Looks like it is possible after all.
It involves putting a second airport at sea level that consists solely of a network of unmarked "apron routes" arranged in a net pattern - then putting this fake airport at a higher scenery priority than Saba. The taxiways in this invisible airport pull the mesh down to sea level and allow Martin's constructed landscape for the airport to look as it was intended.
After that, its just a question of getting the Islanders' CFG correct, since their turning circle is a little too wide and their braking somewhat unenthusiastic at Saba.
I called the fake airport "BABA". It consists of just the afcad installed as a separate airport. This technique has already been used to create oilrigs that don't pull the sea up beneath them, and other off-airport landable areas. The folder for BABA could contain any number of fake Afcads to pull down the terrain at other airports in future.
Note that it must also trace the outline of the runway so that the turning circles for aircraft (traced out using apron routes in my AFCAD) doesn't pull up the terrain at each end.
My afcad also doesn't connect the middle of the runway to the apron. This is to stop the AI aircraft from departing from the mid point (on short runways they do this because they're close to the end of the runway and the AI engine decides that it is okay for them not to backtrack before departing). The gates in my AFCAD also use the hook mechanism (developed by Tom Gibson) that stops aircraft from pushing back on departure. Planes should taxy into their spots and on departure, start and go straight out.
The AFCAD still needs a few tests and I'll submit the files to Martin first.
Gates at Saba - working.
Mage
I sent Martin some file fixes also for SABA. My approach was a little different to maintain the mesh.
I set the hard floor of the dummy to 196.1 instead of 196.8 and then built the AFCAD to 196.8. this maintained the mesh for the most part if the max_vertex is defaulted to 19.
I also added a taxilink down the border of the runway so planes would taxi toward the end keeping a midfield takeoff from occuring.
The only AI Plane that can stop is the Cessna 172. I extended the runways (invisible) to force a touchdown at the "X" but brake settings on AI are to terrible.
I sent Martin some file fixes also for SABA. My approach was a little different to maintain the mesh.
I set the hard floor of the dummy to 196.1 instead of 196.8 and then built the AFCAD to 196.8. this maintained the mesh for the most part if the max_vertex is defaulted to 19.
I also added a taxilink down the border of the runway so planes would taxi toward the end keeping a midfield takeoff from occuring.
The only AI Plane that can stop is the Cessna 172. I extended the runways (invisible) to force a touchdown at the "X" but brake settings on AI are to terrible.
Hello Jim -
I didn't notice any problems with the floor set at sea level. I get the feeling Martin modelled that entire small headland and chamfered it into the hill rising away to the SW. My vertex level is currently set to 20, lowered temporarily from 21 (which is the setting used for 20 meter meshes). I've no idea whether the setting has any meaning for less detailed meshes or whether Martin uses a particularly detailed mesh at 20 meters or better. I sent the files on to Martin earlier today (I'm in the UK). They're not perfect, I lengthened the runway a tad too much and it needs about 20 meters taken off of it.
Since I have a 19/20 meter mesh for the Alps, I'll be putting that vertex value back up to 21!
I assume you chose an intermediate height for the dummy so that people could have something solid to collide with!
The only value these files of mine might have is if anyone wants AI, but until the aircraft landing behavior can be amended to get Islanders and Twotters doing short stops at Saba it is never going to look very realistic.
Heck, if I can get my Twotter stopped by the taxiway and just turn straight off to park, AI should manage it! Having tried the Barths 10 approach too I only wish I'd gotten that DGAC certificate all those years ago. Great airports!
I didn't notice any problems with the floor set at sea level. I get the feeling Martin modelled that entire small headland and chamfered it into the hill rising away to the SW. My vertex level is currently set to 20, lowered temporarily from 21 (which is the setting used for 20 meter meshes). I've no idea whether the setting has any meaning for less detailed meshes or whether Martin uses a particularly detailed mesh at 20 meters or better. I sent the files on to Martin earlier today (I'm in the UK). They're not perfect, I lengthened the runway a tad too much and it needs about 20 meters taken off of it.
Since I have a 19/20 meter mesh for the Alps, I'll be putting that vertex value back up to 21!
I assume you chose an intermediate height for the dummy so that people could have something solid to collide with!
The only value these files of mine might have is if anyone wants AI, but until the aircraft landing behavior can be amended to get Islanders and Twotters doing short stops at Saba it is never going to look very realistic.
Heck, if I can get my Twotter stopped by the taxiway and just turn straight off to park, AI should manage it! Having tried the Barths 10 approach too I only wish I'd gotten that DGAC certificate all those years ago. Great airports!
Ah, I tried 1.2 for the toebrakes in the Islander and it made no difference at all - maybe they were getting worn! 
I'll have another dig around the CFG when I get the time.
I think at least part of the problem is not that the braking strength is insufficient, more like the aircraft just don't start braking/reverse pitch as soon as they could. If there's some hardcoded rule for AI that they only pop spoilers/apply brakes/reverse when all the gear is on the ground, that might have something to do with it.
