Printing layouts containing an image service from ArcGIS Image Server

Note:

ArcGIS 10 is the last release of the stand-alone ArcGIS Image Server product. The image service definition (.ISDef) has been replaced by an improved geodatabase data model—the mosaic dataset—which can be published as an image service using the ArcGIS Server Image extension.

You can print large-format plots with large images served from image service definitions; however, you will need to calculate the print driver quality parameters and resampling factor and adjust the Maximum Columns image service definition property.

Setting the print driver quality and resampling factor

When imagery from an image service definition is displayed on the screen, only the pixels required for the displayed area are processed by the server and sent to the client's display. Typically, this is about one million pixels for a full screen, which can generally be processed in a subsecond. When creating a large-format, high-resolution print, the size of the required image is considerably greater.

There are many different factors that define the print quality when printing vectors and imagery on an ink-jet printer, including the paper type, ink, and dots per inch (dpi). In general, the higher the dpi, the finer the details that can be plotted. You should also be aware of the pixels per inch (ppi) of the image being printed.

Printers are normally classified by their dpi. Older ink-jet printers have up to 600 dpi, and newer printers claim approximately 2400 dpi. On most high-resolution printers, the driver's print quality can be set to a lower number than the achievable maximum. Similarly, when printing with ArcPress, the dpi (up to 600 dpi) may be set. In ink-jet printers, a dot in dpi represents a small drop of ink (in one of the colors of the printer ink). The dot can be either placed or not placed but generally does not have a size range for the color it adds. A dot is therefore very different from an image pixel, which can have a huge range of colors and densities. A pixel is represented on the paper as a large number of different dots. There are different techniques by which different printers and raster image processors (RIPs) convert pixels to dots—these are often referred to as dithering techniques. Generally, at least 16 different dots are needed to create a full range of tone values for a specific color; therefore, the relation between dpi and ppi should be approximately 1:4.

Dpi is often used to define the resolution of devices that display pixels, such as computer screens, which can be misleading. A more accurate way to define a computer screen is to use ppi. A computer screen normally has a resolution of about 100 ppi. A high-quality plot has a resolution of about 200 ppi or 80 dots per cm (dpcm). For most applications, 150 ppi is suitable. If using 150 ppi, this would mean that a high-quality image with a 1-meter pixel size can be plotted up to a scale of 1:6000. Similarly, an image with a 10-meter pixel size can be plotted to a scale of 1:60,000. Plotting the same imagery at a larger scale would result in the pixels appearing slightly blurry. Plotting at a smaller scale may result in a slightly better quality but larger image, but the volume of imagery required to cover the same size page would increase. At 200 ppi, a reasonably large color plot (for example, 30 x 30 inches) requires an image of approximately 108 MB (6000 rows, 6000 columns), which is quite easily managed.

When plotting an image from ArcMap, the resolution of the image requested from ArcGIS Image Server is dependent on the scale of the plot. This is determined by two values—the dpi of the printing device and the output image quality (resample ratio) that is set on the Print dialog box. The ppi is computed as the dpi/resample ratio. Therefore, if you have a printer driver that is set to 2400 dpi and the resample ratio is set to 1, then the image requested will be at 2400 ppi. In the 30 x 30 inch example, this would request an image of approximately 15 GB (72000 rows, 72000 columns). This will take a long time to process and may kill even the best client application. To make requests from ArcGIS Image Server that are not too large yet still result in a good image quality, large requests are split into smaller requests of about one to two million pixels each, in the banding process. If the resample ratio is set to 5, the image resolution will be approximately 480 ppi, which is still far above the high quality of 200 ppi, but the image size would be about 600 MB, which is achievable with most computers.

In conclusion, when making plots that have large volumes of imagery, it is better to set the plotter resolution to a maximum of 600 dpi and the resample ratio to 3, which would result in 200 ppi and an image request size of about 108 MB. An acceptable quality can be achieved by setting the resample ratio to 4 (150 ppi). Setting the resample ratio to 2 or 1 will create very large requests with little gain. If using a plot driver with 1200 or 2400 dpi, the resample ratio should be set to 5.

Setting the Maximum Columns image service parameter

Each image service definition has a Maximum Columns parameter that is defined on the Image Service Properties dialog box on the Client Interface Control node. This defines the column width of the largest request that the image service will accept from a client application. The default value is 15000. This setting is similar to one in ArcIMS, where there is a standard screen request size, but very large requests are excluded. If you intend to plot using very large image requests, it is necessary to first increase this value, then recompile and publish the image service definition. A value of 20000 should be more than sufficient for most applications while automatically ignoring the accidental 15 GB requests.

Note:

The Maximum Rows property does not need to be reset for printing because the banding will automatically split the request into narrow bands.

When working on a client machine with low bandwidth, you should also consider setting the compression factor used for transmission prior to printing to reduce the volume of data transmitted. Using a JPEG compression with a quality as low as 50 percent will have little effect on the plot quality but will substantially reduce the bandwidth requirements in most cases.