Better Galaxy-map stars

Development of artwork, requests, suggestions, samples, or if you have artwork to offer. Primarily for the artists.
Message
Author
User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

#61 Post by Geoff the Medio »

eleazar wrote:Also has the max size of Cores changed between RC4 and RC5, or is it dependent on screen size?
I'm not sure, but I don't specifically recall changing anything relevant to max zoom. At least intentionally...

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#62 Post by eleazar »

eleazar wrote:
Geoff the Medio wrote:
eleazar wrote:does zooming happen in chunky increments, or is that just the way my mousewheel works?
There is a fixed ZOOM_STEP_SIZE, and the zooming function is written such that it only ever multiplies or divides the zoom level by one increment of this value to change the zoom level (ignoring minor rounding quirks). This could be changed, however.
I just wanted to verify that what i'm seeing is normal.

Also has the max size of Cores changed between RC4 and RC5, or is it dependent on screen size?
geoff wrote:For zoom level 8, the highest zoom, this produces a factor of about 1.9, which is used as the scaling factor of the halo relative to the star core. So, if the star core was 20 pixels across, the halo would be 38 pixels, which is just under twice as big by diameter. I found this logarithmic-type scaling gave a nice variation with zoom.
According to my on screen measurements, (of RC5) this is not quite the case. Imagining that stars and halos were solid squares.
At max zoom the core is about 110x110, while halos expand to ~400x400.


Also i'm concerned about the quality of the image expansion:
OK, some of the grunge is the result of the JPG compression of an unfamiliar PC screen capture app, but most of the choppiness was on screen.

Image
Magnified 2X for convenience.

While the problem would be mitigated by expanding the halo less, i'm still concerned. Is this a neccesary evil of a galaxymap that can scale in realtime, or something else?

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

#63 Post by Geoff the Medio »

eleazar wrote:
geoff wrote:For zoom level 8, the highest zoom, this produces a factor of about 1.9...
According to my on screen measurements, (of RC5) this is not quite the case. Imagining that stars and halos were solid squares.
At max zoom the core is about 110x110, while halos expand to ~400x400.
I was actually using the size of the system selection indicator to generate the size of the halo, not the star core itself... So the results were bigger than described. Here's a hopefully-fixed version:
Image
Also i'm concerned about the quality of the image expansion:
OK, some of the grunge is the result of the JPG compression of an unfamiliar PC screen capture app, but most of the choppiness was on screen.
I agree about this... It's even more noticable on the sidepanels with the larger star graphics. It's not something I can do anything about though... Likely tzlaine will have to make any changes. Possibly it can be worked around by using larger graphics, though...?

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#64 Post by eleazar »

Geoff the Medio wrote:
eleazar wrote:
geoff wrote:For zoom level 8, the highest zoom, this produces a factor of about 1.9...
According to my on screen measurements, (of RC5) this is not quite the case. Imagining that stars and halos were solid squares.
At max zoom the core is about 110x110, while halos expand to ~400x400.
I was actually using the size of the system selection indicator to generate the size of the halo, not the star core itself... So the results were bigger than described. Here's a hopefully-fixed version:
Image
:) That's the relative proportion i was looking for.
Geoff the Medio wrote:
eleazar wrote:Also i'm concerned about the quality of the image expansion...
I agree about this... It's even more noticable on the sidepanels with the larger star graphics. It's not something I can do anything about though... Likely tzlaine will have to make any changes. Possibly it can be worked around by using larger graphics, though...?
OK, there are ways to make things look better without replacing the subroutine that actually scales the image, and these are valid even if the scaler looked as good as Photoshop's bicubic interpolation.

* images look better when scaled by powers of 2
* images look best at full size. (assuming the original is good)
So it would be best if the zooming increments hit all the powers of 2: 6.25%, 12.5%, 25%, 50%, 100% etc.

Let's take 8x and 1x zoom as the control points.

:arrow: At 8x, we can assume the player is basking in the glory of our graphics (since unless he has a huge screen, it's impractical to play at that size). So at 8x, let's simply use the core images at 100%, rather than ~85.9% percent as currently. It will make development more straightforward anyway. So star cores will be(considered as solid squares) 128px wide. Halos will be 256 on-screen px wide.

