1.5.0 prerelease 2

[chris]

Dos anyone at all like anything about the new star rendering? Or should I just revert it completely?

--Chris

[ElChristou]

Dos anyone at all like anything about the new star rendering? Or should I just revert it completely?

--Chris

I'm not the indicated to really test your rendering (no OGL2 yet) but seems to me it's a question of tuning than anything else now...
During the galaxy work, we spend days/weeks/months of work with also quite some critics but the present result is here, not so bad... (quite good indeed in the specific frame we draw)

After all, those problems of atmosphere/glare will be part of the past once you will implement HDR, so perhaps in the meantime and to not lose too much time, why not pointing us to the specific values in the code to play with and let's try all together to find a quick fix to those minor problem?

[selden]

Personally, I like the glare effect for stars if it could be made continuous -- by that I mean that when it's on, it's on for all brightnesses and sizes of stars, not turned off in the middle ranges making the stars' surface features visible.

I don't think the star surface features need to be visible when the glare effect is turned on -- when "scaled discs" is selected. However, surfaces should be visible when "fuzzy stars" or "point stars" are selected, as in previous versions of Celestia.

The current albedo effect, however, just isn't working for me. I think it needs to be like the star glare effect I've described: when it's on, it's needs to be on for all sizes of reflecting regions (not the entire body as for emissive objects like stars), but when it's off, objects should just fade away.

[t00fri]

On the one hand I certainly appreciate Chris' effort to try and improve the star rendering! I also appreciate Chris' high effectiveness in coding, of course.

On the other hand, he based the results on some parametric modelling of the stars that has never been really revealed viz exposed to the critics of people with expertise in optics.

What I know is that the new stars invoke some Gaussian (luminosity) ansatz with a certain clipping prescription superimposed, in case of saturation.

We never had a chance of discussing this ansatz from a PHYSICS point of view, notably as concerns various important limiting situations. It was never clear whether the ansatz makes sense, physicswise. Suddenly, there was the new code without much explanation of what went into it...

++++++++++++++++++++++++
All that was available to the public was a "yes-no option": do you like the new stars or not?
++++++++++++++++++++++++

Sorry, but this is NOT my style of discussing such non-trivial tasks. So I typically tend to stay out of such achievements...

Bye Fridger

[rthorvald]

About the Sun, With this version, the Runar Thorvaldsen's texture does not work well

I??ll look into that when 1.5 is finalized.

- rthorvald

[ElChristou]

more opinion:

- I intentionally removed the atmospheres, because I think that stars look better without. It also makes eclipses look better.

I don't think it looks better without atmosphere, even if the rendering with atmosphere is not ideal, without IMO it's less credible, now for sure eclipses looks much better but it's a hard dilemma in the actual state of the rendering...

- The glare texture is being overlayed on the star, washing out some of the detail in the star texture. Personally, I think that this is an improvement.

I agree because in all case, details of the surface should not be visible in normal light.

[chris]

After all, those problems of atmosphere/glare will be part of the past once you will implement HDR, so perhaps in the meantime and to not lose too much time, why not pointing us to the specific values in the code to play with and let's try all together to find a quick fix to those minor problem?

HDR rendering is not a panacea. It will produce very realistic images on systems with modern graphics hardware. But, the images may be *too* realistic . . . For example, for a simple spacecraft visualization, you don't necessarily want the sun to fill the screen with glare when it comes into view. Also, I guarantee that some of the things will see in an accurate HDR view will appear strange even though they're realistic. Since few of us have ever spent any time in space, our expectations of what it looks like are based largely on movies and space art. A few people here will remember how strange Celestia's ring shadows looked on Saturn before Cassini started returning image after image of them (Voyager images only showed a thin dark band because Saturn was near equinox during the Voyager encounter.)

A few things that people find undesirable about the new star rendering code are also cases of people confusing reality with expectations. There are a lot of objections to the glare halos around planets. The halos are used to give the impression of brightness, since the dynamic range of monitors is so limited. But if you agree that stars should have halos, then so should planets--to an observer on Earth, Venus appears like a very bright star. The fact that planets reflect rather than emit light doesn't matter. The objection is a psychological, not a physical one. Of course, you could argue that halos shouldn't be used at all, but I think that this would be visually unsatisfying.

--Chris

[PlutonianEmpire]

After all, those problems of atmosphere/glare will be part of the past once you will implement HDR, so perhaps in the meantime and to not lose too much time, why not pointing us to the specific values in the code to play with and let's try all together to find a quick fix to those minor problem?

