Can we NOT have stars during daylight?

[Delay]

9 minutes ago, PDCWolf said:

The excuse offered is always the vulgarity of the vast majority of the public. I insist that this is absolutely false.

You can quote all you want, dismissing it as an excuse is not an argument.
Fact is that the general public has an idea of what space travel looks like. KSP already broke many of these expectations, leading to severe frustration in players who don't necessarily know a lot about space travel. I can imagine this ended some people's interest in the game as being "too hard" (I know we don't have these problems, but that does not imply the complement is empty!).

The goal should be to strike a balance between meeting expectations and teaching real-world physics. And again: I'm supporting the middle-ground options of either allowing full control of every visual aspect or have mods do this in case people like to.

I don't want to repeat myself too much, but I play with DOE by choice, because I want realism over artistic license. But said artistic license is absolutely fine to have and I would accept, play and love the game just as much without it.

Just because I like bloom and slight overexposure doesn't mean you do either. Doesn't mean I should get to decide how your planets look. Or your skybox. Or your rockets. Or your anything.

Edited November 23, 2021 by Delay

[kedrednael]

11 minutes ago, PDCWolf said:

Quote

If sunlit objects/ the sun keep the same exposure, while only the skybox dims, then that is really distracting and useless. The skybox should dim to prevent the actual sunlit objects/ the sun from being overexposed.

This is a programmatically implemented limitation, and I believe it is easier to not implement it than to bother doing what KSP seems to do

It is definitely easier to just fade the skybox texture in and out like KSP DOE does and keep all other objects the same. I could code that myself.

It is really hard to give all objects an accurate luminosity and adjust exposure for everything, making some objects over or completely under exposed. I have no idea where to even start if I had to make that. Space Engine is the Only program that I know that does that correctly.

Edited November 23, 2021 by kedrednael

[PDCWolf]

3 minutes ago, Delay said:

You can quote all you want, dismissing it as an excuse is not an argument.
Fact is that the general public has an idea of what space travel looks like. KSP already broke many of these expectations, leading to severe frustration in players who don't necessarily know a lot about space travel. I can imagine this ended some people's interest in the game as being "too hard" (I know we don't have these problems, but that does not imply the complement is empty!).

Fact is that the general public loved KSP to the point the devs decided to evolve the product and support it for 10 years, selling it to one of the biggest gaming companies.  It's also one top rated games on Steam, and has been for a while, which is reflected as well on it's metacritic score and global critical acclaim from almost all gaming sources and networks. Let's not diminish KSP's accolades just to have a point, not to mention again how this is not a gameplay mechanic, but rather a visual effect, which is also present on almost all modern games nowadays, even across genres, in the form of eye adaptation.

Quote

The goal should be to strike a balance between meeting expectations and teaching real-world physics. And again: I'm supporting the middle-ground options of either allowing full control of every visual aspect or have mods do this in case people like to.

I don't want to repeat myself too much, but I play with DOE by choice, because I want realism over artistic license. But said artistic license is absolutely fine to have and I would accept, play and love the game just as much without it.

Just because I like bloom and slight overexposure doesn't mean you do either. Doesn't mean I should get to decide how your planets look. Or your skybox. Or your rockets. Or your anything.

I don't mind it being an option and I've said this before, but yet again, we don't know if they've arrived to where they are right now following artistic vision or being ignorant of the principle, heck, it's been 10 years and we still don't know this of the original KSP either.

What you like is completely up to you, and I don't disagree with your tastes, I disagree with some of the arguments you're using.

7 minutes ago, kedrednael said:

It is definitely easier to just fade the skybox texture in and out like KSP DOE does. It is really hard to give all objects an accurate luminosity and adjust exposure correctly, making some objects over or completely under exposed.

Once the lighting issue is settled, what'd be hard is adjusting exposure, indeed, yet almost every modern game having some form of "eye adaptation" kinda tip the scales against how hard it could possibly be.

[kedrednael]

10 minutes ago, Delay said:

The goal should be to strike a balance between meeting expectations and teaching real-world physics.

This is the difficult balance the devs have to find.
Fighting with light exposure controls like in space engine would be stupid when you're also having difficulty actually controlling a spaceship.
 

