Using Blender for Geospatial Projects

Introduction

Blender is a free and open-source 3D content creation application. Although most typically used by artists, animators and game developers, its user base extends far beyond that into professional, academic and hobbyist fields. It offers a broad feature set, even when compared to its commercial competitors, perhaps putting it more into an ‘all-rounder’, rather than specialist, category. Versions are available for Windows, OS X and Linux platforms, with the core application being relatively small and able to function on modest-specification hardware. In addition, perhaps due to its ‘open’ status, a huge amount of training support is available, a lot of which is free. As such, it has a comparatively low barrier of entry.

Given all of this, it’s perhaps not so surprising that when geospatial specialists which to include a creative 3D tool as part of their workflow, they often choose Blender.

OS NGD vector data, visualised in Blender

Why use Blender for Maps and Geospatial Data?

Blender is principally used to create three core forms of output – images, animations and 3D models. Each of these have relevance to geospatial. The requirement for images and animations needs no explanation, whereas 3D models are a resource that can be usefully ingested by GIS tools.

It should be acknowledged from the outset that GIS tools like QGIS and ArcGIS Pro are already well equipped with visualisation tools – and we’d certainly recommend using these for the typical GIS work they’re intended for. However, where Blender shines is in situations where something more creative or out-of-the-ordinary is required. Of course, being a 3D orientated application, Blender is suited to projects that incorporate 3D data – but that’s certainly not the limit of what’s possible.

Many GIS tools also offer some level of support for animation, but Blender’s capabilities for this far exceed these in many respects. Related to this, Blender also includes its own video editing tools, which can be handy if you’ve not got the reason or budget to be investing in separate tools for this.

Finally, should you require custom 3D models to act as ‘3D symbols’ within your GIS scene, then Blender is, of course, a great option.

Learning Blender

Some people are apprehensive about learning Blender, believing it to be a large and complex undertaking. They do have a point – it IS a very expansive and complex application! However, many users only need to learn and use a small fraction of its capabilities – and that’s OK. As it’s free, so it’s not as if you’re paying for unused features. In practice, once you get the basics down and learn to focus on what’s relevant for your own purposes, it becomes a lot more manageable.

The purpose of this document isn’t to teach you Blender. There are already many resources out there dedicated to that (we’ve listed a few suggestions in Annex A). Instead, the focus is to document subjects and techniques that are specific to working with geospatial data .

If you’re starting with Blender from scratch, with a view to using it for geospatial content, you’re unlikely to find training resources that cater directly to your needs. As such, you should expect to spent time learning to create more mainstream content (art, gaming assets, motion graphics, simulations…), with a view to later applying what you’ve learned to your own discipline. If that means creating doughnuts and spaceships, so be it…it’ll help you out in the end.

Blender work typically requires a mix of technical knowledge and art. As such, practical experience is at least as important as theoretical knowledge. It’s not sufficient to just read or watch a selection of tutorials – you need to be prepared to actively experiment and practice!

Viewing a wireframe object in Blender's user interface

Importing Geospatial Data into Blender

For anyone contemplating using Blender for geospatial projects, you first need to consider how you’re going to import your geospatial data. Blender, by default, doesn’t support geospatial-specific file formats, so work-arounds are required. We are aware of four proven methods for achieving this, which are discussed below. Each has different strengths and weaknesses, so your choice will very much depend on the requirements of each individual project.

Third-Party Apps

Use Third-Party Applications to Transform Geospatial Data into 3D Models

Some GIS applications (e.g. QGIS, ESRI’s ArcGIS Pro) allow users to export spatial data to common 3D formats that can be in turn understood by Blender.

Safe Software’s FME is a particularly good option for this flowline, as it supports not only a multitude of spatial data formats, but a useful selection of standard 3D formats also. Should you, for example, wish to create a 3D TIN from standard terrain products, then texture it with geotiff images, then this approach is a great way to achieve that.

Terrain5 and draped imagery, prepared in FME and visualised in Blender

We’ve found that .OBJ files (aka Wavefront Object files) are a good choice for importing models into Blender. The .OBJ file specification is simple and well-established, so implementations of it within applications are usually consistent. It doesn’t have the more advanced features found elsewhere, but for geospatial data, this is often not relevant.

The limitations of this approach are that, in the translation from spatial data to 3D, any references to spatial coordinate systems are lost, as is all attribution.

