Deadstick - Development Update
Greetings Deadstick fans!
Building a new Flight Simulator from the ground up was never going to be easy, however I don’t think any of us quite anticipated the challenge that was building our living, breathing environment!
We always knew we wanted to create beautiful and detailed worlds that would compliment low level bush flying and have steadfastly refused to compromise on that ambition. The reality however, has been that many of the tools that we had been using to develop the game simply were not capable of handling the size, scale, volume or complexity of the world we have been trying to develop.
Try as we might to push these to their limits, we reached a point where it became clear that even simple progress was being held back by the tools and something needed to be done.
After some deliberation, the difficult decision was made to de-prioritise progress on the gameplay and instead, focus our efforts on the tools. Designing and building new tools from the ground up which are purpose built for the task of populating the world of Deadstick, that can run fast, take full advantage of all of the hardware we can throw at them and allow us to realize our ambition without compromise.
Sometimes it is necessary to take a step backwards before you can move forwards, however this had left us in a difficult position. Many of you have watched the trailers, participated in the livestreams or even had hands-on experience with Deadstick. Stripping our world back to its bare bones made it very difficult to show screenshots of ‘progress’ when, to outside viewers, it would look as if things had gone considerably backwards. We therefore made the decision to focus on getting the tools done and letting the results speak for themselves as and when the time came. The reality, as with developing any new set of technologies, particularly with a small team, is that things took longer than anticipated and the community, rightly, grew anxious waiting for updates.
The good news however, is that whilst there will always be minor tweaks and improvements to be made, these tools are now done, give us better functionality and performance than we could have ever imagined and will serve as a foundation on which to build our current and future worlds.
In addition to the development of these tools, we have also had much opportunity to think about and refine existing gameplay mechanics and systems within Deadstick. Prototyping new ideas and incorporating functionality into the simulator that we had not anticipated prior to Early Access. Functionality we have been itching to show off, but without a world to show it off in, have been unable to do so, until now! But more about that later...
[h2]The Technical Bit[/h2]
For those who are interested in exactly what goes on behind the scenes, the following is an attempt to break down the technical challenges we have faced over the past months of silence and our solutions to those problems. For those that want to see the end result, simply skip ahead to end.
‘Styx’ and ‘Stones’ may break my bones...
The task of generating a beautiful, dense and believable world is split into three problems:
We have been hard at work tackling each of these head on with two purpose built tools - ‘Styx’ and ‘Stones’.
[h3]Introducing Stones[/h3]
Working much like a traditional 2D layer based paint package but manipulating terrain heights instead of pixels, Stones is the initial building block for developing and crafting our detailed world, allowing us to stamp and blend various terrain features such as mountains, rivers and valleys together in realtime. Don’t like a particular mountain’s location? Stones allows our artists to drag and drop them to a desired location and then quickly and easily blend them into the surrounding terrain. What was once a slow, laborious and error prone process which struggled to cope with the size of the worlds we envisaged, can now be done in realtime with instant visual feedback.
Stones was designed from the ground up to support and even combine both fictional and real world features. This functionality gives us powerful flexibility to explore different opportunities for future worlds once in Early Access.
[h3]Introducing Styx[/h3]
Once we have our base terrain, the next challenge is how to populate it. The traditional flight sim approach has been to overlay aerial imagery. Whilst this technique is very effective at high altitude, it quickly falls down at low level where resolution becomes a challenge. As the emphasis in Deadstick is on low and slow bush flying we wanted to steer away from this solution and instead represent the world entirely from geometry, where every tree, rock and blade of grass within the world is a 3D object the player can both see and interact with.
It simply isn’t practical to place every tree by hand, let alone every blade of grass, and so a system was needed with which artists could define rules for different surface types and the computer could generate and place the appropriate features accordingly. This isn’t entirely new, we have shown off similar technology in all of our previous demos. What we hadn’t anticipated however, was the sheer volume of assets that needed to be created and placed.
As it stands, there are currently 2.3 billion trees which are generated and exist within our world. To arrive at that figure, the tool must first test many more locations to determine which ones are valid based on rules such as surface, slope, texture type, etc. Once a valid location is found, it must then be tested against all other trees/objects to ensure that they do not overlap. This testing takes time and subsequently our terrain generations quickly ballooned from several hours to several days. Try as we might to optimize this process with our existing tool it was simply becoming too burdensome for our artists to make even minor changes without having to endure very long regeneration times.
By moving this functionality to a separate tool, and by taking full advantage of multi threading and clever caching of previously built rules, we have managed to remove much of the generation overhead and waste, reducing our build times from days down to minutes or in the worst case, hours. This gives our artists the freedom they need to experiment with new ideas and maximise the visual quality they can achieve.
As the majority of this work has been rewriting / redeveloping much of our build system’s inner workings, it is probably the least interesting to show visually, but has by far the biggest implications with regards to efficiently authoring and populating our terrain.
This may look like a fairly mundane screenshot but this tool has dramatically improved the way in which we work and in turn, the detail and quality we are able to achieve within Deadstick.
[h3]The Result[/h3]
Beautifully detailed worlds at any altitude no longer limited by scale or build times but instead, only our imagination!
[h2]What’s Next[/h2]
As mentioned above, in parallel to our work on the terrain functionality and tools, much has been done to develop and refine the gameplay systems and interactions within the simulator. Next week we shall release Part 2, in which we start to explore some of these systems in more detail, giving you a glimpse into what you can expect as a day in the life of a pilot within Deadstick!
