>> What do you mean when you say the raster files up to 10 Gb should be "converted"? Converted to what?
You mentioned in post 8 that your TIF files have up to 10GB, so that's why I tool the number 10GB
"Converted to what" ==> converted from the raster image resolution to the plotter resolution
That means you don't send the TIF to the plotter as it does not have the same rastering as the plotter has, so it has to rerender the image to fit to the resolution of the plotter for the page-size defined in the layout.
Small sample: if you have a image with 100x100 pixel (i know that unrealistic, but easier to understand) and you want to plot it to 10x10 inch using a printer resolution of 600dpi AutoCAD (or what ever software you use for printing) has to resize the image from 100x100 pixel to 6000x6000 pixel (10 inch * 600 pixel per inch).
So while a compressed tif with the 100x100pixel image might have just a few bytes the raster built in memory (before sending it to the printer) needs 6000x6000x3 byte = 108MB (the value 3 comes from an image with 24bit=3byte for true color).
While 100 pixel are easy and fast converted to 6000 pixel your 10GB image needs a lot more ram and processor power to be transformed to the resolution of the plotter.
Hope that is more clear now, - alfred -
Thanks Alfred. Now can you check my math for me?
I have a typical image that is 12,400 x 12,400 pixels.
So if I want to create a 36 inch square plot at 600 dpi that would require 36x600 = 21,600 x 21,600 x 3 = 1,399,680,000 bytes of memory?
That is 1.4 Gigabytes. Am I correct?
That seems reasonable, but it seems I have more trouble sometimes when I zoom into a small area of the image (i.e. 2000 pixels square) and try to plot that at 36 inches. So I wonder how memory needs are affected in that situation?
>> I zoom into a small area of the image (i.e. 2000 pixels square) and try to plot that at 36 inches.
>> So I wonder how memory needs are affected in that situation?
In most cases the output resolution is the bottleneck and as I wrote in my sample: even if the input is small (your sample with 2000 pixel) the output is what raises the memory needs, so the 36 inch square output is it, that results also in 1.2GB.
It might be necessary to calculate more memory if the input has also a big filesize/resolution.
It is possible to calculate less memory if you lower the output (e.g. half resolution results in a quarter of the memory)
It is possible to reduce the memory needs if the calculation results are done on a file on the harddisk instead within memory (but then real slow).
HTH, - alfred -