Blender GIS

Blender GIS is a free and well-established Blender add-on that can be downloaded from https://github.com/domlysz/BlenderGIS . Once installed, it adds a new ‘GIS’ option on the 3D Viewport’s main menu that offers various geospatial tools. Despite its name, Blender GIS doesn’t turn Blender into anything resembling an actual GIS application. However, it does provide a means for importing a small, but useful, range of spatial data directly from formats like ESRI shapefiles or ASCII grids.

A notable feature of Blender GIS is its support for spatial reference systems – something missing from the other approaches listed here. Whilst this functionality isn’t quite as versatile as you might ideally expect, what it does offer is certainly useful.

Another unique feature of this tool is its functionality for generating base maps from online mapping APIs such as Google Maps, Bing Maps, Open Street Map, etc. If you wish to set them up, additional custom API connections can also be made.

One thing that Blender GIS can (mostly) not do is support feature attribution. Certainly, there’s no bridge between the source data and Blender’s own internal attribution system. What it does offer is support for building height data, which is read and processed into geometry at the time of import.

Image Orientated Flowlines

Image-Orientated FlowlinesWhilst Blender has no default support for geospatial data formats or CRS, it does have extensive support for working with image files. As such, if you use a GIS tool to convert feature layers into a stack of aligned images, these images can then serve as the geospatial content of your project.

A typical flowline for undertaking this in Blender might be as follows:

- Use the Import Images as Planes add-on to import the basemap image.

- Use Blender’s displacement tools to derive a terrain surface from a heightmap image.

- Add additional image layers, using additional Texture nodes in the Shader Editor.

Of course, an image-based flowline is not going to offer CRS or attribution support.

A visualisation created from GIS-derived image resources

Blender Python

Blender is supplied with an integrated Python interpreter, script editing tools, a selection of commonly used Python packages and its own Blender Python package. As such, it’s possible for users to author custom functionality in the form of either scripts or add-ons.

As such, you can create scripts that will read spatial data from an external file and convert that into Blender objects. The most straightforward method for doing this is to read data from CSV. The most basic versions of this method only support point geometries. However, more ambitious variants may implement WKT, which thus can support lines and polygons.

An advantage of this approach is that it allows attribution to be read and incorporated into Blender’s own attribution system. This opens the way for the Shader Editor and Geometry Nodes tools to make active use of your attribution, something that quickly becomes an extremely powerful and versatile tool.

Unfortunately, as it stands, there are still limitations regarding this approach. Most significantly, Blender’s own attribution system doesn’t fully support strings or dates, meaning that some attributes must either be ignored, or will need to be encoded.

This map uses Geometry Nodes to derive complex 3D models from attributed point features

Geospatial Coordinates

As is the case with most mainstream 3D tools, Blender includes no native support for Coordinate Reference Systems (CRS). For anyone coming from the geospatial world, this might initially seem to be a deal-breaker. However, whilst CRS support would certainly be a welcome addition, the impact of this is often not quite as significant as you might have initially feared.

One reason for this is that when used to produce rendered images, Blender effectively sits at the end of the flowline. As such, correct geospatial positioning often isn’t necessary, so long as relative positions are retained. Furthermore, as there’s usually no reason to subsequently export the data, alterations to the coordinates will have no knock-on impact.

If you’re using an image-based flowline, as mentioned previously, then issues around geospatial coordinates aren’t likely to be of any significance to you. Likewise, if you’re using the Blender GIS plug-in, issues around geospatial coordinates will be handled for you by default. However, if you’re going to be importing 3D objects that have been externally converted from geospatial data, or are perhaps writing an import script in Python, then you’ll certainly need to be more aware of how such data needs to be handled.

Working with British National Grid (ESPG:27700) Data

Here at Ordnance Survey, the coordinate system we most frequently work with – and supply in our products – is British National Grid (BNG). As such, we’re very familiar with coordinate definitions that might look like (414322, 130464). Whilst there doesn’t appear to be an absolute ceiling on the values that can be used within Blender, working with six- or seven-digit ordinate values can be confusing and tricky.

If you’re going to be importing BNG geometries (or similar…), the best approach is to apply, outside of Blender, a universal offset to all coordinates, so that the resulting ordinate values bring the objects much closer to the scene origin (0,0,0). Doing this is often sufficient to ensure an imported model is immediately usable.

