Is this 4k? I don’t know why but when I play (1080p), the orbs become rectangular-ish the further they are… And yes they don’t really get smaller in size and just become “blurry” and out of focus.
Disorienting to some extent, but definitely better than update-5.
Then we might misunderstand each others points, or the language barrier playing tricks on us.
Did you understand what I wrote to you in those posts?
Let’s try again.
Which is exactly why I tried explaining to you, given the wildly different visibility ranges, you’ll have this effect at a very close range with the old lamps, and up to a much bigger distance with greater lamp visibility. If the lamps were to scale linearly, we’d either have lamps vanishing after just a few miles, or lamps the size of houses when zooming in. That’s simply not how it works.
Look at my screenshots, as well as CptLucky8’s posts. Feel free to measure the pixel size in those.
It’s simply the same size. Is that something you dispute?
Look in my 2nd screenshot. you have lamps in the distance nearly the size as houses, exactly the same angular size as the ones much closer. (which is something I’ve remarked on ever since those got implemented mid-alpha.)
I’m sure of how you interpret your prints, but looking closer, albeit heavily compressed, it’s very easy to see the extremely short lamp visibility range, with lamps fading away after just a mile or two.
But the “same size” phenomenon is clearly visible.
With longer lamp visibility range, you will have exponentially more same sized lamps until the fading starts in the distance, and with short visibility range, you’ll get this effect much closer to you, as your pictures clearly shows.
I’m not sure if we’re talking about two different things here, but try to at least respond to the parts of my post you either don’t understand, or disagree with regarding the lamp scaling shown in both mine and CptLucky8’s posts.
Yes that’s a 4k screenshot from before Update 5 to show we have the same sized bulbs now since release (albeit much longer visibility)
Could you post a screenshot?
Different renderscales can induse some squareness due to sharpening/downsampling, but I’m not certain that is the case in your situation.
Look, from your own print, if I scale them equally, the release build is way smaller, and your print is already zoomed in.
See what I’m saying?
If they were half the size, I’d call 'em way smaller…? But take a look, they’re practically the same size, but I can see with blurry resample you’ll catch some of the glowing halo like you did. For a fair comparison, maintain pixel-scale.
Here are the exact measurements of my picture comparison between the 3 versions.
At pixel level, I even counted the pixels for you to make it easier for you to see/understand the actual size difference, instead of some blurry mess due to resampling.
And here you can see lamps getting smaller with different zooms/distances in the latest build as well.
Although at a much longer distance than pre update 5, as I explained in earlier posts.
Here is a series of images to show what is creating the illusion of a bigger size, i.e. white clipping the halo.
The source image is one of the texture from XP11. It is has a bright center and a falloff. The image shows the original in the middle and the effect of just adjusting the brightness level:
NB: This is the same light bulb texture in all cases but showing as 5 different sizes in appearance.
Since it can’t be whiter than white, it will clip when adding brightness and make the semi transparent halo less transparent and whiter, adding to the ‘core’ size.
What is happening in update 7 like I’m showing is that they are increasing brightness nearby, and decreasing it farther (most likely nor inc/dec at middle distance). this gives the illusion of brighter lights close by and dimmer lights farther away, but this is wrong. They shouldn’t change brightness, they should only change size.
Now compare the light bulb above with the comparison with update 3 and 7 at RJTT:
Update 3 was still showing the halo not far away (this is a crop from the lower right of the original image) while update 7 is white-clipping.
I like the way you do your zoom-in boxes!!
Could the decrease in brightness done by devs to simulate atmospheric effect to some extent?
But yeah they don’t need to decrease the brightness, only reducing the clipping (the core light area) the further lights away, so lights appear smaller and “losing brightness”, while actually the brightness is similar, but the size gets smaller.
I believe this is wrong too! Look at the rail lamps in the XP11 screenshot and how the halo is reducing in size too. This looks much better.
Now look at the 5 bulbs I’ve posted above and how color shifts with -100 brightness and relate this to the red color shifting at the edges of the runway in FS2020…
There is a difference changing size and changing brightness also because the shader code is running in a linear space (the textures and screenshots are in a gamma space). This makes maths a little bit different for these and more complex, whereas changing size is simpler, done in hardware at the silicon level (not the shader code level) and is closer to the truth. (so it seems)
There are certainly good reasons they’re doing it the way they do, yet, this doesn’t preclude investigating other ways and how they compare visually and performance wise. It is also possible they just didn’t put additional thought on the whole night lighting for other reasons (no previous experience with realistic aeronautical lighting but experience in close-up lighting only over limited distance, or art director pushing for photo looking instead of pilot looking lights, or any other speculation you could add to this list).
