While Albert is out for his short vacation Zak asked me, Daniel Tuseev, the Project Manager in 1CGS, to help with the new Dev blog note. I think I have lots of stories to tell you, but today I'm writing a couple of words about this pretty disturbing question: what's the difference between TRUE flight models and table-driven FMs? This one should be interesting because I bet you're wondering what's so special in the FM used in IL:2BOS (the improved version of FM in ROF) in comparison to all previously known FMs.
Let's start with dispelling the myth about "old", "table-driven", "scripted", or "rail-guided" flight models of the early days. Those myths turned out to be very sticky despite numerous explanations made by developers of different projects.
Myth #1: All old FMs are scripted
A completely script-based game engine can be used only for limited genres: strategies, check-mates and card games. These are games that have no objects which would need to move constantly and interact with permanently changing external forces. For instance, the recently released "Bad Piggies" mobile game - it is NOT scripted! After you create another wagon, continuous modeling of interconnected and interacting objects begin and every object in the system is described with numerous different equations, whose parameters keep changing with every interaction, every player's action and other events that occur on the wagon's path.
So, if even "Bad Piggies" is not scripted, then "Retaliator" and, of course, all other, more complicated games cannot be called a "scripted simulation".
Myth #2: All old FMs are rail-guided
First of all – what do you mean by rail-guided exactly? It's a flight model where a player using the flight stick directly controls plane's direction and speed. Take free camera view in ROF or DCS - that's how it works basically. To be more specific: tilt the stick to the right for 10%, and the plane will start rolling with 10% of its maximum roll speed. Well, some additional index may be applied, so the roll may start with a certain delay from actual stick's position change (so that 10% speed will be gained over time, not instantly). And same story for the rest of the active axis.
As soon as a flight model receives calculation of forces, influencing the plane (aerodynamic force, gravity, thrust, reaction of support, etc.) and its speed is calculated as a result of these forces every next moment, taking into account planes mass and mass moment of inertia - well, you can celebrate because you've created a non-rail-guided flight model.
For example: "Top Gun" had rail-guided FM. "Janes-ATF", "Flanker 1.0" and "IL-2", and of course "Bad Piggies" - they all have a non-rail-guided flight models.
Myth #3: All old FMs are table-driven
And this is not a myth at all. The thing is that ALL flight models are table-driven: new ones, old ones, and those still in development. The question is what does "table-driven" mean? Every model of a physical object is described with a set of parameters, presented in a table or in a list. The more complicated and deep the model of a physical object is the more master data these tables contain. The next question is at what level is this data defined? The more complicated such a model is, the lower the stage where master data is applied, while parameters at a higher grade are calculated during continuous calculations (ref. to description of "non-scripted" and "non-rail-guided" above) of their changes at every certain moment of time.
In the case of a simple table-driven flight model the data table will define max thrust of the engine depending on altitude and speed. In case of a complicated flight-model, the table will contain: heat generation level produced by combustion of a certain mixture density; temperature; compression index and so on, also aerodynamic indexes that define the force produced by every prop blade's segment on different speeds, altitudes, air temperatures and pressure volumes.
And what's the most important, a complicated flight model after calculating thrust on maximum and optimal engine rates would give same results as a simple FM, which only uses 1 basic table instead of a couple of dozen of tables and several difference equations. So, why would anyone need a complicated FM in this case? Let me tell you.
Advanced Flight Model
The term AFM was born a long time ago, in the early 2000s, when LockON was being developed, when users, despite that the recent flight sims were neither scripted, rail-guided nor table-driven, began to ask for MORE. Basic aircraft parameters such as max speed, altitude, climb rate, turn rate, roll rate and range were reconstructed quite precisely, however players started to feel the need for a much more realistic SENSATION of a virtual flight. The realistic sensation of flying combines the same components as realistic driving experience does in racing games: believable waggling on road bumps, suspension behavior, side slipping, engine revs control, etc. - plus add to that realistic graphics of the plane and the world and plausible sound. So, users started to request all that from new flight sims - new level of feasibility of planes' reaction in various situations such as bucketing, stall, spin, cross coupling (when a plane yaws left and right if you pull up sharply), engine response, correspondence of its rates to real engines' rates and modes, and other small and specific details in which the devil is hiding.