Should applying an offset, outside of Blender, not be an option for you, then you may then need to consider the following steps…

• It’s likely that, upon import, your model won’t be visible in the 3D Viewport, even though it will (hopefully!) be listed in the Outliner. The is because the viewport camera will be pointed at the scene origin, whereas the imported object’s mesh will be located very far from that, even though the object’s overall location will be set to (0,0,0). This discrepancy between the object location and the position of its mesh is a big factor in why things may not behave as expected. To move the viewport so that it’s in sight of the mesh, you need to select the imported object, enter Edit mode, select ALL, then choose View->Frame Selected (or numpad <.>).

• Moving the viewport camera, will take you nearer to the object’s mesh, but it’s still likely that you won’t be able to see it. The main reason for this is usually that the viewport’s clipping parameters – specifically, the Clip End value – is set at far too small of a value. This can be viewed and changed using the View tab in the Viewport’s right sidebar. So, change the default value (1000m) to something much larger (100000m?). Hopefully, you’ll then be able to see the mesh. If you can’t, try rotating the view…and if even that doesn’t work, try doing View->Frame Selected again.

• Even when you can see the mesh geometry, it’s very likely to appear corrupted and will flash when the viewport is moved. This is clearly unpleasant and not something you can work with. The issue, as mentioned above, is that the position of the mesh is so far from the object’s Location. Blender just doesn’t seem able to handle this situation. The solution, though, is simple – you either need to move the origin to a location that’s in the vicinity of the mesh; or move the mesh closer to the origin. In practice, the first of these is usually easier to accomplish. To do this, select the object, enter Edit Mode, select a part of the mesh (say, one vertex) and then snap the 3D cursor to it (using <Shift><S>, ‘Cursor to Selected’). Now, return to Object Mode and use Object->Set Origin->Origin to 3D Cursor. The object’s origin should now be situated within the vicinity of its mesh – and everything should now be behaving normally. Of course, you might then want to move the object back to a more sensible place in the scene (maybe (0,0,0)?), but this isn’t absolutely necessary.

Exporting Geospatial Coordinates from Blender (using FME)

At first sight, the idea that you can export geospatial coordinates might seem fanciful, as Blender obviously contains no such coordinates. However, it turns out that if you also use FME (or maybe some other tools?), it actually is achievable.

The trick to this is that you’ll need to treat Blender’s local coordinate system as if it was actually a CRS. In our case, this would be BNG, potentially with an offset applied. So, you then export your models in .OBJ format, then read these into FME with the .OBJ ‘reader’. At that stage, you’ll get the option to choose which CRS the data represents, so you’d choose, in our case, BNG (ESPG:27700). FME will then assume that the coordinate values are BNG coordinates, even though they’re actually native Blender coordinates. Although exporting geospatial coordinates from Blender probably isn’t something you’re likely to be doing often, it’s a useful trick that’s worth remembering.

A visualisation of the Beamish museum using NGD feature data

What Blender Features are Most Useful for Geospatial Work?

With enough imagination and ambition, it’s probably the case that there’s scope for any of Blender’s features to be usefully used in the context of a geospatial project. However, it’s probably true to say that some features are going to be more useful for geospatial work than others. The following section, then, suggests the skills that are most likely to be of interest to anyone getting started.

Core Skills

Before you can do anything much at all with Blender, you need to need learn the basics of how the user interface works. This includes the tools for arranging the panels, selecting editors, navigating the 3D Viewport, installing add-ons and working with parameters. If you follow any online beginner tutorial, you’ll likely pick up all of this as a matter of course. Although there are alternatives, efficient use of Blender often depends upon the use of quick keys. You should certainly endeavour to learn at least the most basic of these.

Modelling (Hands-on)

Many people primarily think of Blender as being a ‘3D modelling’ tool. As such, it’s no surprise that the manual creation of 3D geometries is extensively supported. As a beginner, it is certainly worth learning how simple ‘low poly’ mesh modelling works, a the very least. Having said this, many geospatial projects incorporate relatively little hands-on modelling – so don’t assume that this will always be an essential skill. When it comes to skills that are more orientated towards ‘organic’ modelling (subdivision surface, sculpting, etc), it’s less likely, still, that these are going to be relative to geospatial project. So, you might wish to de-prioritise these.

