Night lighting issues still present - The community solutions

In doing it the way you’re doing (bicubic or similar) you’re further distributing and enhancing the sampling errors due to FS2020 rendering a larger point sprite into a smaller size.

@Grinde81 is showing you the correct way to resizing the images, using nearest neighbor so that you can really enlarge the drawn pixels alone.


Here is the screenshot I’ve taken from the paint program with the guides put as close as possible and lights aligned as close as possible for comparison, and it shows the same: size didn’t change:

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.


Here is what is happening in the screenshot above which looks like the left light is smaller but in fact it is not:

bulb_variation


See the same image when I enforce white clipping in changing the level input:

Lights are the same size

1 Like

The shot you’re referring to is a reduction from original size, and you’ll get compression artifacts in the forum upload as well, this is why I included the enlargement with pixel size ratio maintained, for a fair comparison.

This to remove the guesswork and blur-inducing resampling of a downscaled and compressed forum image.

I’ve posted the enlargement multiple times now, uncompressed & unchanged.
But you’re saying I’m lacking basic photoshop skills and have no idea how to resize images side by side…?

I think the basic point of this thread, and one everyone should be able to agree on, is that the lighting looks worse not better. They’ve made the draw distance too great or changed the atmospheric effects somehow so there is not enough falloff in the atmosphere to create a decent looking atmosphere. This world simulator game is surely getting more gamier and less simulatier with each new update.

2 Likes

In my opinion there is no need to answer provocation by another provocation. We’ve been at least two comparing in the most factual way we could, lights didn’t change size (or only marginally) and I believe there is no need to prove more about this here.

However, I believe the only persons we have to convince are Asobo devs and find a way for them listening we’re not satisfied with the current lighting for the many reasons we’re explaining and commenting on. And this is not about marginal light size changes (if any) but more generally about how the different techniques are used and implemented.

1 Like

Currently MSFS is lacking in terms of meteorological visibility. Having this would definitely fade the lamps with distance. Regarding the lamps, pre-update5 lamps was visible just a fraction of what you can see in the real world, so bringing back lamp more realistic distances was definitely the right move.

But doing so without also correcting the less than ideal “bulb” texture gave us a badly tuned appearance, no one is disputing this. It also made other shortcomings very obvious, many of which was earlier hidden by the extremely short lamp visibility range of earlier versions.

Pre update5 was beautifully polished over an extended period of time and had a natural appearance, but was based on a simplified way of emulating night. Update 5 is almost like a rework, making a more realistic foundation (in terms of lamp visibility range, and ground not glowing in the dark) but in current unpolished state I totally get that many misses the pre-patch version, regardless of what made it unrealistic, since it “felt” more beautiful, natural & therefore real

I am not sure if all these observations would be more useful if you, and others from this thread, would sent them to zendesk with screenshots attached

1 Like

I’ve zendesked pretty much all of my findings repeatedly with screenshots, but unfortunately many of those where sent before release-build (and still present), so those screenshots cannot be shared here due to NDA, but I can try to reproduce new ones with current build and in some instances compare to pre update5 in the cases I have screenshots. Anything in particular you would like to see (since I touched quite a few aspects in that post)?

I’m convinced that no matter what the update brings. There will be people on both sides who will argue that it is better or worse. EVERY SINGLE TIME.

There’s is absolutely no possible way Asobo can please them all. It’s as simple as that.

Having said that, the lighting has always looked fine to me.

1 Like

Exactly my point, except I’m not seeing it as two sides “new vs old version” , I’m more seeing it “real life vs any version”

So instead of getting a median value of all opinions (pretty much every user will have one), a simulator should try to match real world.
Many non-pilots gamers & simmers where perfectly fine with lamps disappearing after just a fraction of the distance lamps are visible in the real world, and had no problem with daytime textures glowing in the dark in huge areas that would IRL be completely dark and featureless.

A simulator would benefit the most with setting the bar as close to real life as possible, and not at where a majority (gamers / non pilots) think looks pretty. Both pre & post patch versions had things needing to get fixed, so I’m hoping for real world properties as a foundation and some artistic tuning to make it as beautiful as the eyes actually sees it.

3 Likes

They are not. You are taking a print zoomed in. If you see my post, I said that the new lamps don’t get bigger as we zoom in (the original lamps did get bigger). This was a trick to make lamps bigger on the landscape and we could zoom in without them to balloon in.

The difference is when we are flying zoomed out. And he posted three landscapes and I showed that the first lamps were smaller, and they are. I’m not sure why all the fuss.

Prints below are about the same altitude, same city. The first was before the lighting update, the second, after the update. The size difference is clear.

MFS_22

This is still one of the biggest issues for me:

Hopefully the population density Seb mentioned in the Q&A will help.

7 Likes

I believe this explains why you’re not agreeing. If I’m understanding correctly, you taking screenshots “about” the same position but you’re comparing sets of lights from the screenshots.

If you’re looking at the RJTT example I’m giving, I’m taking the exact same light bulbs, from the near exact same position and angle therefore near exact same distance. I’ve just zoomed the view in the paint program so that the guides are visible, but this is a nearest neighbor zoom so that for example 1 source pixel at 100% displays as 16 times the same pixel color (4x4 pixels) at 400% zoom, therefore eliminating any resampling artifact.

Se again: the guides are clearly showing the same bulbs in the same location in the scenery from the same camera position are the same number of pixels:

Lights are the same size

At this stage you might want to either try finding the same before/after camera shot (same conditions position angles zoom) and do measure graphically the number of pixels, otherwise it is not necessary to further trying convincing me I’m wrong, let’s move forward if you don’t mind.