1 minute ago, PDCWolf said:

yet almost every modern game having some form of "eye adaptation" kinda tip the scales against how hard it could possibly be

No they don't I think? Do you have an example?
Space engineers and Elite dangerous try and fail. Those are pretty big titles. Space engineers seriously adjusts to become darker when you enter dark areas, so, wrong way around. Elite dangerous is pretty static, has stupid things like stars actually becoming brighter when entering a thin atmosphere.
Universe Sandbox, 'Children Of A Death Earth',  Battlefront II, star trek online, X4-Foundation, and 'In the black' (from impeller studios) don't even try, they just have static exposure.

Games on planets can cheat the lighting easier. It's harder in space because then you can actually have the dim and bright objects in view at the same time.
Only the most expensive games of the last few years seem to do it well: Star Citizen and Microsoft FlightSim 2020 (but that's on planet so maybe cheating). And Space Engine.

[PDCWolf]

26 minutes ago, kedrednael said:

No they don't I think? Do you have an example?

Games on planets can cheat the lighting easier. It's harder in space because then you can actually have the dim and bright objects in view at the same time.

If you limit yourself to the space genre, then yeah, but also most games in the space genre definitely do not qualify as modern. Elite came out 8 years ago, with Horizons & Odyssey barely making the graphics better. Space Engine is the one that definitely tries the hardest save for some very big oversights, but at least gives you the option to fix it, even then Space Engine originally came out almost 12 years ago, though it is continually being evolved.

 

 

[sturmhauke]

On 11/22/2021 at 3:44 PM, PDCWolf said:

I meant that it seemed fantastic and laughable to me, that someone would be quickly able to recognize a couple stars on a place that is not Earth, from a single, almost reference-less picture.

...You're looking at a reference-less sky from a place that is not even on your same world, no human has ever seen that sky directly, nor is a single human accustomed to search for stars there, nor do we know the FOV of the camera to justify how much sky is actually visible, or any other useful data... what I just described makes it fit the idea that the picture shows an enough different sky (from what we see every day) that trying to play stargazer is, again, laughable and fantastic.

So you do contend that the night sky from Saturn orbit looks substantially different than the night sky from Earth orbit. (Let's ignore the sky from here on the Earth's surface, since there are significant atmospheric effects.)

The problem is that except for objects in our own Solar System, that's just not true. The average distance from Earth to Saturn is approximately 1.28 x 10⁹ km (8.5 AU). The distance from Earth to Proxima Centauri is 4.02 x 10¹³ km (4.25 ly), which is about 31,400 times further away.

For comparison, say you're looking at a mountain peak 50 km away. What you're saying with the view from Saturn is the equivalent to saying that the mountain peak and surrounding terrain will look different if you step 1.5 m left or right.

Edited November 24, 2021 by sturmhauke

[PDCWolf]

1 minute ago, sturmhauke said:

So you do contend that the night sky from Saturn orbit looks substantially different than the night sky from Earth orbit. (Let's ignore the sky from here on the Earth's surface, since there are significant atmospheric effects.)

The problem is that except for objects in our own Solar System, that's just not true. The average distance from Earth to Saturn is approximately 1.28 x 10⁹ km (8.5 AU). The distance from Earth to Proxima Centauri is 4.02 x 10¹³ km (4.25 ly), which is about 31,400 times further away.

Point taken, yet we still we ended up needing the help of an orrery software and overblown exposure to correctly identify stars. Whilst you're right that there's negligible change in the shape of the constellations, if I handed you an image with 17 random dots, taken from an unknown position, in an unknown point in time, with unknown field of view data, I'd really find it fantastic and laughable again that you'd come pointing out stars and constellations with such certainty as it was done. The picture, with all that missing data taken into account, and with what very little data it presents., represents an alien sky, both in dictionary definition and concept.

[Delay]

4 hours ago, PDCWolf said:

if I handed you an image with 17 random dots, taken from an unknown position, in an unknown point in time, with unknown field of view data, I'd really find it fantastic and laughable again

The date of the image Cassini took is known. From that you can calculate where Saturn was in its orbit. Based on the position of Saturn's rings relative to the planet you can estimate where Cassini was. Based on that you can reconstruct the sky at the time.

6 hours ago, PDCWolf said:

not to mention again how this is not a gameplay mechanic, but rather a visual effect,

Exactly! It does not affect gameplay in any way, shape or form. So why do you, or anyone by extension, care so much that the developers should decide for everyone?
Either make it a highly customizable option or leave it to the modders. There is no correct side here. You can make a case for realism, you can make a case for artistic license, or both or none.