HDR rendering is not a panacea. It will produce very realistic images on systems with modern graphics hardware. But, the images may be *too* realistic . . . For example, for a simple spacecraft visualization, you don't necessarily want the sun to fill the screen with glare when it comes into view. Also, I guarantee that some of the things will see in an accurate HDR view will appear strange even though they're realistic. Since few of us have ever spent any time in space, our expectations of what it looks like are based largely on movies and space art. A few people here will remember how strange Celestia's ring shadows looked on Saturn before Cassini started returning image after image of them (Voyager images only showed a thin dark band because Saturn was near equinox during the Voyager encounter.)

A few things that people find undesirable about the new star rendering code are also cases of people confusing reality with expectations. There are a lot of objections to the glare halos around planets. The halos are used to give the impression of brightness, since the dynamic range of monitors is so limited. But if you agree that stars should have halos, then so should planets--to an observer on Earth, Venus appears like a very bright star. The fact that planets reflect rather than emit light doesn't matter. The objection is a psychological, not a physical one. Of course, you could argue that halos shouldn't be used at all, but I think that this would be visually unsatisfying.

--Chris

Well, IMO, the halos for planets seem a bit too big. how 'bout making them a little smaller?

[Hungry4info]

Honestly, I think that the atmosphere should remain. And if I may add a suggestion? Is anybody interested in making the flare more opaque and perhaps bigger? In my opinion, that would make it appear more realistic. Furthermore, has anybody watched NASA TV where the shuttle's or the station's camera is panning back and forth and the sun comes into view? The lens flares or whatever they are (the polygons of colour that extend in a line from the sun) would be a nice addition. Is it possible to get the flare to rotate around the star as the observer changes view?
For example, if the sun is in the lower right corner of the screen, but still visible, lense flare things (sorry, I know this isn't too descriptive) extend towards the top-left corner. And at the top left corner, the flares extend towards the bottom-right corner. When the sun is at the centre of the screen, the flares would be "on top" of the star, so they wouldn't be as visible. Also, another idea? Is it possible to make the sun "brighter" as it gets closer to the centre of the screen? Perhaps increasing flare size and opacity with proximity?

Well, those are just my crazy ideas.

[ElChristou]

After all, those problems of atmosphere/glare will be part of the past once you will implement HDR, so perhaps in the meantime and to not lose too much time, why not pointing us to the specific values in the code to play with and let's try all together to find a quick fix to those minor problem?

HDR rendering is not a panacea. It will produce very realistic images on systems with modern graphics hardware. But, the images may be *too* realistic . . . For example, for a simple spacecraft visualization, you don't necessarily want the sun to fill the screen with glare when it comes into view. Also, I guarantee that some of the things will see in an accurate HDR view will appear strange even though they're realistic. Since few of us have ever spent any time in space, our expectations of what it looks like are based largely on movies and space art. A few people here will remember how strange Celestia's ring shadows looked on Saturn before Cassini started returning image after image of them (Voyager images only showed a thin dark band because Saturn was near equinox during the Voyager encounter.)

A few things that people find undesirable about the new star rendering code are also cases of people confusing reality with expectations. There are a lot of objections to the glare halos around planets. The halos are used to give the impression of brightness, since the dynamic range of monitors is so limited. But if you agree that stars should have halos, then so should planets--to an observer on Earth, Venus appears like a very bright star. The fact that planets reflect rather than emit light doesn't matter. The objection is a psychological, not a physical one. Of course, you could argue that halos shouldn't be used at all, but I think that this would be visually unsatisfying.

--Chris

Ok, nice to read you, things get much clear. Is the actual rendering (I'm talking about halo for planets) based on observations (photos of planets seen from shuttle or probe)?

Concerning the HDR, IF it's suppose to be so perturbating I think we should have it as an option for hightly realistic rendering over a basic rendering; it could be like a layer the same way as wavelenght filters...

It would be cool to have some feedback from astronauts on this topic, nobody knows one?

[t00fri]

...
It would be cool to have some feedback from astronauts on this topic, nobody knows one?

H?¶, h?¶....

Sure Christophe : My father was an astronaut, my mother was an astronaut, I was created in space, my syster is an astronaut ...

...and my dog was an astronaut.

Cheers,
Fridger

[ElChristou]

...
It would be cool to have some feedback from astronauts on this topic, nobody knows one?