By that moment, LockOn already had very good graphics and sound and by these characteristics it was among the leaders in the genre. To achieve what community was intensively requesting from the in-game flight model the studio hired Andrey Petrovich Solomykin.
By 2004 when LockOn: Flaming Cliffs was released, Andrey had completed developing the new flight model for two ground assault planes Su-25 and Su-25T "Frogfoot". Who knew that working at ground assault planes (“sturmovik” in Russian) would be sign for the future?
So how was Frogfoot in LockOn: Flaming Cliffs different from the previous game in the series - LockOn: Modern Air Combat? It is that it received Advanced Flight Model instead of Standard Flight Model. Here are the general differences between then:
- In SFM all parameters that influence the aerodynamic force are set for the entire plane, and it’s calculated for the whole plane at a time. It used to be common for all flight sims of those days.
+ In AFM an aircraft is separated into numerous parts, and parameters are set individually for every piece; aerodynamic force is calculated separately for each of them, and the plane’s motion results from multiple indexes taken from these forces. It allows it to reproduce such delicate effects as irregular flap vortex on different portions of the airframe, asymmetrical airfoil flow while spinning, etc.. This allowed us to model quite precise reconstruction of how all parts of a real plane interact with air: waggle before stall, dynamic of a spin situation, spinning, reaction to sharp maneuvers, mush when recovering from a power dive, reaction to sharp moves with stick and levers – thanks to the much bigger depth of the FM we managed to make its behavior extremely close to the one in the real life. Also it helped us to get rid of additional physics model that was used to recreate plane beyond stall angle.
- SFM used 3 physics models: first for the flight itself, second for moving on solid horizontal surfaces, third for maneuvers beyond stall angle. However some flight-sims didn’t even have the second model for moving on the ground.
+ AFM allowed having landing gears as an individual and properly working part of the plane. All forces were calculated including its own mass, shock absorbing, momentum of wheels and the absorbers, transitioning of loading from each gear to the plane, etc.. Firstly, it allowed us to remove the annoying switches from one physics model to another, secondly provided realistic taxiing, landing and take-off.
Also, modeling of devices and systems improved drastically: we began to create the planes from lots of separate parts with clamps and modules, like in LEGOs, which later allowed adding details and inner complexity of each system without significant technical difficulties caused by architecture limitations. We got to work with lots of cause and effect relationships between faults: a break down in one system now is able to influence lots of other systems. The way this idea evolved can be tracked, for example, in Ka-50.
After LockOn: Flaming Cliffs released, Andrey decided to forge ahead, and he switched to another project – Rise of Flight. There he was able to further develop his ideas, but this time on a brand new level. AFM structure of ROF was particularly targeted at the damage model. In the earlier version of AFM only the plane itself had realistic damage system and all detaching parts were modeled as rail-guided. But ROF gives the same fair AFM even to those parts that were broken off of the damaged aircraft. This brought up the level realism of the aerial damage system to a whole new unprecedented level.
In addition to that a system of non-rigid links was added to the ROF damage model. As a result, damaged wings, airframe, fins were able to fracture and wobble which allowed to model a realistic avalanche type of damage that accumulated and more and more influenced the entire structure: every fracture could turn into tearing and then breaking over time, and so on. And the visuals improved accordingly, so the way a wing was falling apart looked better than in any other flight sim.
