Celestia Forums

I attach two mockups for alternate (close range) star rendering, proposed by ElChristou. What do you people say? Would this be an improvement over current rendering or not?

Yeah, I suppose that looks comparatively good.

The idea is to have a better rendering waiting for HDR or other complex rendering. It would be a transition rendering. The principal point is to remove the hard edge of the sphere. Also the sphere for any classes apart dwarves is supposed to be pure white and use a colored corona more or less strong to define the color. In the shots above the coronas are color exaggerated to increase the disturbing effect of brightness but seems too artistic for Celestia. Perhaps something smoother would be better...

Personally, I don't think that Toti's disks are much of an improvement. I think it's all way too regular and geometrical.

Bye Fridger

t00fri wrote:Personally, I don't think that Toti's disks are much of an improvement. I think it's all way too regular and geometrical.

Fridger, as said above, it's not question here to do something really fancy, just to remove the hard edge of the sphere that makes no sense at all and why not playing a bit with the corona.

Don't forget that HDR will change the game as we already see on osX with DW work...
Now, more complicated stuff concerning regularity of stars' shape should be discuss in depth for future versions (1.9, 2.0 ?), like the well known case of betelgeuse...

ElChristou wrote:

t00fri wrote:Personally, I don't think that Toti's disks are much of an improvement. I think it's all way too regular and geometrical.

Fridger, as said above, it's not question here to do something really fancy, just to remove the hard edge of the sphere that makes no sense at all and why not playing a bit with the corona.

It's important to bear in mind that the hard edge does make a lot of sense. The sharp edge of the solar disc is apparent whenever the eye isn't overwhelmed by the its brightness--near sunset, filtered by clouds, etc. The apparent blurriness of the sun is a result of the eye (or camera's) limited sensitivity and of scattering in the optical medium.

This is not meant to discourage experiments with star rendering--it's just a reminder that however we decide to render stars, we're making some assumptions about light adaption of the eye or exposure and aperture of a camera.

--Chris

chris wrote:
It's important to bear in mind that the hard edge does make a lot of sense. The sharp edge of the solar disc is apparent whenever the eye isn't overwhelmed by the its brightness--near sunset, filtered by clouds, etc. The apparent blurriness of the sun is a result of the eye (or camera's) limited sensitivity and of scattering in the optical medium.
...
--Chris

I thought we have agreed a long time ago that we shall basically model the appearance of the Universe as recorded by the naked eye (or an equivalent sensor) !?

This then should include typical acceptance effects of the eye-based vision. So if I am looking out of a bull's eye of a space ship approaching another star, I certainly would NOT see a sharp edged star disk, but rather some blinding wobble of brightness...

Bye Fridger

Chris, I do agree (of course!) about the neat disk through atmosphere, and for sure we must find a way to marry both aspect of stars, from the vacuum (deep space or moon for example) with a bright halo (much bigger in reality than this) and once in an atmosphere under precise condition (sunset , clouds) fading the halo to reveal the disk.

Now as I suppose this will be only possible via some complex maths through an interaction with HDR and Mie theory, for now we must do with what we have and I'm not sure the 2 points where the sun should appear as a disk worst the bad rendering in all other cases which must represent 90% of Celestia rendering...

t00fri wrote:Personally, I don't think that Toti's disks are much of an improvement.

These are mockups made in Photoshop by ElChristou, not by me.

t00fri wrote:I thought we have agreed a long time ago that we shall basically model the appearance of the Universe as recorded by the naked eye (or an equivalent sensor) !?

This then should include typical acceptance effects of the eye-based vision. So if I am looking out of a bull's eye of a space ship approaching another star, I certainly would NOT see a sharp edged star disk, but rather some blinding wobble of brightness...

As I wrote to ElChristou, this is the key point here. We need to simulate light perception based on a given sensor's characteristics; otherwise results would present many incoherences. The sun should look hard-edged when its brightness is diminished by a filtering medium, as Chris illustrated. But rendering in the vacuum should be thought as an entirely different effect.

To me the rendering in vacuum should be the priority because after all as far as I know, the specialist for sky rendering is Stellarium meanwhile Celestia is principally based for larger scale rendering... What I have some trouble to imagine is how a cloud map can modify a rendering (in this case fading the corona/halo)...

t00fri wrote:Personally, I don't think that Toti's disks are much of an improvement. I think it's all way too regular and geometrical.

t00fri wrote:I thought we have agreed a long time ago that we shall basically model the appearance of the Universe as recorded by the naked eye (or an equivalent sensor) !?