[PDCWolf]

18 minutes ago, Delay said:

The date of the image Cassini took is known. From that you can calculate where Saturn was in its orbit. Based on the position of Saturn's rings relative to the planet you can estimate where Cassini was. Based on that you can reconstruct the sky at the time.

That goes way beyond simple stargazing and recognition of constellations, and is exactly what it took to reconstruct the position of Cassini in Celestia.

Quote

Exactly! It does not affect gameplay in any way, shape or form. So why do you, or anyone by extension, care so much that the developers should decide for everyone?
Either make it a highly customizable option or leave it to the modders. There is no correct side here. You can make a case for realism, you can make a case for artistic license, or both or none.

If this was an actual argument, and people didn't care about products evolving to include new features, anything but a flat procedural surface and christmas rockets would be a mod and we'd still be at 0.1.

[Delay]

5 minutes ago, PDCWolf said:

If this was an actual argument, and people didn't care about products evolving to include new features, anything but a flat procedural surface and christmas rockets would be a mod and we'd still be at 0.1.

So visuals being left up to the community is the same as implementing new rocket parts, planets and gameplay features? You yourself distinguished between VFX and gameplay before, why the sudden equivalence?

KSP 2 is in the unique position of being the successor to a game that already did leave these decisions up to the individual to decide and customize. Following this idea is not bad.

Edited November 24, 2021 by Delay

[PDCWolf]

9 hours ago, Delay said:

So visuals being left up to the community is the same as implementing new rocket parts, planets and gameplay features? You yourself distinguished between VFX and gameplay before, why the sudden equivalence?

KSP 2 is in the unique position of being the successor to a game that already did leave these decisions up to the individual to decide and customize. Following this idea is not bad.

Except they've now learned from what the community did, and we have stock implementations of effects that mods had to bring to the table: RealPlume, Scatterer, Clouds, PlanetShine, Engine Light, etc. 

[Delay]

1 hour ago, PDCWolf said:

Except they've now learned from what the community did, and we have stock implementations of effects that mods had to bring to the table: RealPlume, Scatterer, Clouds, PlanetShine, Engine Light, etc. 

How did you just gloss over my question?

Plus, you make it sound like I'm against improving these systems purely because they are visual. I welcome every improvement in both gameplay and visuals, just to make that clear. However, unlike replacing a particle system with a mesh-based one (which, among other things, has performance benefits), removing stars is not an improvement, just a simple change that doesn't add or subtract anything.

Edited November 24, 2021 by Delay

[PDCWolf]

5 minutes ago, Delay said:

How did you just gloss over my question?

Plus, you make it sound like I'm against improving these systems purely because they are visual. I welcome every improvement in both gameplay and visuals, just to make that clear. However, unlike replacing a particle system with a mesh-based one (which, among other things, has performance benefits), removing stars is not an improvement, just a simple change that doesn't add or subtract anything.

I differentiated graphics from mechanics when the "bad game design" was brought up, as the video used for exemplifying talked about mechanical bad game design. 

You're now saying that increasing the realism of the visual systems and bringing them up to gaming industry standard doesn't add or substract anything. It adds realism, substracts unintuitive behavior (remember how the stars not being in the Apollo pictures were a big part of the conspiracy theory?). On top of that, you still keep referring to it as "removing stars", when that's clearly not how it works, and you definitely know that, having played with DOE. Stars will be there, you'll see them come and go exactly as you do in real life when lighting conditions change.

[Delay]

4 minutes ago, PDCWolf said:

It adds realism, substracts unintuitive behavior (remember how the stars not being in the Apollo pictures were a big part of the conspiracy theory?

You got it the wrong way around. Stars in KSP aren't unintuitive because they don't show up in the Apollo pictures. They don't show up in the Apollo pictures because that is unintuitive, because that is something we are not used to seeing.

I'm more or less actively involved in debunking that conspiracy theory, so I'd know what, how (and if) they think.

6 minutes ago, PDCWolf said:

On top of that, you still keep referring to it as "removing stars", when that's clearly not how it works

Fine, "Replacing the star-filled skybox with a black one in the presence of any bright object". I can understand how under non-superficial interpretation, this statement can seem ambiguous.

 

7 minutes ago, PDCWolf said:

Stars will be there, you'll see them come and go exactly as you do in real life when lighting conditions change.

