People often ask what the motivation was behind making Deadstick and with that I have to recall my own flight training. One of the biggest surprises was how little of the course was dedicated to physically learning to fly an aircraft, and how much focus was placed on procedures and the discipline involved to operate one safely. There are many sims that do an incredible job of simulating aircraft, but I always found that the experience of simulating the pilot and airmanship was somewhat lacking. To that end, the concept of Deadstick was born, to give players the experience of what it’s like to be the pilot. [img]{STEAM_CLAN_IMAGE}/32753672/3ee2104af296ac55f342f56875a2969a65c7ec55.png[/img] Those that have been lucky enough to fly in a light aircraft, will also likely have experienced the sheer disappointment of having your trip or lesson cancelled at the last minute due to poor, unforecast or, in my case, British weather! And with good reason. The weather presents one of the biggest risks to general aviation pilots and continues to catch pilots out. Gasco (The General Aviation Safety Council) here in the UK have identified that weather related accidents, be that loss of control in poor weather or controlled flight into terrain, often as a result of hitting rising terrain whilst in cloud, accounted for 20% of all aircraft accidents between 1980 to 2006, with the figures remaining much the same to this day, and similarly echoed around the world. The dangers are numerous - [list] [*] Clouds can quickly lead to spatial disorientation and rapid loss of control. [*] Icing can quickly lead to loss of aerodynamic lift and the ability to maintain altitude/flight. [*] Fog and poor visibility can entirely obscure a runway, leaving you trapped in the air. [*] Wind can wreak havoc on take off and landings, as well as ground speed and fuel planning. [*] Combine wind with mountains, and dangerous downdraughts on the leeward side can pull you down into the terrain. [*] Not to mention the implications of flying into storms. [/list] [img]{STEAM_CLAN_IMAGE}/32753672/412ad0cb483d4c0ce167fd8fd86b7483d366e6d1.gif[/img] Is Deadstick a game or a simulator? I’ll leave that to the community to debate but, if it is a game, then the weather is certainly the enemy! For Deadstick to provide a compelling pilot experience, it is essential for us to create a dynamic and realistic weather model with which the player can interact with, plan around and hopefully avoid. This created its own unique challenges. As other simulators have increasingly turned to grabbing live weather data or giving the user a choice of fixed scenarios, the former isn’t an option for us in our fictional world, and the latter doesn’t give us the challenging unpredictability that we want the player to experience. We have therefore set about creating a plausible dynamic weather model which can create an environment in which the player can experience all of the hazards described above. Those that have had the opportunity to play Deadstick will have encountered some of these before - Thick overcast layers of cloud making it difficult to let down for an approach amongst mountainous terrain, strong crosswinds presenting challenging approaches and dangerous topographic effects when trying to fly over mountains in strong winds. [img]{STEAM_CLAN_IMAGE}/32753672/16b759360e44e151f38e89a5a3ede76814de9df7.png[/img] It isn’t just enough to randomly select from various weather parameters and change them throughout a given playthrough. Whilst weather can be chaotic, over time, pilots can often learn how to interpret or predict how the weather will change based on current conditions. What’s more, certain locations will often have their own region specific weather conditions or micro-climates - be that prevailing winds, a predictable early morning fog, which will burn off as the sun warms the ground, or building afternoon cumulus clouds which could catch out any unsuspecting pilots. These are all elements we wanted to ensure we could somehow capture and feel plausible to the region we are most inspired by when building our fictional world - Alaska. So how exactly does it work? Well, we decided to cheat. We might not have access to live weather data given that we are using a fictional location. However, what we do have access to is years of historical weather data for Alaska, so we opted to analyze that data over a 10 year period, looking at how the weather evolves each day and, from that, build a statistical model with which we could use to represent similar effects. [img]{STEAM_CLAN_IMAGE}/32753672/c97383fe39017dc5b6bcdf9db6f57a4f34b4bbc9.gif[/img] [h3]Introducing Markov Chains[/h3] Markov chains have often been used to predict the weather but, in their simplest form, describe how things move from one state to another using probability. Grossly oversimplifying, a sunny day might have a 60% chance of continuing to remain sunny, a 25% chance of turning to light rain and a 15% chance of storms. We can roll a dice to determine which outcome occurs based on the above probability and then in turn do the same for our new state. The reality is far more complex as we include many more parameters to describe how the weather might evolve over given days, weeks, months and, in turn, how these affect wind, cloud base, temperature, dew point, surface pressure, visibility, etc. All of this is generated using historic real world data to create a statistically accurate model. [h3]Game vs Simulation[/h3] This alone isn’t enough however. Deadstick provides a sandbox environment in which players can take on various flying jobs in an ever changing weather environment. Having an accurate weather model is nice but we also want to be able to prod our