Frequently asked questions
Image to print: DPI, ppi, and pixels. How do they fit together? (continued)
There are two cases to consider:
1: you have too few pixels
Suppose I cropped an image to, say, 1400 pixels X 1132 pixels.
But I intend to enter this photograph in a competition as a 10” x 8” print. The judge (and all other critics) are likely to pick it up and eyeball it at a distance more like 35 millimetres than 35 centimetres!
I want to use my printer’s ability to print at 300 dpi so the critics can’t see any dot grain.
A 10” X 8” print, at 300 dpi, will have 3000 X 2400 dots. But that’s many more than my cropped image has pixels.
Basically, we need a way to distribute colour and tone information among 3000 X 2400 dots so that it looks the same as the colour and tone information we had in our 1440 pixel X 1132 pixel image.
The process is called interpolation, or up-sampling, and its a bit like putting round pegs into square holes: there is data missing. We have to make an informed guess about the best data to fill in the gaps to make the picture look OK.
This process cannot be done successfully if there are too few pixels.
2: you have too many pixels
For this example, let’s assume you have a bigger camera and you’ve cropped an image to, say, 5400 X 3780 pixels. You want to print on A3 paper at 16” X 11.7”
A 16” X 11.7” print, at 300 dpi, has 4800 X 3360 dots. But your image has 5400 X 3780 pixels – too many.
You’re going to have to resample the image to squeeze your information into fewer pixels. This process is called down-sampling, and it’s a bit like shaving the corners off square pegs to fit in round holes.
This process is easier than up-sampling because we have plenty of data available that we can use to calculate the new pixel data.
How to resample?
This problem has been around since digital photography began, and people have invented clever software programs which, using today’s powerful computers, can measure and redistribute the colour and tone data among the new pixels so the compromises are hardly perceptible.
Nevertheless, whether you need to up-sample (create new pixels) or down-sample (squeeze into fewer pixels), your image is going to suffer to some extent. Following resampling your image will never be exactly as it was in the camera or previously in the computer. It could become soft and fuzzy or, on the other hand “jaggy”, as a result.
Different computer ‘algorithms’ are required in each case. Down-sampling is somewhat simpler because the new data for each pixel can be worked out be a kind of averaging of data that was in the surrounding pixels previously. Up-sampling is more difficult because we are trying to make up data for places where there was no colour or tone previously.
How to handle resampling in practice
There are basically, 3 options:
1. Let the printer driver do it. Because the printer can print only at specific dpi resolutions (240, 280, 300, 360 dpi etc), the printer driver will always resample your image if it does not get the right number of pixels. However, you have no control whatever as to which dot resolution it picks nor the accuracy of the program that does it. That is probably the least satisfactory option.
2. If you use Photoshop or a similar program, you can resample manually under the Image/Resize menu. You tell the program the number of dpi that you are going to print to, and the size that you want the final print to be, and Photoshop calculates the dot size the final print should be, then it resamples your image to provide 1 pixel for each dot.
If you choose to manage the process using Photoshop you have a further choice. Photoshop uses different algorithms for up-sampling and down-sampling to deal with the different problems that arise.
3. If you use Lightroom all you need do is specify, in the print menu, both the image size and the dpi you want. Lightroom works out whether your image needs to be up-sampled or down-sampled, and applies the appropriate resampling and output sharpening routines as it is preparing to send the file to the printer.