Most of the time stars would not be visible, unless you are located in the shadow of a planet or moon. Even in interplanetary space, the spacecraft would reflect too much light to make stars visible. Thus, we're not really talking about "stars unless object" and more about "no stars unless shadow".

In that case, why bother with a skybox at all? It requires long exposure times, even in LEO.

[PDCWolf]

12 minutes ago, Delay said:

You got it the wrong way around. Stars in KSP aren't unintuitive because they don't show up in the Apollo pictures. They don't show up in the Apollo pictures because that is unintuitive, because that is something we are not used to seeing.

Except you try to look at the sky and turn on a light and suddenly there's less stars.

Quote

I'm more or less actively involved in debunking that conspiracy theory, so I'd know what, how (and if) they think.

Fine, "Replacing the star-filled skybox with a black one in the presence of any bright object". I can understand how under non-superficial interpretation, this statement can seem ambiguous.

Insisting on trying to be snarky, I see.

Quote

Most of the time stars would not be visible, unless you are located in the shadow of a planet or moon. Even in interplanetary space, the spacecraft would reflect too much light to make stars visible. Thus, we're not really talking about "stars unless object" and more about "no stars unless shadow".

In that case, why bother with a skybox at all? It requires long exposure times, even in LEO.

It's been like 15 posts and you're still trying to misrepresent the idea to have an argument at all. You'd have stars during night time, during the dark side when orbiting a planet (if you're not looking at anything else that's bright enough), when an eclipse happens, when your own craft or another passes in front of the sun, when in interstellar space, when looking from inside a capsule in the shade, etc. Plus I never locked the method to dirty skybox tricks, and they could do it in other ways, not sure why you keep bringing up skybox tricks when that's just one possible method.

[Delay]

23 minutes ago, PDCWolf said:

Insisting on trying to be snarky, I see.

Actually, no. I was genuinely trying to correct this ambiguity. Don't worry, it's not the first time I've been misinterpreted / have presented my texts as misinterpretable like this.

 

25 minutes ago, PDCWolf said:

Plus I never locked the method to dirty skybox tricks, and they could do it in other ways, not sure why you keep bringing up skybox tricks when that's just one possible method.

I mean skybox as a generalization of all possible techniques, as skyboxes are the most ubiquitous way to achieve such an effect.
No perceived snark this time, hopefully. I can see how that might not be apparent. I'm not insisting on this method, I've simply chosen it as a representative of all.

 

29 minutes ago, PDCWolf said:

You'd have stars during night time, during the dark side when orbiting a planet (if you're not looking at anything else that's bright enough

That is literally everything. I'd have to look exactly up in the middle of Kerbin's shadow to even try to see a few stars.

Now, you could fix that by increasing the simulated exposure time, but that would necessitate the implementation of at least motion blur, and not just over 1-2 frames.

[PDCWolf]

15 minutes ago, Delay said:

That is literally everything. I'd have to look exactly up in the middle of Kerbin's shadow to even try to see a few stars.

Now, you could fix that by increasing the simulated exposure time, but that would necessitate the implementation of at least motion blur, and not just over 1-2 frames.

Hey, where's the rest of my list lmao. If you look away from the possible halo, or at least shielded your eyes from it, you'd see stars too. That's how astronauts see a excrementston of stars at night on the ISS, and also auroras. 

[Delay]

37 minutes ago, PDCWolf said:

Hey, where's the rest of my list lmao.

Rather easily refuted. A spacecraft is too small to completely cover the sun most of the time. Just how close do you want the camera to be?

Interstellar space is an obvious one. You're not arguing that, either. Even the title of the thread excludes interstellar space from being considered.

As for eclipses, I find it difficult to find any photographs of that, neither from the ISS nor the ground. I know that it's possible to observe stars behind the sun during a solar eclipse (one of the first demonstrations of general relativity), but no video or photo seems to exist that demonstrates the phenomenon. I could just not look deep enough, though. Would you like to share a photo?

...unless you were talking about lunar eclipses, which happen at night (see title) and it wouldn't be possible to observe a lot of stars with at most a 1/60s exposure time (assuming no motion blur is desired. Though star trails do look pretty...). The sky from the ISS is quite bright.

Edited November 24, 2021 by Delay

[PDCWolf]

2 hours ago, Delay said:

Rather easily refuted. A spacecraft is too small to completely cover the sun most of the time. Just how close do you want the camera to be?