That’s not the original print, that’s the forum resized and heavily compressed, and then very blurry resized version.

I finally understand your confusion now since you seen to be basing it all on messing around with downsized and heavily compressed versions instead of looking at the enlargements (occupying more pixels to prevent forum compression artefacts).
I made those enlargements exactly for this reason, using original data before forum resizing/compression to prevent confusion and incorrect comparisons exactly like the one above using compressed and downsized forum versions.

The pixels are right there.
No way around that, the pixel count is right there for anyone to see.
6x7 pixels, 7x6 pixels, 6x6 pixels. That’s the size of the bulb texture both at release, after update 5 & 7.
(+/- one or two pixels depending on aliasing effect)

Since you don’t seem to believe me, I’ve included original non-resized/compressed version attached below, so you and anyone else can see what I’ve been telling you the whole time.

Here’s the unchanged original sized tiff to avoid forum compression and degradation.

So you and anyone else can see those enlargement are a completely correct representation of the original size lamps.

The lamp bulb texture is the same in both release and post patch versions. The visibility/brightness range is different. And shows now more than ever that it’s too big, and desperately needs a smaller texture with much better scaling for size & brightness.

1 Like

agree 100% . Rayleigh scattering probably. All of the discussion on lamp size aside, the night scene i remember early sept was stunning, i did a good few approaches and flew around glasgow being amazed. It now looks harsher and has less sublty between side streets and main roads.

What concerns me is these changes never get documented in the release notes. So, it burns 100’s of hours of peoples time objection handling them into either admitting it changed, or, as an example the LOD, people finding the settings in the cfg file and putting back was was there before. All the smaller bugs still also dont get fixed (high alt airport temp/pressure etc), just stick it backlog and see how many posts we get voted about it. This isn’t a great way to fix, develop or tune any software product. The huge gap between the # of fixes per month vs the revenue taken in is certainly worth thinking about.

1 Like

Thank you. You’ve got me covered in all of your posts!

There is no confusion. And why your enlargements are so low res? I can count just a bunch of pixels, very differently from the cutout that I posted.

So this talk of ‘downsized and heavily compressed versions’ is more applicable to you.

I’m not sure why are you trying to skew the argument.

Funny thing is, only reducing brightness in Freestyle won’t make the orbs smaller… I got that effect by using a special FX filter called “sketch” a tad bit. It sort of grey-washes the entire image over, but it has an effect of “chipping away” the edges of the orbs to some extent.

And of course since your scene is now grey-washed, then you’d need to play with shadows, gamma, brightness, colour temperature etc so it looks darker instead of grey.

A very limited band aid temporary solution for me in the time being… But I prefer the original (non-filtered), with smaller orbs that don’t get fuzzy/blurry in the distance. A tad fuzzy can be understandable maybe due to atmospheric effect etc… But not to the extent of myopic blurry.

1 Like

It is a temporary solution. After some experimentation it feels fake. For now back to the original.

This will probably be my last post about light bulbs sizes, but I’ve experimented trying to make them more visible and this is really interesting comparing the different approaches vs the photo. Here is:

Real:

XP11:

FS2020 Update 3:

FS2020 Update 7:

NB: you have to open these images in their full size to really see the details!

A few quick comments on what I see:

  • fs2020 update 7 is rendering almost all “street light” bulbs at the same size, whatever the distance to the viewer. In addition the emboss filter is showing sharp edges which means they are very bright with very small halo. There seem to be “types” of lamps each displayed the same size and same way (all taxiway bulbs are rendered the same, all street bulbs the same, etc…)

  • fs2020 update 3 is rendering smaller street lights in the distance which is good, and the emboss filter is showing more fussiness revealing a wider halo. However it looks like their size vs distance is not linearly distributed like in the xp11 shot. Instead it seems they are displaying by LOD rings (all lamps in LOD ring 1 are the same size, all lamps in LOD ring 2 are same size). Otherwise if not by LOD rings there is some form of ‘clustering’ of same size lamps.

  • xp11 is rendering street lights with size varying with the distance and all lights are softer than fs2020 indicated the rendering of the halo around the light even for the closest ones.

  • real photo is showing more variation in light sizes but it shows there is a correlation still with light size and distance to the viewer. Furthermore, some of the lights are as sharp as the ones in fs2020 but most are fuzzy like xp11

2 Likes

That right there is part of the confusion.

You’ve taken a picture that first has been made smaller & degraded with high compression, and then made a blurry enlargement out of that.
That introduces a bunch artefacts and destructive defects blurring the whole thing, not longer an exact representation of the original size difference being compared.

The lamp size compared in the originals at native resolution where 6x7, 7x6 & 6x6 pixels / size in the versions compared.
Or do you refute that?

It becomes more obvious looking at the enlargement. (Since and the forums quality compression wrecks the original as your enlargements clearly shows.)

The enlargements I made are showing/representing the exact same numbers of pixels as the original native scale being compared, but “bigger” to make the comparison easier without blurring/messing up due to forum compression, as you now can see exactly how many pixels (if any) the size differs at native MSFS resolution (after all, that’s what we’re getting on the screen when flying.

That’s what the comparison is about. Showing the lamps (within range before falloff starts) at the same location, same distance, and same conditions actually have the same size.

But the distance where the lamps start to fade is much further away now, due to the increased lamp draw distance.
And because of this the size of the lamp texture definitely needs shrinking and scaling needs fixing too.