In any case, it is certain they’ve tried to respond to community demand in a hurry with the integration of the ‘night lightning enhancements’ since release, but even if I can’t tell whether it was better pre-release because of the NDA I can at least tell it is not the same (from the look of the publicly available pre-release screenshots) and many seem to want pre-release lighting back (at least the point light part of it).
I must be missing something, why is it wrong to reduce brightness for more distant lights? That is after all how it works in real life.
I’m no expert on any of this and have been reading along with some interest. So I may be totally wrong here but I would guess it’s because the lights are a constant brightness no matter how far away you are from them. What changes is the viewers perception of scale based on the distance and the diffusion/refraction of the light in the atmosphere before it reaches the viewers eyes.
What you are hinting at is that “brightness” is an ambiguous term, and it is.
In more precise terms, radiant intensity and luminous flux do not depend on distance, but irradiance and illuminance do. I think what the sim depicts (or should depict) for lights is illuminance?
Some useful infos to get started:
Yeah that is interesting… but what does it have to do with this?
Are you talking about how the sim (+ GPU + monitor) should render the less bright more distant lights, is that your point?
My point is in the video. If you change the colors (indirectly in changing the brightness) you’re not averaging the same as if your reducing the size. It is clearly explained in the video at 1m 22s, at 2m 19s, at 3m 07s.
I understand it is a little difficult if you’re not into computer graphics programming and shader code, but just looking at the runway at EGNX I’ve posted above, you can see how different it is when changing size vs brightness, and I personally find XP11 point lights more realistic and they do not show the same flaws than the ones in FS2020.
I believe at this stage, if not looking for the details behind the maths with light in CG, it would be a matter of taste only and the only way to solve this is having a Legacy/Modern choice for lighting as we do have for the flight model…
Yeah but what averaging are you talking about.
Take a step back – we agree irradiance and illuminance decreases with distance, i.e., fewer photons hit your eyes (irradiance) and therefore your “perception of brightness” (illuminance) is lower when the light is more distant?
I think, reading between the lines, that what you are talking about is how that reduced illuminance of more distant lights should be reflected in the rendered image, and that issue might well be non-trivial. Is that the point you are trying to make?
You achieve reduced luminance in the distance in shrinking down a point sprite image of a light bulb. That is the point I’m trying to make. If you just approach lighting in painting white pixels, sure, you’d change brightness. But in painting sprite with a semi transparent halo, you let the hardware filtering the texture at smaller size do the magic for you… And this will give a correct size appearance based on the distance too.
Again, download, open in a paint program, then zoom the XP11 and the FS2020 images of the EGNX runway and compare, measure RGB levels, etc… Then zoom out and compare the overall distribution of lightness from front to back of the runway etc… This is self-evident to me but I might be alone looking things the way I see them (of course only me can see what I see anyhow LOL)
(like Legacy / Modern flight model setting)
Regardless it is more accurate perceptually, some prefer left, some other right…
PS: of course there are different performance implications doing one approach or the other but ultimately it appears FS2020 is rendering point sprites anyhow, so in the end it might not change much between doing what they do which looks like:
Got it, yeah then I know what you mean.
Where I am coming from is the halo and glow effect is caused mainly by our eyes (especially in the zero aerosol setting used by live weather), or by the camera if it is a photo, and therefore the size of that halo and glow is independent of the intensity of the light and is determined by the point spread function of the imaging system. Thus it would seem more natural to reduce the intensity to simulate light propagating a longer distance, and then let the processing steps that simulate the eye fade the halo and glow effects (and colors) accordingly. With a final clipping step for parts of the halo and glow that are now too weak to be seen the result is a perception of a smaller halo.
But I get that in actually rendering that in computer graphics there might be more clever and better looking ways to achieve the same result than the simple physics approach I outlined. If you say shrinking a point sprite is the more clever way then I am sure that is true.
In terms of end result, I certainly think the right picture is more believable.
At least we’re now 2 ! 
I honestly don’t know whether I’m right or not, just that using the halo texture for a bulb light and reducing its size proves to be better in practice in a specific simulator, and like you’re saying it is more believable because the halo and the size are tricking our senses.
PS: and I shall add I find XP11 in VR with this lighting system, in addition to real car lights moving on roads, very immersive.
PS: I’m not bashing FS2020 devs either. I find their graphics engine really amazing given the material complexity (mesh density, textures, PBR) and the light model. What we’re inferring from comparing screenshots is only based on what we see mostly and for some of us also intimate knowledge about how 3D engines are working and coded. It is possible the issues I’ve hilighted in my few latest posts here are only due to a few bugs in the shader code for example and the implementation is supposed to show these lights differently.
Add to that, in XP11 you can change some light bulb properties in a file, and using dataref editor…
If MSFS have that also, then some in the community can just give us their “tweak” on the light file(s) for people to copy / download.
