Following on from parts 1 and 2, here’s some more composer changes which are coming in QGIS 2.2

• Rotation support for all composer item types. Now anything you draw in a composer can be rotated, including scale bars, legends, attribute tables and html frames! Rotation of text labels has also been improved by making the border and background of labels respect the rotation of the label.

Every composer item can now be rotated…

• Resizing of rotated items has been improved. Now it’s possible to easily resize rotated items while keeping their correct shape. (There’s still one missing ingredient for complete support here – shear/perspective transforms. Unfortunately this will probably have to wait till 2.4).

Better resizing of rotated items

• Rulers can be shown or hidden in compositions
• The ruler appearance has been tweaked, adding smaller divisions and better text placement

New tweaked appearance for rulers

• A zoom to actual size button and short cut (Ctrl + 1) have been added

New Zom to 100% button

• Lastly, the status bar has a new zoom combo box, which shows the current zoom level and allows for quick zoom to several predefined levels. You can also enter an exact zoom level in the box for precise control.

New zoom levels combo box in the status bar

As you can see, the print composer in QGIS 2.2 just keeps getting better! There’s a few other really exciting new additions which have landed recently too, but they deserve their own blog posts. Stay tuned…

One of the big features I worked on for QGIS 2.2 is gradient fill symbols for polygons. In my view QGIS’ symbol support is one of its biggest strengths — the versatility of its symbol layers coupled with the powerful data defined properties support allows for so many effects which just aren’t possible in other GIS packages. Gradient fill support is a nice addition to these features and should help make QGIS even more attractive to cartographers. In this post I’m going to give a quick run through of how gradient fills work in QGIS, and some of the options available for tweaking them.

Gradient fills are enabled through the Style tab in the properties for a vector layer. The default fill for a polygon in QGIS is “Simple fill”, so to switch a layer to a gradient fill you first need to select the “Simple fill” layer, then change the “Symbol layer type” dropdown  to “Gradient fill”:

As you can see, there’s a lot of options in QGIS which can be tweaked for gradient fills. I’ll run through each of them now and explain a little bit about how each one can be used.

Colour modes

QGIS supports two different types of colour modes for gradient fills. The first is a simple “Two color” gradient, where the colour smoothly blends from the first colour to the second. The second mode, “Color ramp” allows you to use any of the standard or user-defined QGIS colour ramps, which can consist of multiple colour stops:

Colour options in gradient fills

So, when would you use these options? Well, any time you need more than two colours or need to tweak the position of any of the colours in the gradient you’ll have to use a colour ramp.  If instead you’re just wanting a quick-and-easy gradient then the two colour option might be more suitable.

One last important distinction is that the colours in a two colour gradient can be set using a data defined expression:

Data defined gradient colours. Please try to use them more tastefully then this!

Gradient types

The next option for gradient fills is rather self-explanatory: gradient types. QGIS supports linear, radial and conical gradients:

Coordinate modes

The coordinate mode option is a little trickier to explain. The default setting, “Object“, will cause the gradient to be drawn entirely within each separate feature. You can see in the example below that every lake feature is coloured with a gradient which starts with light blue in the top left and darkens to a deeper blue in the bottom right. This gradient fill is repeated for all the lake features:

The “Object” coordinate mode for gradient fills

In contrast, the “Viewport” coordinate mode causes the gradient to be drawn across the entire current view of the map. So only the lakes in the top left of the map are drawn with the light blue colour, and the lakes in the bottom right with the deeper blue:

The “Viewport” coordinate mode for gradient fills

The choice of coordinate mode will depend entirely on your cartographic desires for your map!

Reference points

QGIS gradient fills allow the setting of two “reference points“. These points control where the gradient fill begins and ends. It’s easiest to visualise how these work by imagining a square defined by the points (0, 0) in the top left and (1, 1) in the bottom right. The two reference points fall somewhere within this square. So, the default reference points of (0.5, 0) and (0.5, 1.0) represent points mid way along the top edge and and the bottom edge, respectively.

Now imagine that this square forms the bounding box for the feature being drawn (or the current map window, if in “viewport” coordinate mode). The default reference points mean that the gradient will be drawn from the middle of the top edge to middle of the bottom edge of the feature. Reference points of (0, 0) and (1, 1) would mean the gradient is drawn from the top-left to the bottom-right. Similarly, reference points of (0.5, 0.5) to (1.0, 1.0) would draw a gradient from the middle of the feature to the bottom right (good for radial gradients).