By the way, Jim. Did you consider an angled approach for Saba? I'm guessing not, since it was intended as user-only. In reality the crews fly at the hill and turn fairly short final if they're landing to the SE, the reason being to stay out of the bad air that's close to Saba. I think they usually turn at the small rocky outcrop that Martin modeled. I visited there one time (as pax) and that was the approach they flew.
I'll have another dig around the CFG when I get the time.
I think at least part of the problem is not that the braking strength is insufficient, more like the aircraft just don't start braking/reverse pitch as soon as they could. If there's some hardcoded rule for AI that they only pop spoilers/apply brakes/reverse when all the gear is on the ground, that might have something to do with it.
By the way, Jim. Did you consider an angled approach for Saba? I'm guessing not, since it was intended as user-only. In reality the crews fly at the hill and turn fairly short final if they're landing to the SE, the reason being to stay out of the bad air that's close to Saba. I think they usually turn at the small rocky outcrop that Martin modeled. I visited there one time (as pax) and that was the approach they flew.
- Clement312
- Posts: 7
Sure -
I put this up in the first post but the image got linked to another Imageshack user and it was removed.
Progress is good. Martin has suggested a technique that does away with needing a separate scenery layer - which makes this addition simpler to use. I'm glad to say it works and is a one-file addition to the scenery (2 if you also count an AFCAD with parking spots).
Note that you'll still have the problem of getting AI planes stopped within the length of the runway - its very short. Obviously Martin's scenery does its part and now the FDE tweakers have to do theirs!
I put this up in the first post but the image got linked to another Imageshack user and it was removed.
Progress is good. Martin has suggested a technique that does away with needing a separate scenery layer - which makes this addition simpler to use. I'm glad to say it works and is a one-file addition to the scenery (2 if you also count an AFCAD with parking spots).
Note that you'll still have the problem of getting AI planes stopped within the length of the runway - its very short. Obviously Martin's scenery does its part and now the FDE tweakers have to do theirs!
MageBy the way, Jim. Did you consider an angled approach for Saba?
If SABA becomes a reality for AI in the TNCM scenery then a GPS approach will have to be written so the airport does not close if real weather falls below 3 mile visibility. AI can't land on a IFR FP if some sort of approach does not exsit in the database.
If I curve a GPS approach it only occurs when weather is IMC. If weather is VMC AI default back to a VFR hard code approach with a FAF of 2000 ft.
Fake ILS is the only way to fool FS to curve the approach when weather is VMC.
Hi Jim - I actually realized that it was a dumb question when I thought for a moment. If the airport isn't used for AI there's little point making it super-accurate in every respect, because originally the a/c would do these arty approaches and then vanish the moment they came to a halt (in mid air 200 feet beyond the end of the visible runway like a Bugs Bunny cartoon character but without the plummet).
Looks like the option is a fake ILS pointing straight at the rock. On my one visit years ago the weather was fairly typical, puffy clouds, about 4000' bases, 50nm vis and flying straight at the biggest bit of rock in the area in perfect VMC.
I encountered pretty bad vis in the Caribbean from time to time but seldom below 3nm, more like about 10-15 miles, enough to hide any nearby islands, and enough to make a GPS very useful. With a GPS approach you'd keep Saba nominally open. At present the AI vectors itself onto a very long final. An approach straight to the hill might actually get traffic tracking fairly realistically into Saba from further out because the 30-degree fan each side encompasses St Martin and gets fairly close to St barths too.
I'm assuming the approach into 30 is more mundane.
Looks like the option is a fake ILS pointing straight at the rock. On my one visit years ago the weather was fairly typical, puffy clouds, about 4000' bases, 50nm vis and flying straight at the biggest bit of rock in the area in perfect VMC.
I encountered pretty bad vis in the Caribbean from time to time but seldom below 3nm, more like about 10-15 miles, enough to hide any nearby islands, and enough to make a GPS very useful. With a GPS approach you'd keep Saba nominally open. At present the AI vectors itself onto a very long final. An approach straight to the hill might actually get traffic tracking fairly realistically into Saba from further out because the 30-degree fan each side encompasses St Martin and gets fairly close to St barths too.
I'm assuming the approach into 30 is more mundane.
- RobdeVries
- Posts: 32
Hi Rob -
Although I can't speak for Martin & Jim, the revisions I made are with them (I'm assisting Jim regarding aircraft FDEs, and Jim is looking at a combination of that and the AFCAD options available). I didn't really want to post anything straight on here because that opens a whole can of worms regarding support - "whose version of the changes are you using?" kind of questions. I'd rather Martin & Jim were happy first and the changes were officially sanctioned and released as a small update. It just keeps things tidy.
My own minor supporting role has been somewhat slower because of the UTIL.DLL problem I've had with my installation of St Barths. That stopped me cold for about 5 days and possibly slowed Jim's own involvement because I couldn't offer useful and timely feedback to him.