Materials/Shaders

Knowledge of Blender’s materials/shaders system is most certainly something that’s going to be of value when undertaking geospatial projects. You should endeavour to become familiar with the Shader Editor and its node-based workflow as a matter of priority, as it’s often this that can deliver the impressive graphical effects that likely attracted you to using Blender in the first place.

An example of what can be achieved with Blender's materials.

Rendering, Cameras and Lighting

If you wish to use Blender to create high-quality visualisations, it’s fundamental to understand how the rendering process works. This also requires a good understanding of cameras, lighting and ‘environment’ settings. Whilst you can learn the absolute basics of each of these quite easily, it’s certainly worth putting in the time to explore each of these further, as they’ll have a direct influence on the quality of your output.

Procedural Modelling (Geometry Nodes)

Procedural modelling – that is, the creation of models via a sequence of rules – is relatively new to Blender. It’s not easy to learn, being more technical and maths-orientated than other disciplines. However, it can be extremely powerful, so is certainly worth the effort. Blender’s implementation of procedural modelling is enabled by the Geometry Nodes modifier, which itself is paired with a custom node-based editing environment.

If you use geospatial vector data directly within Blender (ideally, with attribution intact) then Geometry Nodes can be of particular value. For example, a single Geometry Nodes network might have the ability to automatically convert a multitude of building footprints into a whole 3D city.

Using attribution and procedural modelling to create a 3D bar graph

Animation

For some geospatial specialists, animation may be the primary focus of their Blender work. Blender’s support for animation is extensive but may be simplified if a separation is established between simple object animation (relatively simple to learn) versus character animation (challenging). As character animation probably won’t be required for most geospatial projects, this does suggest that you needn’t worry too much about the more involved aspects.

Video Editing

Some people are surprised that Blender incorporates its own video editing functionality by default. Although it doesn’t have the broad range of tools that you might find in specialist editing applications, it’s still a useful tool. If you’re rendering animations and wish to add things like audio, transitions and text, then this is a simple way of achieving that whilst remaining within Blender’s environment.

A screenshot of Blender's video editing interface

Annex A - Recommended Learning Resources

There’s a huge goldmine of free and commercial resources out there that will help you develop your Blender skills. Whilst we can’t even begin to ‘scratch the surface’ of this in a document of this type, we’ve included, below, some links that we’ve found particularly helpful in the past.

Official Blender Documentation

As should be obvious, your first source of information about Blender should be the Blender website itself (www.blender.org). Here, under the Support header, you’ll find extensive user manuals, along with a small selection of free tutorials. Further official tutorials can also be found at the separate Blender Studio site (https://studio.blender.org/). This requires a paid subscription, as it’s partly this that funds Blender’s development costs.

YouTube

Blender Guru

When it comes to third-party learning resources, then YouTube is very much a goldmine. The most famous (and, most certainly, the most viewed) tutorial series for beginners can be found on the Blender Guru channel (www.youtube.com/@blenderguru). Specifically, it’s the tutorial series for creating a model of a doughnut that is best known. These tutorials cover a broad range of Blender’s functionality, so serve as a good introduction to what Blender is all about. This does happen ‘at pace’, though, so for some people, this can be ‘too much, too soon’.

Grant Abbit

If you prefer a slower pace, the beginner-level tutorials on the Grant Abbitt channel (www.youtube.com/@grabbitt ) are a good option. Grant is a teacher by profession, which is evident in his approach to presenting content. As with Blender Guru, the focus is orientated towards mainstream 3D art, so the content isn’t going to be directly applicable to GI orientated work. However, many of the core skills are certainly transferable.

Ducky 3D

One rather different channel worth mentioning is Ducky 3D (www.youtube.com/@TheDucky3D). You’re unlikely to find beginners tutorials here. Instead, he focuses on ‘motion graphics’, which are typically short and abstract animations. Whilst, again, this is not directly GI related, he often features ideas and techniques that could potentially be incorporated into a ‘spatial data’ context. If you’ve already got an understanding of Blender’s basics, this channel is very much recommended.

Erindale

Be warned, this one (https://www.youtube.com/@Erindale ) is very much at the ‘advanced’ end of the learning spectrum and is most usually focused on procedural modelling techniques. However, it provides a lot of advice that really showcases what’s possible with Geometry Nodes – and this is certainly something that can be built upon in a geospatial context.