weather model, whether that be for good or for evil. In the early game, perhaps we might want to have the weather system lean towards clear sunny skies whilst players learn the ropes of navigation. Similarly, for more experienced players, we might want to have the weather be far more changeable to present more of a challenge, all whilst remaining plausible. Similarly, whilst jobs are often procedurally generated for a unique and dynamic experience, we wanted to add in specific jobs and events with their own exciting challenges. To that end, it was essential to refine our weather system such that it could be influenced or entirely overridden by specific events within Deadstick. This is exactly what we have been working on over the past few months and we think you’ll like the results! [h3]Planning and Forecasting[/h3] Having an accurate weather model is great but we also need our virtual pilots to be able to plan for how best to fly in it. In the real world, this is done in two ways - by looking at the current weather at our local airport, en route and at our destination, and by looking at forecasts where available for all of the above. Actual weather is generally delivered to pilots in the form of METAR’s (Meteorological Aerodrome Reports) and forecasts via TAF’s (Terminal Area Forecasts). These are encoded messages for brevity which pilots are taught to decode in order to best understand a given day’s weather. It was essential for us to be able to represent these in-game with particular attention on the forecasts. Forecasting has improved over the years, but it isn’t an exact science - sometimes forecasts can be very accurate, sometimes less so. This is one of the exciting areas of Deadstick - forcing players to deal with the unknown and unforecast and seeing how they react. One of the benefits of our weather model is that, as we are simulating it, we know exactly how it is going to evolve and can in turn deliver perfect forecasts… But where is the fun in that!! In much the same way that we can use Markov chains to determine how the weather will actually evolve in our simulation, we can also analyse the probabilities for the other states we could be transitioning to and use that to drive our forecast system. If we are feeling particularly nice, we can just forecast the most likely outcomes which would probably represent what will actually happen with the weather. However, for more experienced players, we can start to have fun with the forecasting and instead forecast very plausible changes which could occur, but which ultimately won’t, forcing players to think on their feet! [h3]Fly By the Seat of Your Pants[/h3] [img]{STEAM_CLAN_IMAGE}/32753672/4c99ae95881bb11d3bdfabe5b8254c25236ed612.png[/img] It’s all very well being able to simulate the weather under the hood, but what about being able to feel it? In Deadstick, every aerodynamic surface of the aircraft is broken down into tiny sections and each one is simulated individually. Each section flies through its own parcel of air and, as such, will react uniquely to the air that it encounters, be it rough or smooth. This means that as you fly through the weather, each part of the aircraft is affected differently, making for a much more dynamic environment which truly feels ‘alive’. These parcels of air are aware not only of the global weather, but also the local terrain topography, which can make for some very interesting dangerous flying conditions if flying low in the mountains in high winds. It’s entirely possible to encounter downdraughts which can exceed the climb rate of the aircraft, the consequences of which can be severe! This system has been designed to be as extensible as possible such that, over time, we can continue to add and refine effects. It’s not uncommon in the real world for certain airstrips to have their own unique weather quirks and dangers, particularly where trees are involved close to the runway and the subsequent wind rollover effects that can occur. These are all details we hope to develop and refine over time. [img]{STEAM_CLAN_IMAGE}/32753672/141573e8395dfd547303bced4359be0070f698a8.gif[/img] [h3]Every Cloud Has A Silver Lining[/h3] How many of you have been flying an instrument approach on your simulator of choice in limited visibility, only to find yourself instantly pop through a sheet of cloud into perfect weather - breaking the immersion and destroying the challenge of your zero visibility approach? This is a common problem which occurs when trying to represent clouds using traditional meshes or billboards. These meshes, constructed from triangles or quads, have no thickness and so, whilst it is possible to produce some beautiful looking results, they can only ever be seen ahead of you and will instantly pop away as soon as you pass through the camera. Traditionally, it was too computationally expensive to consider alternative approaches. However, with the advent of modern graphics cards, volumetric rendering effects are now a reality. This allows for true fluffy clouds with realistic depth and lighting which the player can fly in and be fully immersed inside. We don’t just want to stop with cloud and atmospheric effects however. In one of our previous posts, you will have seen the progress we have achieved with our dirt system. The next step is to marry the two together such that the weather can dynamically affect the terrain surfaces you are flying over which, in turn, will have drastic implications on the ground handling of the aircraft. A challenge that always needs to be considered when flying from strips or operating off airport! We hope you’ve enjoyed reading this month’s update and digging into the details on how weather will work in Deadstick. See you next time :)