H?¶, h?¶....

Sure Christophe : My father was an astronaut, my mother was an astronaut, I was created in space, my syster is an astronaut ...

...and my dog was an astronaut.

Cheers,
Fridger

well it cost nothing asking... I pretty sure someone over there would be pleased to help...

[neo albireo]

A few things that people find undesirable about the new star rendering code are also cases of people confusing reality with expectations. There are a lot of objections to the glare halos around planets. The halos are used to give the impression of brightness, since the dynamic range of monitors is so limited. But if you agree that stars should have halos, then so should planets--to an observer on Earth, Venus appears like a very bright star. The fact that planets reflect rather than emit light doesn't matter. The objection is a psychological, not a physical one. Of course, you could argue that halos shouldn't be used at all, but I think that this would be visually unsatisfying.

I understand that the halo is used to give the impression of brightness. But let's take Earth/Moon as an example.
What happens now?

At 1 mio km distance:


The moon (bottom right) does not yet have a halo (realistic).

At 2 mio km distance:


The moon is much brighter than earth (not realistic).

At almost 10 mio km distance:


The earth closed on in brightness (realistic). I changed my point of view to compare the two better.

At 10 mio km distance:


Suddenly, both objects become invisible at once, after not having given much the impression of loosing brightness with distance (not realistic).

It's that change of realistic and not realistic views that makes it very annoying. I like the halos, but I don't think you would p.e. see a halo of a small moon of jupiter when it is crossing in front of jupiter (the difference in albedo is just not big enough to account for such a difference in reception). You would rather not see that moon at all due to the lack of contrast. I don't know of a solution though. As it is now, I prefer the old way without halos around moons and planets.

[t00fri]

Chris,

could you please reveal what you star code finally assumes wrto the tunable parameters of your star modelling??

++++++++++++++++++++++++
No one of us can judge the physical correctness of your code if you don't communicate what you have assumed and on what physics scenario you based the rendering...
++++++++++++++++++++++++


I quote some essential comments of people from before you started with the coding:
cf.
http://www.celestiaproject.net/forum/viewtopic ... 30&start=5

For bright stars, the central part of the disc is saturated. The height of the Gaussian is clipped to some maximum level, resulting in a wide saturated region surrounded by a relatively small unsaturated fringe region.

Of course, there is clipping in your above star profile. But given that a Gaussian is determined by 2 parameters, its central height, and its width, I am trying to find out whether both are fixed by physical facts or whether perhaps one is fixed by "fiat"

Just trying to "knock" at the hidden degrees of freedom in your approach

Indeed, there are some tunable parameters that need further consideration. There's the FWHM, which is set to 6 pixels in these images. There's also the saturation magnitude. Stars below this value have centrial peak brightness values less than the maximum representable pixel value. The meaning of the limiting magnitude parameters is obvious; how to choose a saturation magnitude isn't clear; for the moment, I'm just using the same calculation as in 1.4.0.

Fridger that's an interesting approach. For the same lens system shouldn't all background stars have the same half width, saturation notwithstanding? So one might let the halfwidth be fixed relative to the FOV; under the assumption that higher FOV is equivalent to higher resolution, and thus smaller halfwidths (though constant when measured in pixels.)

For the moment, I'm assuming that the halfwidth is a constant size in pixels, not a constant angular size. So, smaller FOV implies higher resolution.

Bright stars fill out the outer fringes of the Gaussian PSF to the point where they become visible above the noise level (here, that's just 1/255, the faintest non-black pixel value.)

and so on...

Bye Fridger

[chris]

Chris,

could you please reveal what you star code finally assumes wrto the tunable parameters of your star modelling??

++++++++++++++++++++++++
No one of us can judge the physical correctness of your code if you don't communicate what you have assumed and on what physics scenario you based the rendering...
++++++++++++++++++++++++

I described in some detail how scaled disc stars are implemented, but there was no response:

http://www.celestiaproject.net/forum/viewtopic ... c&start=15

--Chris

[t00fri]

Chris,

could you please reveal what you star code finally assumes wrto the tunable parameters of your star modelling??

++++++++++++++++++++++++
No one of us can judge the physical correctness of your code if you don't communicate what you have assumed and on what physics scenario you based the rendering...
++++++++++++++++++++++++

I described in some detail how scaled disc stars are implemented, but there was no response:

http://www.celestiaproject.net/forum/viewtopic ... c&start=15

--Chris

I am sorry,

that post I have never noticed. Too bad. Some posts by Malenfant were in-between which made me turn away, I guess

While your cited post contains some formal description of the math that you implemented into the code, there was no comment about the underlying physics!

