Image Quality

From Deskins Group Resources

This page describes the background of making high-quality images for your report or final publication. A related topic is Making Molecular Images. It's good to get in the habit of making good images so you can save yourself time and headaches down the road when finishing up your report/article/thesis. Bottom line: save images as high-resolution tif (or png) files for raster images or emf (or svg) for vector images.

Raster Images

Images come in two different types, raster and vector. You can read more about raster images here. Raster images consist of many dots or pixels put together to form a picture. A camera, for instance, makes raster images from what it views in the lens. The size of the image determines how big the picture is and how much detail is in it. For example, an image that is 300x300 means that it has 90000 different pixels. The bigger the number of pixels, the more you can put in that image. Common formats for raster images include gif, png, jpg, bmp, or tif. The problem of raster images is that changing size can reduce quality. For instance, if I have a small image (300x300), but I want to make a poster of the image I may need to resize it to much bigger (maybe 3000x3000). In order to resize, interpolation must be performed to determine what all those new pixels should display, which degrades the quality of the image.

Resolution

Raster images should be at least 300 x 300 dots-per-inch resolution for publication. See here for a discussion on resolution. Dots per inch is exactly that: the number of dots there are in one inch. A typical figure is about 3 1/2 inches wide in the printed format of most journals, meaning that images should be at least 1050 dots wide (3.5*300). More is better, so make your images much bigger if possible! Of course more resolution means bigger file sizes so there is a trade-off at some point. However it is much easier to take a high-resolution image and convert it to lower-resolution, than vice-versa. So start with high resolution and make smaller if necessary.

Compression and File Format

One popular method to make images smaller is compression. There are lossy and lossless methods, which will not be discussed further (google if you want to learn more), but you should avoid lossy formats. This means avoid jpg, but use png or tif formats. Tif is usually preferred by journal articles.

Vector Images

The second type of image is a vector image (more details here). Vector images are based on mathematical formulas to draw pictures, so changing size is not a problem. For instance, the formula to draw a circle is x^2 + y^2 = r^2. I can draw a circle for r = 5 and r = 10 with no loss in image quality. Common formats for vector images include emf, wmf, svg, and usually eps. If you have a choice for an image type, vector is usually better than raster, but this is not always possible with the software you are using. Some journals also will only take submissions with raster images.

Microsoft Word

When you paste an image into Word, it may not be very good quality. Word has a weird way of dealing with images and sometimes compresses them to save space. Emf and wmf files should be ok, since they are vector graphics. You can choose Paste/Paste Special, and then the format of the pasted image. You should also turn off compression and increase the resolution of images, as described here, here, or here. Choose dpi of 300 or better.

Editing/Making Images

You may want to edit an image, such as adding labels to a chart, or labels to an image of some molecules. Or you may want to combine several images together. You'll need some special software to do this.

Image Editing Software

For raster images, you can use paint.net, gimp, or Adobe Photoshop. Both paint.net and gimp are free. paint.net is probably the easiest software of the group to learn. You should learn about layers to use the software effectively. One software that isn't really for editing image, but lets you crop them or save in different formats is IrfanView, which is free. Irfanview is very quick and easy to use.

For vector images Adobe Illustrator and Inkscape are good choices. Inkscape is open source and free. Other free vector drawing programs are LibreOffice draw and OpenOffice draw.

Powerpoint

One cool feature with Powerpoint is the ability to save a slide as a raster image (like bmp). You can paste an image, add arrows, labels, or whatever, and then save the image. By default, however, the quality of the images is low. You'll need to increase the resolution as discussed on this page about high-quality powerpoint images. Be careful when resizing a raster image within Powerpoint since the quality may degrade if you do so.

Excel

This plugin will allow you to export high-resolution charts from Excel. Technically, Excel charts are vector graphics, so you could also copy/paste into a vector editing software (like Inkscape), then export from that software into a high-resolution tif file. Another method is to paste a chart into Powerpoint and then export as high resolution tif file (described above) or save a slide as an emf file. You can also copy, then Paste/Paste Special into Word as an emf, wmf, bmp, etc. file. You may need to try several pastes to get it right.

Other Options

Besides Excel, there are other software for creating graphs that can export directly to various file formats. Gnumeric is an open-source spreadsheet that you can save charts directly to pdf, svg, high-resolution png, and others. It also opens Excel files. SigmaPlot is a specialized software for making graphs. It has many advanced features and you can control graphs much more than other software. It also can save to many formats. The downside is that there is a little learning curve to using it. WPI has a license for SigmaPlot. OpenOffice and LibreOffice both have Excel clones. You can create charts in either software package, then copy/paste into the draw program, and save from the draw program into various formats, including high-resolution raster images. Finally, if you are feeling adventurous you can make great plots with matplotlib.