Roman Design

Now that my CYYJ Victoria is released, I'm once again looking for feedback on what airport to do next. Please let me know if you are aware of other developers working on Canadian airports, what is expected to release soon, what is missing, and cast your vote for what you would like to see next. I Would like to identify airport targets that would be in demand, that are either not done or being worked on by anyone else, or exists in much inferior quality - and definitely not anything that's already available on MSFS Marketplace. I prefer International/Regional airports that have jetways and can accept at least smaller airliners like 737 or ATR42 etc. Let me know what you think...

CYYJ Victoria International Airport is now available on MSFS Marketplace. Please be aware though that this is the first released version and not the latest update. MSFS QA/Release cycle takes weeks, during which time I can’t submit an update without restarting the cycle, so any updates will appear in the marketplace up to 6 weeks later than on my website, which is why I recommend my website if you want to get updates right away. Of course, nothing beats the convenience of the MSFS Marketplace, and the QA/Release cycle is there for a reason, so use the platform of your choice. Eventually, the Marketplace version will be the same. Just be aware that updates take time to get there.

Such test won't show much: of course any custom airport would be much more complicated than a default one - it always has a higher polygon count and more textures, because it provides more details, so it takes more computational resources. So in pure FPS it will always be somewhat lower than defailt. So, not much of a test. The question is if it still within a headroom that provide smooth flying experience, and here is where the level of optimization of the Asobo photogrammetry comes to play. So if the default area mesh brings your PC to the border of it's abilities, a custom airport can easily bring it over the threshold. But if the area is well-optimized, a custom airport may not bring a noticeable difference in performance. By comparison, photogrammetry area has a much higher polygon count than a custom airport, so the area and photogrammetry optimization level can affect performance a lot. I can suggest a simpler test: disable photogrammetry, then test default airport vs. my airport. You would probably still see a couple of FPS drop of course, just because there's more detail. But likely the performance will still be very smooth because there's a lot of headroom without photogrammetry, which you will see impacts performance by a huge margin when you enable it. I don't see any performance issues in 4K and in VR, everything is very smooth. However I do have 7950X3D CPU and 4090GPU, which is basically the best you can get now, running on DDR5 6000 RAM. So it's not surprising it runs great. However things are far from perfect in VR in KJFK and over Manhattan.

I don't know for certain, but it does make sense, because they updated photogrammetry, and it's the photogrammetry optimization that was the source of many performance problems before, according to Asobo interviews. They optimized large cities many times to increase performance. So I suspect when Canada photogrammetry got expanded and denser in the update, it was not optimized enough. In fact I'm certain it's not well-optimized because when I was working on CYYJ I found there are very large amount of "polygonized trees" - faceted green blobs thet did come with photogrammetry. But they are supposed to be virtually eliminated by Asobo AI, so that only buildings remain. They look ugly in photogrammetry, and AI is supposed to replace them with nice autogenerated trees. However it wasn't the case in many places in that area. There are plenty of photogrammetry tree areas everywhere that just waste polygons. I can't provide options to cut back on objects, as there's no way to do it in the MSFS Marketplace anyway. But it won't help. Compared with photogrammetry that's everywhere around, it's a miniscule number of polygons. It won't make any preceivable difference.

I don't really see any significant FPS hit from the airport. I suspect that the most of the FPS hit you are seeing is actually coming from the photogrammetry surrounding the area. The polygon count is what's loading your CPU and creates stutters. I will be releasing another update, however there's nothing I can really further optimize for performance. All objects are done with 3 LODs (level of details) and it's all really very efficient. I believe it's as efficient as can be, considering there is interior modelling etc. As low polygon count as possible without sacrificing quality too much. I think overall there's a lower polygon density in my scenery than the phorogrammetry mesh surrounding the airport. Texture use is also very efficient. I'm not just slapping 4K PBR textures on everything - there are only a few of 4K textures where they are needed, the rest are sized according to what they cover. Objects share textures a lot, so it's very memory-efficient. For example I don't create a a 4K texture for every building, even if the building uses similar materials as other buildings. Objects share materials. You can see by the package size that it's very modest. Most users say my airports compare well in terms of FPS impact. I suggest you compare to a simiarly-sized airport located in heavy photogrammetry area. You will probably see the same or worse performance. That's why I always tweak my MSFS to perform well taking off in my target plane from JFK and flying over Manhattan in dense clouds, for example. If it can handle that, it will hande everything. Otherwise you could be fine in rural areas, then hit stutters on landing at any decent airport in a photogrammetry area, because that's the most demanding scenario.