Building a new Flight Simulator from the ground up was never going to be easy, however I don’t think any of us quite anticipated the challenge that was building our living, breathing environment!
We always knew we wanted to create beautiful and detailed worlds that would compliment low level bush flying and have steadfastly refused to compromise on that ambition. The reality however, has been that many of the tools that we had been using to develop the game simply were not capable of handling the size, scale, volume or complexity of the world we have been trying to develop.
Try as we might to push these to their limits, we reached a point where it became clear that even simple progress was being held back by the tools and something needed to be done.
After some deliberation, the difficult decision was made to de-prioritise progress on the gameplay and instead, focus our efforts on the tools. Designing and building new tools from the ground up which are purpose built for the task of populating the world of Deadstick, that can run fast, take full advantage of all of the hardware we can throw at them and allow us to realize our ambition without compromise.
Sometimes it is necessary to take a step backwards before you can move forwards, however this had left us in a difficult position. Many of you have watched the trailers, participated in the livestreams or even had hands-on experience with Deadstick. Stripping our world back to its bare bones made it very difficult to show screenshots of ‘progress’ when, to outside viewers, it would look as if things had gone considerably backwards. We therefore made the decision to focus on getting the tools done and letting the results speak for themselves as and when the time came. The reality, as with developing any new set of technologies, particularly with a small team, is that things took longer than anticipated and the community, rightly, grew anxious waiting for updates.
The good news however, is that whilst there will always be minor tweaks and improvements to be made, these tools are now done, give us better functionality and performance than we could have ever imagined and will serve as a foundation on which to build our current and future worlds.
In addition to the development of these tools, we have also had much opportunity to think about and refine existing gameplay mechanics and systems within Deadstick. Prototyping new ideas and incorporating functionality into the simulator that we had not anticipated prior to Early Access. Functionality we have been itching to show off, but without a world to show it off in, have been unable to do so, until now! But more about that later...
[h2]The Technical Bit[/h2]
For those who are interested in exactly what goes on behind the scenes, the following is an attempt to break down the technical challenges we have faced over the past months of silence and our solutions to those problems. For those that want to see the end result, simply skip ahead to end.
‘Styx’ and ‘Stones’ may break my bones...
The task of generating a beautiful, dense and believable world is split into three problems:
- Authoring the base terrain model for our world.
- Populating the terrain with natural and man made elements.
- Efficiently rendering that populated terrain within the simulator.
We have been hard at work tackling each of these head on with two purpose built tools - ‘Styx’ and ‘Stones’.
[h3]Introducing Stones[/h3]
Working much like a traditional 2D layer based paint package but manipulating terrain heights instead of pixels, Stones is the initial building block for developing and crafting our detailed world, allowing us to stamp and blend various terrain features such as mountains, rivers and valleys together in realtime. Don’t like a particular mountain’s location? Stones allows our artists to drag and drop them to a desired location and then quickly and easily blend them into the surrounding terrain. What was once a slow, laborious and error prone process which struggled to cope with the size of the worlds we envisaged, can now be done in realtime with instant visual feedback.
Stones was designed from the ground up to support and even combine both fictional and real world features. This functionality gives us powerful flexibility to explore different opportunities for future worlds once in Early Access.
[h3]Introducing Styx[/h3]
Once we have our base terrain, the next challenge is how to populate it. The traditional flight sim approach has been to overlay aerial imagery. Whilst this technique is very effective at high altitude, it quickly falls down at low level where resolution becomes a challenge. As the emphasis in Deadstick is on low and slow bush flying we wanted to steer away from this solution and instead represent the world entirely from geometry, where every tree, rock and blade of grass within the world is a 3D object the player can both see and interact with.
It simply isn’t practical to place every tree by hand, let alone every blade of grass, and so a system was needed with which artists could define rules for different surface types and the computer could generate and place the appropriate features accordingly. This isn’t entirely new, we have shown off similar technology in all of our previous demos. What we hadn’t anticipated however, was the sheer volume of assets that needed to be created and placed.
As it stands, there are currently 2.3 billion trees which are generated and exist within our world. To arrive at that figure, the tool must first test many more locations to determine which ones are valid based on rules such as surface, slope, texture type, etc. Once a valid location is found, it must then be tested against all other trees/objects to ensure that they do not overlap. This testing takes time and subsequently our terrain generations quickly ballooned from several hours to several days. Try as we might to optimize this process with our existing tool it was simply becoming too burdensome for our artists to make even minor changes without having to endure very long regeneration times.
By moving this functionality to a separate tool, and by taking full advantage of multi threading and clever caching of previously built rules, we have managed to remove much of the generation overhead and waste, reducing our build times from days down to minutes or in the worst case, hours. This gives our artists the freedom they need to experiment with new ideas and maximise the visual quality they can achieve.
As the majority of this work has been rewriting / redeveloping much of our build system’s inner workings, it is probably the least interesting to show visually, but has by far the biggest implications with regards to efficiently authoring and populating our terrain.
This may look like a fairly mundane screenshot but this tool has dramatically improved the way in which we work and in turn, the detail and quality we are able to achieve within Deadstick.
[h3]The Result[/h3]
Beautifully detailed worlds at any altitude no longer limited by scale or build times but instead, only our imagination!
[h2]What’s Next[/h2]
As mentioned above, in parallel to our work on the terrain functionality and tools, much has been done to develop and refine the gameplay systems and interactions within the simulator. Next week we shall release Part 2, in which we start to explore some of these systems in more detail, giving you a glimpse into what you can expect as a day in the life of a pilot within Deadstick!
