This is starting to get off topic from matplotlib, but it is relevant to
creating good EPS figures...
When using ImageMagick to transform from to an EPS, your results will be
much improved by using the parameter "-density 288". This increases the
resolution (and thus results in a much bigger file). I find this is
adequate for producing publication quality plots.
···
> Quick warning about ImageMagick and EPS files (and any other
> vector-based graphics format)...
>
> ImageMagick is a raster-based system. When it loads a vector-based
> graphic, it implicitly rasterizes it, performs the requested
> operations, and then outputs the file format. So, even if you
> output the result as a EPS file, you would have lost all
> vector-based info (like text and lines). The result looks horrible.
This is getting off-topic, but this is also something that I feel quite strongly about. There is no point in submitting an EPS file that comes from ImageMagick for publication. Because of the rasterization, you might as well send them a PNG file.
There are a few reasons why publishers request vector-based graphics formats. First, they can change the font properties and the text of the figure as needed. If the image has been rasterized, they then have to edit out the rasterized text and put in their own text. Second, there are various anti-aliasing techniques that can be utilized for text and lines as one scales the image for different publishing mediums. If you take your high resolution rasterized eps file and try viewing the text when the figure is scaled to a small size, and then try the same for an unrasterized eps file, so long as your viewer is decent, the unrasterized image will remain readable at smaller sizes than the high-resolution rasterized version.
Don’t rasterize your vector images… your publications will thank you for that!
/rant
On a related note, you should rasterize your pcolor() and imshow() using “rasterize=True” in those function calls. This will allow the image to be rasterized to its native resolution, while still letting everything else in the figure remain vector-based. This results in beautiful plots with sane file sizes.
Ben Root
···
On Thu, Aug 26, 2010 at 11:27 AM, Jonathan Slavin <jslavin@…2999…81…> wrote:
This is starting to get off topic from matplotlib, but it is relevant to
creating good EPS figures…
When using ImageMagick to transform from to an EPS, your results will be
much improved by using the parameter “-density 288”. This increases the
resolution (and thus results in a much bigger file). I find this is
adequate for producing publication quality plots.