There’s also the option to set either of the reference points as the feature centroid, which again can come in handy for radial or conical gradient types.

Gradient spread

If you’ve got your head around the reference points concept, then the next setting for gradient fills affects how the gradients spread. This takes effect whenever a gradient starts or ends before the bounds of the feature. The default setting of “pad” means that the gradient will simple “pad” out any extra space with the start or end gradient colour:

“Pad” spread – notice how the darker blue is stretched across the right side of each feature

“Repeat” mode will tile the gradient across the feature:

“Repeat” spread

Finally, “reflect” mode will draw a reflected version of the gradient to fill up any extra space:

“Reflect” spread

Angle

Last of all, there’s a simple “angle” parameter, which allows you to rotate the entire gradient fill. This option is included mostly for use with data defined symbols, since a similar effect can be achieved by changing the gradient reference points. Amongst other effects, this is useful for achieving a “sun glint” on water, where each gradient is drawn in a random direction (more on this in a later blog post):

Random data defined gradient angles

This leads me into my final note… all of these properties can be data defined! So you could have a column in your data controlling whether each feature is drawn with a radial or linear gradient, or whether the gradient in a given feature should be drawn at a specific angle, or that the gradient in a feature should start at the centroid and end at the top right of the feature!

I’m excited to see what the QGIS user community is able to create using this new gradient fill feature when 2.2 is released. If you’ve already had a chance to play with the dev version of 2.2 and have something to show off, make sure you submit your map to the Flickr QGIS Map Showcase!

Following on from Part 1, here’s some more recent usability improvements which have been made to QGIS’ map composer:

• Multiple items can now be selected by drawing a box around them. Holding Shift while drawing a box adds to the current selection, holding Ctrl will subtract from the current selection. In addition, holding Alt while drawing the selection box changes from an “intersects” style selection mode to a “within” selection mode. This was my major frustration with the composer in earlier versions of QGIS, and combined with the earlier fixes for multi-item drag and resize it substantially improves the design workflow.

Selecting multiple items with a mouse drag

• Grid lines are always drawn on top of compositions
• There’s new menu options for showing the grid and snapping to grid, and for showing/snapping/smart guides. All these have convenient keyboard shortcuts so you can quickly switch on or off display of the guides and grid or temporarily switch on or off snapping while you work. There’s also a new menu item for clearing all guides from a composition.

Grid and guide controls in the view menu

• The composer window now has a status bar which shows the exact cursor position and details about resizing/moving items

The new composer status bar

• A new mouse-based zoom tool has been added to the composer, which allows you to click and drag to specify an area to zoom to. You can also switch to this new tool at any time by holding Ctrl and space. Holding Shift while using the zoom tool switches to zooming out.

That’s it for now, but there’s loads more to come for 2.2!

In the spirit of depesz’s “Waiting for Postgresql 9.#” series (and BostonGIS’ follow up Waiting for PostGIS 2.#), here’s the first in a new series of “Waiting for QGIS 2.2″ posts…

As I mentioned in my last post, one of my major goals for QGIS 2.2 is to make the print composer behave more like a professional DTP package. A big part of this is making sure QGIS respects all the expected standard behaviour for graphic design programs. While the print composer made huge advances toward this goal in version 2.0, there’s still further we can go.

So, let’s start with a list of all the new usability features the print composer has already gained for QGIS 2.2:

• Selections of multiple items can all be moved and resized together (previously only one item in the selection would be altered)
• Mouse handles are always drawn on the top of the composition – in earlier versions mouse handles could be hidden by items sitting higher in the composition
• Holding shift while resizing items locks their aspect ratio
• Holding control while resizing items causes them to resize from their centre
• Holding shift while moving items locks the movement to the horizontal or vertical
• Holding control while moving items temporarily disables any grid or guide-based snapping
• Items can be removed from a selection by shift-clicking them
• Control-clicking an item causes the next item below it to be selected – this allows you to select items which are hidden behind other items
• Holding shift while moving items using the cursor keys results in a large item movement
• There’s a new “Edit” menu in the composer, with options like
  • Select All/Select None and Invert Selection
  • Select Next Item Above and Select Next Item Below
  • Lock Selected Items and Unlock All

The new composer “Edit” menu

Additionally, there’s a bunch of pull requests which haven’t yet been merged to master, but are ready to go, including:

• Adding a dedicated “pan” tool for dragging around compositions
• Compositions can be panned at any time by holding the space key or middle mouse button
• Compositions can be zoomed by using the mouse wheel

Composer pan tool

Some of these are simple little changes which don’t sound like much, but it’s not until you’re forced to work without something like shift-resizing to lock the aspect ratio that you realise how often you use it and rely on it!

