MSFS20 Introducing Chief Petty Officer Mike Forney and prepare to be projected to the 1950's

[=rk=]

Yep, pedals are already done! The entire project is already 50k over half a million polys, so the panel may have to stay. The problem is that the original artist added tons of smoothing loops in all sorts of places they weren't needed. I mean, rebuilding these models is a study into obscure cultures, eight faced nuts with rounding loops and contoured edges proves this guy has never seen even a lawnmower engine, much less an aircraft up close, yet his cockpit was very accurate. I added the collective, a wiring harness, and some fire extinguishers, but any tuner knows a rounded bolt is toast, so eight sided? No.

Currently hooking up the instruments and tuning in the glow in the dark dials. I settled on a custom transition for the fuel gauge, because although I have seen red and blue glow, they are, or were not nearly as robust as the most highly radioactive "yellowcake green," so they likely would have faded out after just a few years.
Fun fact, radium has a half life of 1600 years and goes dark due to the breakdown of zinc sulfide. This means your grandfather's watch is still zapping you, even if its lost its glow.

As to simulators, I can't really say. This is all being done in 2020 and unless I get some wild desire to submerge myself into a sim in which I don't even know how to slew my view and start developing, there will likely be a 2020 version. Meanwhile I'm getting screens from K-Max users that want to know when can I hook up logging for 2024, because otherwise it looks great. So I'm kind of torn. To be honest, I kind of leave it up to the model. From my perspective, they kind of get a personality and then build themselves through my hands. So it's kind of up to H-5 where it wants to settle, or Mickey Rooney if you prefer.

I kind of think he'd want to go rescue Harry Brubaker one last time for old times' sake and that would be easier to arrange in 2024, so we'll see.

 

[=rk=]

Most of the instruments are completed at this point. You can see there is a bit if instability, but once airborne it levels out and I can show you around the cockpit. The radio display is set up to switch between a magnetic compass and an hsi style course pointer. The transceiver was current in about 1970, so it only has the frequency decimal tenths. I chose to model it that way because the actual frequencies show up in the tooltips and this would be more authentic. I did have to add swap buttons and those are actually using vertex color, I'm toying around with self illumination in keeping with the theme of the gauge faces. There are no emissive multipliers, it's all pallet color, although I supposed I could set up a material with those emissive multiplier parameters set and then bake that to vertex color, pretty sure that would work.

Coding the animated frequency number dials ended up being a study in informational logic. I use the LLM's like ChatGPT to write a lot of my XML and I knew calculating out that first decimal would be a bit of a challenge, because in the component logic it is both a placeholder to represent a hierarchy of frequencies that are established specifically for communication purposes and also a numerical value which must be multiplied and converted into an integer. Beyond that, any sort of an angular equation would inevitably resolve such that some numbers would line up perfectly in the window, others not. It ended up such that I had to add a registry value for each solution, so when the calculation came to say, position 450 in an animation of length 1000, that by trial and error I had found the proper position to be 473, I could then record that into the registry, reload the aircraft and voila.

ChatGPT did this for me, us and it took a while. It was like it was just throwing spaghetti at the wall and actually seemed dumbed down. An interesting thing happened when I tried the other LLM's. First I tried Meta and I even went to GitHub Copilot that we run through Visual Studio. I thought this one would be unique but it's just more of the same. It's like there is this little shot of brilliance and then dull monotony. I interpret this to be an engagement algorithm. The goal is to expend zero energy to keep me engaged and subscribing and open to suggestions of sponsor products. But when I started bouncing around from platform to platform, it was if they reacted with jealousy. I honestly think one of them spit back a snippet of code I'd shared on a different platform, well I know it did, but XML RPN terms aren't nuanced enough to be unique to a specific conversation. So suffice to say the'll be ready next time I try to play them against each other. Or maybe not, obviously human intuition is the only thing they keep us around for. Anyway within very short order, my old standby CGPT spit out a veritable sonnet of a component, monkeys playing Beethoven and we get to enjoy the performance.

Elsewhere I was pushing the boundaries of emissive representation, I noticed the beacon housing has a fresnel lens and it's large enough to contain a rotating flasher, which I interpreted to be way to diffuse the beam so the whole housing glowed in a pulsing pattern. So not problem making the projector, the proper emissive bits and pieces attached effects, how to get that glow...I had it perfect using a semi transparent inverted dome. It only looked this way once, after I restarted the sim the next time I did a build it was almost painful to look at, you can kind of see the little fluorescence in there. It's just a matter of pushing the boundaries, too little emissive effect and the dome is a dark red cloud that blocks the light, too much and it goes disco ball.

 

[=rk=]

Tuning in the localizer needle and also, I got tired of the rotor torque engine torque needles superimposing in every angle and aspect of operation, so I did a dive into the component template and came up with a couple of simulator variables to better plot the two needles against each other.

 

[=rk=]

Here are pictures taken while developing the downed pilot module. The idea is that you will place the pilot 1 mile northeast of the aircraft and then go rescue him. There are no maps or superimposed arrows, there is an extremely large fluorescent dye marker and also the pilot is waving frantically, so that helps.

Here's how it looks from 3/4 mile out and 3,000 ft altitude, so you kind of have to draw a compromise from getting high enough to see it without getting too far away from the earth to see it.

Here we are at 670 feet and you can just start to make out the pilot.

When you get your sling close enough, he'll climb in and you can haul him back up again. I know his arms are looking a little Gumby, thanks Mixamo.

The cool thing about the module is that it will work with any aircraft, there's even a really crude source file on another thread, but yeah it's a standalone exe made from the AI Objects and Waypoints SimConnect sample project, so the lost pilot would spawn for any user aircraft, you could even swap out the model for another SimObject named Lost_Pilot. Then all you'd need is a little bit of component xml to handle the aircraft side of the rescue and you're good. Get rid of William Holden for someone more contemporary, Tom Cruise maybe but he never gets shot down.

Oh yeah I almost forgot to add, for like making forest fires to go put out you would just click the Z key and wait for the smoke to rise, there will be one placed a mile away to the NE at ground level. Add a little proximity trigger to your bucket at the moment you drained it, tied to visibility of the fire simobject and you have an instant mission.