I can see three separable stages of perception/display that are involved:

-- the theoretical illumination profile due to the given light source. This involves standard theoretical optics, where the size* wave number = R * k of the source and the distance of the observer are the crucial parameters.

-- the observational means used by the observer (naked eyes, camera, telescope,...)
-- the familiar dynamics deficiencies of generic monitors used for the displays


Perhaps it is not useless if I ask some simple questions once more from the beginning:

-- what physical effects do you think the fuzzy (Gaussian) shape is supposed to represent? What is the physical basis for your assumed constant FWHM of the Gaussian??

-- what observational means are we assuming the observer is using? His/her eyes? Or is he/she taking (CCD) photographs?

-- the clipping emulates the bad dynamical response of monitors to a large extent. Agreed?

-- how do we have to implement the transition to the pointlike star limit? What are the relevant physics constraints? How is the given star magnitude transferred to the point stars in this limit? Normalization?

--how do we have to implement the halos, given their assumed role of indicating particular brightness? What happens with the halo in the pointlike limit?


What else?

Bye Fridger

[rra]

what physical effects do you think the fuzzy (Gaussian) shape is supposed to represent? What is the physical basis for your assumed constant FWHM of the Gaussian??

I think the Gaussian effect is OK it s just a matter of scaling which should be appropriate.
The Gaussian shape on the edges would be due to either limited optical MTF or in best case diffraction limit of the optics , be it our eyes or some optical instrument.
In the case of our naked eye the scale (FWHM if you like) is constant,
the efect of the Gaussian edge would be constant , the whole profile is merely a convolution of the "constant" Solar lateral intensity , (some clipping must be assumed) and the Gaussian (or Airy disk) of the optic limits.
I am not sure Chris implemented it such a way ??

Ren?©

[t00fri]

what physical effects do you think the fuzzy (Gaussian) shape is supposed to represent? What is the physical basis for your assumed constant FWHM of the Gaussian??

I think the Gaussian effect is OK it s just a matter of scaling which should be appropriate.
The Gaussian shape on the edges would be due to either limited optical MTF or in best case diffraction limit of the optics , be it our eyes or some optical instrument.
In the case of our naked eye the scale (FWHM if you like) is constant,
the efect of the Gaussian edge would be constant , the whole profile is merely a convolution of the "constant" Solar lateral intensity , (some clipping must be assumed) and the Gaussian (or Airy disk) of the optic limits.
I am not sure Chris implemented it such a way ??

Ren?©

Ren?©, Chris,

honestly, I don't think things are that easy throughout, since Celestia is supposed to cover a huge range of source size/distance, which implies quite different optical regimes from Fresnel to Fraunhofer diffraction.

In the diffractive regime, we must once for all agree on the observational means. In case of a telescope or CCD aperture, a new dimension enters and the infalling plane wave is clipped by that aperture. That is the very reason for the Airy disk pattern, the size of which directly reflects the aperture size! Airy disks are not Gaussians, but rather J1(x)/x Bessel functions as you surely know.

These comments are just to illustrate that we have to commit ourselves pretty much to an observational setup, if we want to model physically correct stars in Celestia!

Bye Fridger

[rra]

Fridger , I do agree with you:

we should commit to 1 observator,
in this case I would vote for some optical instrument using CCD
sensor:
this implies also that the size of the Point Spread Function should vary
depending on the FOV Celestia uses, and also clipping will be seen.
As for the Airy-disk:
a properly designed optical system has no (or at least shouldn't have an)
Airy-disk at the usable FOV, thus the PSF should be in the form of a
Gaussian.


Ren?©

[t00fri]

...
As for the Airy-disk:
a properly designed optical system has no (or at least shouldn't have an)
Airy-disk at the usable FOV, thus the PSF should be in the form of a
Gaussian.


Ren?©

Ren?©,

what do you mean? My properly designed 8" telescope certainly shows a beautiful Airy disk

Of course, with some tricky (transmission) profile across my aperture, I can convert the Airy disk into a Gaussian. The respective procedure is called "apodization"

Here is a summary from Wiki (in case you need it):
http://en.wikipedia.org/wiki/Apodization

Bye Fridger