Then what's your problem?

When I look at the sun, it appears perfectly circular, all the same colour (white, in fact, bright white, blindingly so, too). Huge halo, almost as bright as the sun itself. It's rather large too, taking up a significant part of my field of view when I look at it. I certainly don't see any sunspots with the naked eye, and certainly not limb darkening. Now I realise that the a star won't be blindingly bright on a computer program. However the stars can certainly be improved. Look at, for example, Stallarium's star rendering. Deep Space Explorer is the only application I have ever owned that renders stars even more unrealistically than Celestia.

When I see the sun, it is regular and geometrical.

Sure, 1.5.0pre5 is much better than 1.5.0pre4, but seriously. Go outside, look up and see the sun (should be the brightest thing in the day sky). Now compare that to what Celestia shows.

Post Scriptum: I am willing to entertain the idea that there is something wrong with my eyes, and perhaps they do not function as properly as they should. Perhaps I'm oversensitive to light so that bright things seem much brighter to me than they would otherwise seem.

Hungry4info wrote:Now I realise that the a star won't be blindingly bright on a computer program.

Not blindingly bright, but a quite uncomfortable brightness can indeed be simulated:

Hungry4info wrote:Then what's your problem?

Toti wrote:

Hungry4info wrote:Now I realise that the a star won't be blindingly bright on a computer program.

Not blindingly bright, but a quite uncomfortable brightness can indeed be simulated:

This is now getting much more into the regime where I "don't have a problem" (cite: Hungry4info)

Bye Fridger

Toti wrote:Not blindingly bright, but a quite uncomfortable brightness can indeed be simulated:...

Yes! this is the kind of discomfort I'd like to experiment!
Now looking at this shot, all frames here should be plain white, no? So what about increasing a bit more the glare? (in fact it would be cool to experiment an "avant-go?»t" of wavelength filtering with the stars, allowing to switch to a reduced visible light mode to tone down the halo and reveal an eventual texture...)

Now we still have the problem of the rendering through atmosphere...

Eventually my HDR builds will allow you to control the exposure of nearby stars (already works for planets) with the "<" and ">" keys.

ElChristou wrote:Now looking at this shot, all frames here should be plain white, no?

I do believe so. Viewing the sun from close by should saturate the eyes.

ElChristou wrote:in fact it would be cool to experiment an "avant-go?»t" of wavelength filtering with the stars, allowing to switch to a reduced visible light mode to tone down the halo and reveal an eventual texture...)

This would be pretty good. Let the user decide between the two and thus people who are unsatisfied with the more realistic stars can set, as you said, reduced visible light.

And yes, I do realise that not everybody will want their screen completely white with saturation. This is why I advocate allowing the user to switch between the two.

dirkpitt wrote:Eventually my HDR builds will allow you to control the exposure of nearby stars (already works for planets) with the "<" and ">" keys.

I have been watching the thread where you discuss this. I find the idea wonderful and am very supportive of this. I do look foreward to it's completion, and hopefully if succsesful, integration into Celestia. I would love to be able to help, but I am not very knowledgable about coding in Celestia and such.

ElChristou wrote:Now we still have the problem of the rendering through atmosphere...

This is best left for HDR rendering, where the consequences of light attenuation by an atmosphere gets sorted out quite naturally.

--Chris

chris wrote:

ElChristou wrote:Now we still have the problem of the rendering through atmosphere...

This is best left for HDR rendering, where the consequences of light attenuation by an atmosphere gets sorted out quite naturally.

--Chris

So there is zero chance to achieve the union between both rendering (vacuum/atmosphere) in classic render paths?

ElChristou wrote:

chris wrote:

ElChristou wrote:Now we still have the problem of the rendering through atmosphere...

This is best left for HDR rendering, where the consequences of light attenuation by an atmosphere gets sorted out quite naturally.

--Chris

So there is zero chance to achieve the union between both rendering (vacuum/atmosphere) in classic render paths?

It's possible to achieve some of the effects, but everything should work much more elegantly (and importantly, more robustly) with HDR.

--Chris