Nice videos, I didn't find those when working ont he scenery, but got others, from the cockpit - though with less detail. Looks like I was pretty close.

That seems to be a good solution. I can do a slighly better job, looking at their video, so if I decide to make the pack, I may just do that... As "romandesign-" comes before "samscene-" the exclusion file would be necessary, as their scenery would override my exclusions.

Now that poses a dilemma: if I model the Victoria harbour seaplane terminal, it will conflict with SamScene, but if I don't - people will be missing it in an otherwise nice package. Not sure what to do... MSFS Marketplace doesn't support any optional packages - it has to be a single package.

I'll look into that one...

Well, if he can get me a 3D-model file that can be legally converted and used in my scenery - I will be happy to add it 🙂

Hi everyone, I came here to write a post about yesterday's CYYJ release but it looks like this topic is already here and is very active, so why create another one? Anyway, thanks everyone for your kind words. This project took an immense amount of work, much more than expected, over the last 11 months. So I hope enough people enjoy it, so I can keep doing Canadian airports. I love doing that. To address a few points mentioned in this thread: - The download issue is resolved, everyone gets their download. Don't hesitate to contact me if you have any problems, but things are working great now. I missed a glitch yesterday - that's what you get when you work until 2am for the last 5 days straight including the long weekend, to try and finish the release... - I had local sources that provided me photos of the area, the airport, the new terminal building, the aprons and buildings etc. to add to the info I sourced. I'm not naming them because not sure they'd want me to, but you know who you are and thank you so much for your help! As a result, I got as much detail as I could as precise as I could, within the limits of MSFS and common sense (well, sometimes beyond the common sense...). I won't directly compare with the competitor's airport, it's not my place to do that. But there are many details that are not immediately noticeable, here are just a few examples showing the level of detail I was aiming for: The police sruisers have authentic livery of "Victoria Police" with correct decals etc. The taxis are the most common model at the airport (Prius) and have authentic company livery including their phone etc. No school buses at the ariport 🙂 On the other hand, there are BC Transit buses at the airpost and close to water terminal that have authentic livery including all the signage. ATC tower has interior with the actual CYYJ diagrams showing on the monitors There are detailed RVs at the RV rental business' parking lot, and there are a few BC Transit buses there - not sure why, but that's what real photos show, so there they are. There are deicing vehicles with CYYJ livery at the airport and inside a MegaDome hangar Animated Canadian flags actually respond to the wind direction, like a windsock, they are always face the wind correctly There are models for seats and structures for baseball park on approach and it's night-lighted - just because it would look nice when landing... Water Aerodrome has a start position at the dock (as pictured in this topic earlier) and Aerosoft Twin Otter in Harbour Air livery looks almost identical in front of my two static HA Twin Otters, so it would feel right at home. I placed 2 buoys hopefully at correct locations on the water - by analysing several landing videos. The dock start is also connected to an invisible "water runway" and via a taxiway road to the main airport network. So MSFS could spawn planes landing and taxiing. You can also ask ATC after landing to provide taxi instructions to the park and it should get you there, including the blue ribbon if desired, though I haven't tested that. I created many of the roof textures from flying my drone at some local commercial buildings after doing the research on Google Maps finding promising roods. I found most available textures inadequate or too bland, so made a couple of local trips and did some photos, and made seamless textures in Photoshop, then created normal, roughness and ambient occlusion maps and combined them for PBR materials. The list can go on, but you get the idea... - I am thinking about doing a pack of Vancouver Island helipads as there are a few and it would be a lovely area for realistic helicopter flying. There is a set of 4 helipads in Victoria harbour, there's another one next to a lighthouse on a rocky island, about 4 hospital helipads etc. Thre's also a small airport with 3 runways, and a couple of seaplane locations. Not sure there's enough demand to cover development time though. But I'm thinking about it.

That's very impressive. They don't use MSFS though. Maybe it's the corporate version of XPlane or something else. And they have custom Varjo setup with some extra cameras that do amazing hand tracking with full 3D model of your fingers rotating knobs. Such presition is not possible with Leap Motion at least. And hand tracking is not supported in MSFS at all, with OpenXR Toolkit it is possilbe but there are still problems with it. I'm involved with it from the beginning and testing betas. Because it has to overcome MSFS VR implementation problems it's not exacly possible to that degree yet. Regardless, you can model a cockpit with all switches and use virtual displays in the similar way, and just use hand tracking for visually seeing where your hand is, not the actual interaction, it should work fairly well. But building a whole cockpit and a motion rig that can support all that weight is way above most people's budget, certainly mine 🙂 My rig is compact and on a very tight budget. Its value is an amazing bang for the buck ratio 🙂 Now, if someone has a budget for 6DOF rig that can support a full cockpit weight - great gor him, but just the motors/actuators needed would cost several thounsand dollars. The whole rig would cost anywhere from 6 to 15K in parts/materials only. Theirs are more expensive of course, but they are a business and the rig is cerified.