ROF also featured a detailed model of the internal-combustion engine and the aircraft propeller, which turned into a great opportunity to let the player see and hear all the impressive aspects of that system: engine’s cough, cylinders missing a cycle, etc. Engine damage system details including those caused by oil leaks, water leaks and other sorts of damage like bullet holes were added. And we also included the difficulties of starting an engine in cold weather, overheating and overcooling; effects of combustion depending on the mixture quality and the overall condition of the power plant.
But what’s most important is the propeller airflow – for the first time ever in flight sim it has become not an index influencing the rudder, but a fully functional object in the atmosphere and every plane could interact with it properly (not just the plane producing this airflow). And it removed all obstacles to realistic modeling of a propeller-driven aircraft on transitional states like stall and spin and other situations – because peculiarities of controlling a prop-driven plane (various wobbles, swings, pulls, etc.) are significantly defined by the prop airflow. And if we were able to find references for the engines, there was no way to find them for the propeller airflow effect, so this system is 100% our own development.
And finally, some words about Battle of Stalingrad.
Well, speaking of its flight models this is what we have now:
- Huge experience in working with AFM from of SU-25 in LockOn to 40 WWI planes.
- Unique and unrivaled (yet) damage model technology for the airframe including realistic fracturing and natural modeling of detached elements
- Detailed model of a combustion engine allowing representation of various kinds of its typical and faulty functioning.
- Realistic model of an aircraft propeller and propeller airflow able to influence not the plane itself but the atmosphere in the entire game world as well.
Having all that as a base we’ve started to develop more details for the flight models that we needed for WWII planes, trying to make them on the same or (preferably) higher quality level than in ROF. What we have achieved as of yet:
- We studied documents and developed an engine model and its systems allowing automatic control on Soviet and German planes. Engine rates already match the existing references and you can be sure that visually the power plant will be at least as sweet as in ROF.
- A new method of tuning flight models was developed that allows to tune aircraft as precisely as a WWI plane even thou there are much more parameters to pay attention to. Besides, this method allows tuning FMs and engines even more correctly than in ROF.
- Landing gear modeling was reinvented from scratch because what we had with WWI planes was not even close to enough. We will implement key features of cushioning of gear legs, cushioning of pneumatics, dependence on friction and structure of the surface and so on.
- All mechanization will be modeled for correct and realistic functioning of appropriate systems.
- Overall approach to works on damage model was modified. Holes will appear on a particular side that was shot at (if it wasn’t penetrated thru); the quantity of damaged-parts has increased, details of damage model improved in general; more variation added to every part of the airfoil; damage of canopy and instrument panels was added.
- Calculation of aerodynamic force on the airframe and propeller improved. Near sonic speed is possible in WWII setting and how this condition affects the plane was also taken it into account. Same thing about the airscrew which can work at trans-sonic speeds – this aspect was considered as well.
In conclusion I’d like to add that the most of the work on polishing FM technology for the WWII era requirements are now complete. All that remains now is to double check some details and complete the rest of the planes. Three of them are practically ready now.
Also I should add that a realistic flight model is an important part, but not the only part of a flight sim game that makes it realistic and provides you with a SENSATION of flying. Other important features are up-to-date and natural physics of in-game objects, modern immersive graphics, strong positional sound, exciting competitive AI, realistic weapon and damage modeling, a good storyline and a completely user-friendly GUI.
All these modules have already been worked on since the very start of developing IL2:BOS and a significant part of it is complete, however there’s still a lot to do. This time I discussed FMs, next time I’ll pick another topic. As you can imagine, we have a lot to tell you about our work. We’re developing a whole range of aspects simultaneously, we are not stuck on just one thing and as a result it allows us to get a quality, balanced game that you’ll be able to play soon. In general, we’re quite optimistic about the developmental aspects and we’re sure that we’ll deliver you the expected product in time and with all promised features. And of course we hope you’ll love it
ATTENTION! The final decisive poll has been launched in the Founders group - take part in it and vote for your candidate
Go to the poll now.
Also here's a small pack of fresh screenshots showing ground vehicles in the current version
You can discuss the news in this thread