Literally a matter of perspective. It's pretty common to completely eclipse the sun with your craft in KSP.

Quote

Interstellar space is an obvious one. You're not arguing that, either. Even the title of the thread excludes interstellar space from being considered.

That's on me for not remembering KSP2 was gonna include interstellar travel. Multiple star systems were confirmed long ago.

Quote

As for eclipses, I find it difficult to find any photographs of that, neither from the ISS nor the ground. I know that it's possible to observe stars behind the sun during a solar eclipse (one of the first demonstrations of general relativity), but no video or photo seems to exist that demonstrates the phenomenon. I could just not look deep enough, though. Would you like to share a photo?

...unless you were talking about lunar eclipses, which happen at night (see title) and it wouldn't be possible to observe a lot of stars with at most a 1/60s exposure time (assuming no motion blur is desired. Though star trails do look pretty...). The sky from the ISS is quite bright.

I've personaly witnessed a total solar eclipse, stars are visible. Obviously nothing like watching the sky at night, but if you happen to have a good spot and the stars are (pun intended) on your side, you might see the brightest ones. In preparation for that eclipse on July 2 2019, I checked articles like these: https://www.space.com/36721-stars-planets-visible-during-solar-eclipse.html

I can tell you too that if you're purposefully trying to photograph the corona, it's so very dim that you won't see stars, but that's a matter of what you can actually see with your eyes (both corona and stars) vs balancing exposure on a camera. 

[kedrednael]

On 11/24/2021 at 1:59 AM, PDCWolf said:

If you limit yourself to the space genre, then yeah, but also most games in the space genre definitely do not qualify as modern. Elite came out 8 years ago, with Horizons & Odyssey barely making the graphics better. Space Engine is the one that definitely tries the hardest save for some very big oversights, but at least gives you the option to fix it, even then Space Engine originally came out almost 12 years ago, though it is continually being evolved.

 

 

So, no examples of games that do have correct exposure adjustments.

[PDCWolf]

15 hours ago, kedrednael said:

So, no examples of games that do have correct exposure adjustments.

Yeah, not with the genre limit. That's important, as there aren't that many games in the genre to begin with anyways., otherwise the list is as big as we want it to be, since every unity game has an option to natively implement full-scene-luminance-histogram based eye adaptation: https://docs.unity3d.com/560/Documentation/Manual/PostProcessing-EyeAdaptation.html

This is also a feature available to every unreal 4 (and 5) game, and it works exactly on the same histogram method: https://docs.unrealengine.com/4.27/en-US/RenderingAndGraphics/PostProcessEffects/AutomaticExposure/

 

[SOXBLOX]

Interesting that this thread is still going. First, I'll say again that I was wrong that the stars in that image of Saturn were part of Orion. I made a quick guess and assumed that the stars would have to be fairly bright. About five minutes of finding the date of the image and recreating the scene in an orrery proved my guess wrong, which is fine. It doesn't matter. If anything, it shows that a more human/HDR-like camera can capture fainter stars that I thought, even in bright light.

Anyone who thinks the sky in orbit of Saturn is substantially different from that of Earth, however, has no idea what they're talking about. I think we all agree there. So, it was not "fantastic" that I should have made a confident guess. It was completely possible to make an accurate guess based on a knowledge of the night sky. Of course, I didn't, but I haven't memorized all the stars down to the 9th magnitude, either. 

Now, we've also established that people in bright environments (for a certain definition of "bright" and "environment") don't see stars because of A) dark adaptation and B) relative brightnesses. Cameras are a different animal, and have additional issues that can come into play. We all agree there.

The argument was about whether people standing in shadow on the sunlit side of a world could see stars. @PDCWolf's original wording made it seem like he thought all planets were like Earth, where the sunlit atmospheric smog and particles drown out the stars. That would mean stars aren't visible on other worlds if the sun is in the sky, regardless of shadow, and regardless of whether the world has an atmosphere or not. I (correctly) pointed out that, if that was what he meant, then he would be wrong. Emphasis on the "if". Like I said, sloppy wording. You can see stars if you're not looking at a bright light source, regardless of whether the sun is shining on the back of your head or not. The Apollo astronauts (lookie, primary sources!) confirmed this.