VR users are only about 2% of gamers, and I guess the market is just not there for something specialized like that. That's the use case I haven't though of. Yes, that would be very useful. Even let's say, if my encoder box and switch box would be visible in VR they would be easier to work with or could be made more elaborate (though not very realistic). Or Radio/AP panels could just be replaced with your setup, where you can see your hand operating these, while glass panels could be left in VR. That actually sounds like a very useable idea. With a bit of work one could integrate his compact Ratio/AP/Switch panel setup into most aircraft that would look plausible in VR. Throttles could be added that way too, along with yoke/joystick. There are many things that don't need very hi-res display for. And this would be especially valuable for people who already have expensive panels set up (like Logitech radio/AP etc. and large throttle quadrant setups). This would let you integrate your existing setups that you spend many hundreds of dollars for, into VR. I hope when the next generation of HMDs introduce high quality passthrough feeds, this will become more common and easy to do.

I didn't mean that the price is generally unreasonable. It's just unreasonable for me 🙂 This kind of solution is mostly used for corporate and training applications, where it makes a lot of sense. We see that it makes sense at least for some high-end simulation fans, who spend many thousands on their setup, though I don't really get why: Think about how strong people's reactuons are for not perfectly crisp glass panels with DLSS in MSFS SU10 beta, or when recuding TAA to 70%. And then think how even the best glass panel would look after being displayed on external monitors and then passed through cameras, then post-processed and distorted to fit the HMD FOV. Zed mini vertical resolution is not great: 4416 x 1242 but only at 15 fps which is unusable, and with 30fps (which is also far from perfect) it's just HD 1080p x 2 eyes (3840 x 1080). My Reverb G2, as all the best HMDs are 4K. Add resizing and berrel correction distortion etc. and you can see the picture (or not see it LOL). I'm sure it will be far worse that what you'd get in VR. It may barely be OK for steam gauges or military simmers who don't care that much about instrumentation, but then why build expensive and time-consuming panel and limit yourself to a single plane, and then get a low-res image in your HMD? Plus, generally, upcoming Meta and Pico HMDs specifically state high-quality color pass-through cameras for augmented reality. But I'm sure they won't be any better in terms of resolution. Enough for typical augmented reality applications, but not enough for passing through precise instrumentation and other small and high-resolution objects. I think hand tracking makes much more sense. Perfect sense in fact. In my case you would interact with a virtual cockpit and some strategically (but not always realistically) placed hardware, but if you really want to, and you're OK with being limited to a single plane, you could build 1:1 cockpit control replica, and use hand tracking just for seeng where your hand is, operating real knobs and switches placed at same locations, while seeing crisp and realistic instrumentation in VR. In a way that would also be augmented reality.

Oh, I don't deny it's a very cool technology 🙂 However this camera is worth almost as much as my whole motion rig 🙂 The question is - is it worth it for someone's particular goals? Not for me. Although I've never seen it, I doubt the physical glass cockpit would be sharp enough, passed through the cameras, compared to what we get in our virtual cockpits. The plus side is of course the ability to build a true-to-life physical cockpit to interact with, but it kind of defeats the purpose of VR, because you get a inescapingly reduced quality passthrough feed of reality that does look a bit "pasted" over the VR picture which can't be good for immersion, and you are limited by the very specific cockpit that you built. Why build a whole cockpit if you already wear an HMD, where it can be recreated in perfection? In VR there's no limit. Yes, it's cool to see your hand on the joystick. But for "real" interaction with the cockpit, hand tracking should be functional very soon, it's almost there already, and every cockpit elment would then be properly reachable without the use of controllers. A mint condition Leap Motion controller can be found for less than $60. For me the goal is not building the ultimate possible cockpit - I'm not rich enough for that, as there's no limit. The goal is the get the ultimate experience I can get for spending as little money as possible. I beileve that for me I got hte best price/performance point at under $500 budget. The experience can be better of course, but spending $5000 will not get you a 10x experience, but maybe 10% better. Spending $50,000 will get you 30% better, etc. It's the law of diminishing returns. But to each his own, and everybody's budgets and ambitions are different. I couldn't justify spending more than $1000 on a grown-up toy like this. But some people spend tens of thousands on full-cockpit simulators.