While it’s really exciting that the release of QGIS 2.0 is just around the corner, I’m finding it just as exciting that work is already under way on version 2.1! Some great new features have already landed and I can’t wait to see what else is planned. Personally, I’ve got plans for a raft of improvements relating to print composers with the goal of making the print composer behave as much like a full-blown DTP package as possible. In the meantime, here’s a quick teaser of a feature I’ve been working on that will hopefully be landing in master soon:

Here’s a cool trick which you can do in QGIS 2.0. It builds on two new features introduced in version 2.0 – atlas prints and html labels.

Atlas prints (previously available as a plugin, now integrated into QGIS core) were developed by Oslandia, and allow you to create data driven map layouts from a specified coverage layer. You can read more about them here.

Another new feature in QGIS 2.0 is the ability to render composer labels as html (courtesy of Olivier Dalang). This allows all kinds of fantastic effects, such as formatting text in the middle of a label (using <b>, <i>, and <font> tags) or creating labels which contain HTML tables. You can even use CSS stylesheets and rules to format the HTML! I’ve been told JavaScript also works inside the labels, but I’ve yet to try this out.

You can combine these two new features for some great tricks. Let’s say we’d like to create a set of maps of local councils, and we want each map to have a watermarked logo of the council on it. For this example I’ve created a basic basemap of Victorian councils, and I’ve downloaded all the council logos (in a variety of formats) to a local folder. Next, I’ll add an extra column to the council layer containing the name of the logo image:

Adding a logo column to the table

Then, we’ll throw together a simple composition containing the map and set it up as an atlas print:

Generating an atlas

Now for the fun bit. I’ll add a label item to the composer, set it to “Render as HTML”, and insert some specially-crafted html:

The magic HTML label…

For copy/paste purposes here’s the label contents again:

<style>
* {margin: 0px; padding: 0px}
</style>
<img src="file:///home/nyall/GIS/council_logos/[% "logo" %]"
style="width: auto; height: 100%; display: block; margin: 0px 0px 0px auto;" />

There’s a couple of things to note here. First, the magic happens in the image’s src attribute (“file:///home/nyall/GIS/council_logos/[% "logo" %]“). When the composer is exported, QGIS will replace the [% "logo" %] part with the contents of the logo field for each row in the councils table. This means the image source will then point to the local copy of the council’s logo, eg “file:///home/nyall/GIS/council_logos/glen_eira.jpg” for the first row.

Secondly, I’ve styled the image with the css:

width: auto; height: 100%; display: block;

This allows the image to resize to 100% of the height of the label while maintaining its correct aspect ratio. I’ve also added the rule

margin: 0px 0px 0px auto;

to force the image to right align within the html label. This ensures that all the watermarked logos will appear in a consistent size and position for each map.

Lastly, I’ll remove the label’s frame and background by unchecking these options, then set its transparency to 80% under the new “Rendering” section:

Yet another new feature in QGIS 2.0…

Ok, we’re all done. Now, when I select Composer -> “Export as Image”, we’ll get a lovely set of council maps complete with watermarked council logo!

A watermarked atlas!

There we go — all ready for print, with no manipulation in external programs required at all!

Bonus post-credits section

Here’s a kicker — the linked images don’t need to be local. That means you can even piggy-back off an existing web service to generate an image on the fly! Let’s say you were asked to add QR codes to your maps to link directly to the council website. All it takes is adding a new column to the table, then modifying the image src to read:

src="http://qrfree.kaywa.com/?l=1&s=8&d=[% "url" %]"

Now when we export the maps we’ll also get a QR image generated on the fly and inserted into the layout!

Complete with dynamically generated QR code!

Combining HTML labels and atlas prints in this way is extremely powerful. This example is just touching the tip of the iceberg – I’m keen to see what the community can do with this when QGIS 2.0 is released!

I was recently interviewed for an article on open access to geospatial data in the City of Johannesburg. The article has been published online now for all to enjoy...