I also pointed out that it is still possible to see stars directly next to a bright object, so long as the relative brightness is within a certain threshold. (e.g. one can see bright stars next to the full Moon.)* The OP didn't include this caveat in his post, even though it's necessary. Perhaps he didn't realize it?

----------

So, really this argument started because of sloppy wording. That's fine; details escape us all sometimes, and that's what the community is here for. 

As soon as this becomes a matter of subjective taste, or whether or not other unrelated games do a certain thing, we've left the realm of fact and entered the realm of subjective opinion and art. If making the skybox behave a certain way improves the game, by all means, go for it. But if it confuses players or doesn't add anything worthwhile, maybe it can be left out. Best of all, it can be a toggle-able or slider-ed setting. That way, everyone can be happy. (Except people who want others to only do things their way.)

----------

* Pretty sure the Sun's brightness at the distance of Pluto is less than that of the full Moon from Earth. So, on Pluto you could look directly at the Sun and still see stars. It's brighter than that, at around 250× the full Moon's brightness. This is equivalent to the Sun being just below the horizon on Earth. So, twilight-level illumination, from one point in the sky. Planets would be visible for sure, and any abnormally bright stars.

As OHara pointed out, the further out you go, the more stars you can see, so there is a point when you can see stars even while looking directly at the sun. This is another caveat that needs to be included in any broad, sweeping claim like "stars are not visible when the sun is shining overhead".

Edited November 29, 2021 by SOXBLOX
Fixed.

[OHara]

42 minutes ago, SOXBLOX said:

Pretty sure the Sun's brightness at the distance of Pluto is less than that of the full Moon from Earth. So, on Pluto you could look directly at the Sun and still see stars.

The sun is still 100× brighter when seen from Pluto, than a full mon is from here (5 steps difference in apparent magnitude) according to the numbers at wikipedia (which look right to me.) 

But on a journey between here an alpha-Centauri, the sun will dim until it looks like just another star.   The interstellar aspect of KSP 2 might have prompted the developers to simulate some of the varying brightness of stars and all the other things in view. 

[kedrednael]

8 hours ago, PDCWolf said:

Yeah, not with the genre limit. That's important, as there aren't that many games in the genre to begin with anyways., otherwise the list is as big as we want it to be, since every unity game has an option to natively implement full-scene-luminance-histogram based eye adaptation: https://docs.unity3d.com/560/Documentation/Manual/PostProcessing-EyeAdaptation.html

This is also a feature available to every unreal 4 (and 5) game, and it works exactly on the same histogram method: https://docs.unrealengine.com/4.27/en-US/RenderingAndGraphics/PostProcessEffects/AutomaticExposure/

 

That's really cool info!
KSP2 is being developed in Unity right, so those settings should be editable with mods, if we're not happy with them.


I wonder if it can recreate the actual brightness range needed for completely realistic stars though. What I am about to write is probably wrong in many ways.
If I understand correctly, each frame is calculated in high-dynamic-range HDR mode,  which has the max range of brightness values. This HDR image is calculated in 16 bit, which means each pixel can have brightness values between 1 and 2^16 = 16500. This is transformed to a low-dynamic-range (LDR) image to show on the screen, which basically cuts off the lowest and brightest values (and does some squishing & log transformation). The trick to simulated eye adaptation is determining where it should take this slice of the HDR's brightness values. The LDR needs 255 different brightness values, the amount of brightness values of a computer screen. If it has less than 255 brightness values banding artifacts occurs, there are jumps between the different brightness values.
So what's shown on screen can be a slice of the calculated brightness values: 255/16500 = 1.54% the brightness values of the HDR. (This is optimistic because of the log transform: more detail is needed for the darker values of the LDR than the brighter values -> For the LDR you actually need a bigger slice of the HDR values or artifacts will occur.)

Our eyes can see 24 stops brightness difference at any one time, that means the brightest thing you can differentiate is 2^24 = 16 million times as bright as the darkest value you can differentiate.

We want to be able to see sunspots and other cool details when doing a flyby over the sun (as opposed to the sun always being overexposed, even when it fills the screen), and we want to be able to see the milkyway/ the dimmest visible star. I read the dimmest star we can see with our eyes is 6 trillion times dimmer than the sun. So we need to calculate brightness values for stuff 6 trillion times brighter than the darkest thing. Then select a slice of 16 million big out of those 6 trillion values, because that's the range we can see at any one time. That slice is only 0.00027% of the calculated values. This is a way smaller portion than the optimistic 1.54% slice we could select without artifacts out of our 16bit HDR image.