There's also the decision of what to do regarding aircraft FDEs. We can get aircraft stopping on the runway at SABA just fine, but any changes we make can also affect the tricky approaches at Barths. Stopping an aircraft short on the runway also affects the flow at St Maarten itself when those same planes land there. If an AI plane can leave at the first exit from r/w 09 it has to taxy against the flow all the way to the old terminal, so there's some "sweet spot" to find where the aircraft stop fine at Barths & Saba but don't mess up the flow at St Maarten by using the first exit. I know that AI FDEs are theoretically not Martin's concern and should be a third party consideration, but I get a sense that Jim in particular (being the specialist in this area) would like to get the specifics of the change as good as they can be.
Its easy enough (so it seems) to get Saba AI working, but Saba and Barths airfields and the flight dynamics of the aircraft that visit are very closely intertwined. There's a thread elsewhere on this board about the traffic approaching Barths r/w 10 passing through the hill, and so there's probably a side agenda with the FDEs where any changes we do for Saba don't make AI behavior worse (and hopefully improve it) at Barths, thereby solving that issue too. To fly the approaches at Barths, AI aircraft have to be on IFR plans to use the phantom "ILS" and avoid the hill (the glideslope is steep). Fernando's Dornier and Henry's Twin Otter can't fly that angle as they come, so their FDEs have to change to permit the AI aircraft to achieve that steep profile. Any changes for Barths then also need to be looked at for Saba - and so it goes back & forth.
Of course, everybody's AI setup is different so any final AFCAD and FDE combination needs to be as close to "one size fits all" as it can be. In the extreme case, for users who aren't AI tweakers and who don't change their aircraft flight dynamics, the scenery must still work acceptably in their situation. Martin's sceneries sell outside of the AI community, after all.
For someone that can't speak for Martin or Jim I've certainly written a lot, (!) but that reflects just how surprising it was for me to see all the ramifications of a supposedly minor change.
Although I can't speak for Martin & Jim, the revisions I made are with them (I'm assisting Jim regarding aircraft FDEs, and Jim is looking at a combination of that and the AFCAD options available). I didn't really want to post anything straight on here because that opens a whole can of worms regarding support - "whose version of the changes are you using?" kind of questions. I'd rather Martin & Jim were happy first and the changes were officially sanctioned and released as a small update. It just keeps things tidy.
My own minor supporting role has been somewhat slower because of the UTIL.DLL problem I've had with my installation of St Barths. That stopped me cold for about 5 days and possibly slowed Jim's own involvement because I couldn't offer useful and timely feedback to him.
There's also the decision of what to do regarding aircraft FDEs. We can get aircraft stopping on the runway at SABA just fine, but any changes we make can also affect the tricky approaches at Barths. Stopping an aircraft short on the runway also affects the flow at St Maarten itself when those same planes land there. If an AI plane can leave at the first exit from r/w 09 it has to taxy against the flow all the way to the old terminal, so there's some "sweet spot" to find where the aircraft stop fine at Barths & Saba but don't mess up the flow at St Maarten by using the first exit. I know that AI FDEs are theoretically not Martin's concern and should be a third party consideration, but I get a sense that Jim in particular (being the specialist in this area) would like to get the specifics of the change as good as they can be.
Its easy enough (so it seems) to get Saba AI working, but Saba and Barths airfields and the flight dynamics of the aircraft that visit are very closely intertwined. There's a thread elsewhere on this board about the traffic approaching Barths r/w 10 passing through the hill, and so there's probably a side agenda with the FDEs where any changes we do for Saba don't make AI behavior worse (and hopefully improve it) at Barths, thereby solving that issue too. To fly the approaches at Barths, AI aircraft have to be on IFR plans to use the phantom "ILS" and avoid the hill (the glideslope is steep). Fernando's Dornier and Henry's Twin Otter can't fly that angle as they come, so their FDEs have to change to permit the AI aircraft to achieve that steep profile. Any changes for Barths then also need to be looked at for Saba - and so it goes back & forth.
Of course, everybody's AI setup is different so any final AFCAD and FDE combination needs to be as close to "one size fits all" as it can be. In the extreme case, for users who aren't AI tweakers and who don't change their aircraft flight dynamics, the scenery must still work acceptably in their situation. Martin's sceneries sell outside of the AI community, after all.
For someone that can't speak for Martin or Jim I've certainly written a lot, (!) but that reflects just how surprising it was for me to see all the ramifications of a supposedly minor change.
- RobdeVries
- Posts: 32
Hi Mage,
First of all thank you for taking the time to reply to my question in such elaborate way. It's much appreciated.
My stance in this one will be to just wait and let the masters do their job.
I understand the intricacies with the AI. You said:
All the best,
Rob
First of all thank you for taking the time to reply to my question in such elaborate way. It's much appreciated.
My stance in this one will be to just wait and let the masters do their job.
I understand the intricacies with the AI. You said:
It's probably not a feasible solution but isn't it possible make a workaround by adapting the FDE for a particular aircraft for STOL capabilities and making it appear at some point to land at Saba and disappear after take off from Saba at some point . And let another copy of the same aircraft (with appropriate FDE) do the same at St. Barth's? And this in a coordinated fashion so that it appears to be a continuous flight between islands?Its easy enough (so it seems) to get Saba AI working, but Saba and Barths airfields and the flight dynamics of the aircraft that visit are very closely intertwined
All the best,
Rob