it matters at all... I've never reduced the size of an image (in pixels) for print, but I send my stuff out and print on photo paper 95% of the time.

[ravncat]

Originally Posted by sk66

Depends I guess. Are you talking about increasing the DPI to reduce the print size while leaving image size alone? (I was). .

I'm talking about feeding the data to (in this case my epson) the printer at 360 PPI. - Without re-sampling the image. The image size may end up smaller than your print size. I can let The printer then handle the up-scaling before output. This way the file is "parsed" better. (So maybe we mean the same thing? ppi and dpi should not be used interchangeably. (dpi itself isn't the best term since it has more than one meaning itself - both "micro dots" per inch and "cells" per inch (or even dots, when going back to the old dot matrix printers )

[SwissJon]

So I'm curious. Let's take an image, 2880px by 4320px and tell Photoshop to resize it to 720dpi, that would result in a 4"x6". Resampling the image doesn't take place in Photoshop because you're using a multiple of 720. Print this out, and then compare it an image at PS default resolution printed directly to the print driver and in the driver, specify that you want to print the image 4"x6". Which one (if any) produces the better image?

Ok, my guess about the above is that you're going to get the same, or not noticably different IQ on the printed image.

Ok, now we come to the more testing time. Let's take an image at my camera's native size. I shoot with a D7000, that gives me images that are 4928x3264 pixels. I'm going to crop it to 6x4 but preserve the maximum information. So I crop the canvas to 4896x3264px. All lovely, but that doesn't give me exact multiples of 720.

So the only options are for me to resample the image in PS or letting the print driver and printer take the strain.

Resampling in PS gives me an image of 4320x2880 px at 720pixels/inch to give me a 6" by 4" image using whichever algorythm gives the best visible results.. (Bicubic sharper or Bicubic smooth seems the best two options here) and then print out the resulting image using the standard print driver.

Alternatively, taking the image at 4869x3264 px and then tell the printer to print me a 6"x4" image will let the print driver take the strain. Resampling will still take place, but this time the printer does the work, taking into account the algorythms set by Epson rather than Adobe.

There are two arguments here as to which will give the best result, I've no idea which is correct, or if it's dependant on the image..

Leaving PS to do the resampling allows PS to use the Adobe algorythms. You get to see the image as it will be printed outprior to the printout, you get several algorythms to choose from, and it has access to an extremely powerful processor. However it has no idea of the layout of the jets on the printer so it doesn't understand cells.

Pushing the raw data out to the printer driver and allowing it to resample on the fly gives the driver the chance to use it's "knowledge" of the way the printer prints, and should give a better image. However there aren't so many options, and you can't see the image until it's printed out.

I've no idea which is correct, what are your thoughts?

[sk66]

Originally Posted by ravncat

I'm talking about feeding the data to (in this case my epson) the printer at 360 PPI. - Without re-sampling the image. The image size may end up smaller than your print size. I let The printer then handle the up-scaling before output. This way the file is "parsed" better. (So maybe we mean the same thing? ppi and dpi should not be used interchangeably. (dpi itself isn't the best term since it has more than one meaning itself - both "micro dots" per inch and "cells" per inch (or even dots, when going back to the old dot matrix printers )

Ah, yes upscaling should always be done in post, and then resharpen before printing. If you let the print driver upsample for you even multiples give the best results (but I still wouldn't recommend it).

I was talking specifically about downscaling...i.e. feeding your epson the image at 500 ppi/dpi to get a smaller print than "native".

[sk66]

Originally Posted by sk66

Ah, yes upscaling should always be done in post, and then resharpen before printing. If you let the print driver upsample for you even multiples give the best results (but I still wouldn't recommend it).

I was talking specifically about downscaling...i.e. feeding your epson the image at 500 ppi/dpi to get a smaller print than "native".

Jon, in both of your examples you are sending data at a larger size than the requested output. In almost every case just sending the file and letting the printer generate the output will give equal/better results. But there is no absolute "right answer".

Think of it this way. You don't reduce the color (bit) levels for viewing on different monitors. And you don't resample your images to 72ppi for viewing on your monitor do you? You could without effect. But if you also downsized that image so that the print size remained the same at 72dpi you would be discarding a lot of data. Generally that extra data is interpolated during printing and results in smoother/better tonality. There are rare occasions where you would rather have "harder edges" and manually downsampling with bicubic sharper or nearest neighbor could give better results.

IMO, keep the data and let the printer sort it (for downsizing). If that does not generate the results you want make a NEW COPY and downsize it manually to see if you can get better results.

[robert lamont]

Can i throw a curve ball into the equation. If i were to scan negatives ( say 6x7) would the resolution i pick to scan the negatives affect the outcome of the print? that is should i pick a resolution that is a multiple of the printer's native resolution? Print size would be ( in my case) 30 x 40 in.

[sk66]

Usually the problem with scanning negatives is the resolution is too high for many applications (web based) and is actually excessive for a typical print (unless you want 30x40 in).

This is one of the times where downsampling often makes sense....to reduce file size. If you scan at any resolution greater than can be viewed/printed then you are just making a larger file size. Generally recommended for scanning a 35mm for print is 4000dpi which will print up to 16x20 with excellent results.

At 6x7 you could reduce the scan resolution significantly, probably all the way to printer resolution and get excellent results for typical prints. But if you are not impeded by the processing loads and file storage you might as well scan as high as you can.

"Multiples of printer resolution" really only factor in when you are letting the print driver upscale an image for you.... "even multiples" give better results when resampling an image yourself. But you always have to compromise somewhere.

Once again, we are into areas where I am not a technical expert..... I'm just giving the best information I as I know it.

[ravncat]

Originally Posted by robert lamont

would the resolution i pick to scan the negatives affect the outcome of the print? that is should i pick a resolution that is a multiple of the printer's native resolution? Print size would be ( in my case) 30 x 40 in.

Sk66 has a great answer for you. It really comes down to the file size - pixel by pixel is the factor that determines how big you "can" print. In a sense that "resolution" can be whatever you want. How big a file do you want? and At what point are you not getting meaningful detail from your scan?. Sk66's guidelines for scanning are great. You can always rescale the image too. Since you can scan pretty large, you may as well scan down-sample it somewhere, because this can get rid of some of the noise from your scan.

Originally Posted by sk66

"Multiples of printer resolution" really only factor in when you are letting the print driver upscale an image for you.... "even multiples" give better results when re-sampling an image yourself.

Coming from the view of "Fine-Art Printing" This is important if you want highest quality possible. I do a lot of printing at A4 and A3. I can definitely see the difference with the loss of nearest neighbor artifacts in high detail prints at a4 between files sent at 360ppi and 300 ppi. This isn't as important for smaller prints. often you can even just get close to the printer resolution and be ok.

If the image size is close to the native resolution, then the algorithm doesn't matter much You can leave automatic scaling to either Photoshop as part of the print dialogue or the printer driver and in each case you don't know how well the algorithms work. If the image size is close to the native resolution, then the algorithm doesn't matter much - if it has to be considerably up or down sized, then you should do it yourself, because it will affect how much print sharpening should be done., or with a good RIP if you have a lot of images to re-size.