My Conclusion:
We cannot have completely realistic lighting range in game, a 16 bit HDR image does not offer enough detail. I don't think they can switch to higher bit HDR calculation, that may not be supported, would be slower and require more memory.

Like the unity manual explains: "in any given moment of time, the [human] eye can only sense a contrast ratio of roughly one millionth of the total range. What enables the wider reach is that the eye adapts its definition of what is black."   My calculations implied we want the range at any given time to be one in six million of the total range (not far off!  (and I wanted to enable sunglasses for when close to the sun basically, extending the range further than our eyes can)). They do not explain what the maximum range of their HDR/ simulated eye adaptation is.

Edited November 29, 2021 by kedrednael
eyesight 24 stops

[PDCWolf]

Quote

I wonder if it can recreate the actual brightness range needed for completely realistic stars though.

This I doubt we could do realistically, but there are many ways to at least solve the problem programmatically by having them on different layers or something like that, without doing something crazy like actually having 10 to 20 thousand point light sources at different brightness levels.

Quote

What I am about to write is probably wrong in many ways.

me when

Quote

If I understand correctly, each frame is calculated in high-dynamic-range HDR mode,  which has the max range of brightness values. This HDR image is calculated in 16 bit, which means each pixel can have brightness values between 1 and 2^16 = 16500. This is transformed to a low-dynamic-range (LDR) image to show on the screen, which basically cuts off the lowest and brightest values (and does some squishing & log transformation). The trick to simulated eye adaptation is determining where it should take this slice of the HDR's brightness values. The LDR needs 255 different brightness values, the amount of brightness values of a computer screen. If it has less than 255 brightness values banding artifacts occurs, there are jumps between the different brightness values.
So what's shown on screen can be a slice of the calculated brightness values: 255/16500 = 1.54% the brightness values of the HDR. (This is optimistic because of the log transform: more detail is needed for the darker values of the LDR than the brighter values -> For the LDR you actually need a bigger slice of the HDR values or artifacts will occur.)

It's off but not that far off: https://knarkowicz.wordpress.com/2016/01/09/automatic-exposure/

Quote

Our eyes can see 24 stops brightness difference at any one time, that means the brightest thing you can differentiate is 2^24 = 16 million times as bright as the darkest value you can differentiate.We want to be able to see sunspots and other cool details when doing a flyby over the sun (as opposed to the sun always being overexposed, even when it fills the screen), and we want to be able to see the milkyway/ the dimmest visible star.

We can't normally see black spots, not with our eyes alone, unless you have a huge chunk of atmosphere magnifying and also attenuating the sun (i.e. only during sunrise or sunset, and yes, this damages the eye anyways). In any other condition, the brightness just overwhelms the spots. We can, however, see all the way down to zodiacal light when there's no pollution, that bit is correct. Finally, once outside the atmosphere, astronauts are able to see even more, since atmospheric extinction is no longer bringing everything down. 

Quote

My Conclusion:
We cannot have completely realistic lighting range in game, a 16 bit HDR image does not offer enough detail. I don't think they can switch to higher bit HDR calculation, that may not be supported, would be slower and require more memory.

Of course not, but as space engine shows, we can very well reproduce the effect in a credible way. Still, using a skybox instead of actual stars would require different trickery to reproduce the effect at such levels of detail.

@SOXBLOX I see you've edited your comment. First off, YEAH, AMAZING, this is like the 5th or 6th most viewed thread in this subforum, and the most commented thread. Second off, any accusation of sloppy wording, or straight up incorrect expressions I'll take, but only because I'm not a native English speaker. Finally, yeah, I think the objective side of the thing is all cleared up as one follows the discussion, and what we're left with are technicalities and subjectivities. However, if I were to base myself on other suggestions, none of them have the super specific claim that would clear any discussion up as the center of the opening post.

As for the alien sky thing, it's a game we could play: It's a fact that the sky is not different, but my point is reliant on references to find orientation, not constellation wandering via parallax. We could give each other small sectors of the sky, without reference, orientation data, or context, to actually see how much we can identify, if we can identify anything at all. At this point I'll still die on the hill that neither of us, or anyone in this thread, would correctly identify constellations in those circumstances.

Edited November 29, 2021 by PDCWolf