:arrow: At 1x, details of a star blur, and much of the galaxy is in view at once. Haloes are nearly irrelevant. Cores are 16x16 px (again considered as solid squares)— 1/8th of the original image size. Halos have also shrunk to 16x16. According to my scheme they should be virtually invisible.
EDIT: It's possible i've pegged the wrong place for halos to disappear, and that it really should happen at .5x zoom. I can't really be certain without seeing it in action.

If star graphics are made according to my scheme, the game can stop rendering them at or below 1x zoom. If stars are drawn according to a different scheme, there might be a slight hitch as the halos disappear. Of course a coder must evaluate if this bit of optimization is worth doing.


:arrow: Ultimately it may be worth the trouble (in improved performance and enhanced appearance) if artists could provide optimized proxy images at smaller sizes (for stars, icons, whatever), and the game would use whichever version is appropriate, or scale the nearest size. Mac os X icons use this method, and it allows greater clarity at small sizes, and greater detail at big sizes than any scaling algorithm could provide.


:arrow: Image size of Halos:
It may be worthwhile to make the halo source images 256x256, i.e. at 100% while zoomed to 8x. On the other hand, since halos are inherently blurry, a better scaling subroutine should be able to acceptably enlarge them to 200%. If performance is negligibly effected, i'd go with the former.


I didn't provide any illustrations, but it was gorgeous out, and i wrote this outside without a good place to use a mouse, but i think we understand each other now. I can clarify anything that's not clear.
Last edited by eleazar on Wed Mar 21, 2007 9:51 pm, edited 1 time in total.

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

#65 Post by Geoff the Medio »

eleazar wrote:At 1x ... Cores are 16x16 px [and] Halos have also shrunk to 16x16. According to my scheme they should be virtually invisible.
Except the halos and stars don't fill the whole width of their graphic... If there was a particularly large star or a particularly large halo, this size will be proportionally scaled in the on-screen result, and the large halo would remain visible behind a small star at zooms where a large star would hide a smaller halo.
If star graphics are made according to my scheme, the game can stop rendering them at or below 1x zoom. If stars are drawn according to a different scheme, there might be a slight hitch as the halos disappear. Of course a coder must evaluate if this bit of optimization is worth doing.
It already happens actually; if the halo scale factor is less than 0.5, I don't draw it. The effort to so-optimize was and is quite minimal.
:arrow: Ultimately it may be worth the trouble (in improved performance and enhanced appearance) if artists could provide optimized proxy images at smaller sizes (for stars, icons, whatever), and the game would use whichever version is appropriate, or scale the nearest size. Mac os X icons use this method, and it allows greater clarity at small sizes, and greater detail at big sizes than any scaling algorithm could provide.
I don't think this is worth doing for stars... There's not a lot of important or interesting high spatial-frequency internal structure for a star graphic that would be lost by algorithmic blurring / mip-mapping whereas MacOSX icons have lots of little details that really need to be redrawn differently at different zooms...

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#66 Post by eleazar »

Geoff the Medio wrote:
eleazar wrote:At 1x ... Cores are 16x16 px [and] Halos have also shrunk to 16x16. According to my scheme they should be virtually invisible.
Except the halos and stars don't fill the whole width of their graphic... If there was a particularly large star or a particularly large halo, this size will be proportionally scaled in the on-screen result, and the large halo would remain visible behind a small star at zooms where a large star would hide a smaller halo.
I'm not entirely sure i follow.
The 16x16 measurements are again pretending that each graphics was filled with solid color. But i am aware things won't work out quite right if artists try to put together stars and halos of significantly different sizes. Halos of different sizes/colors shouldn't be intermixed with the same core. That would be a bad idea, graphically, anyway. i wouldn't worry about it.
Geoff the Medio wrote:
eleazar wrote:If star graphics are made according to my scheme, the game can stop rendering them at or below 1x zoom.....
It already happens actually; if the halo scale factor is less than 0.5, I don't draw it. The effort to so-optimize was and is quite minimal.
Cool.
Geoff the Medio wrote:
eleazar wrote::arrow: Ultimately it may be worth the trouble (in improved performance and enhanced appearance) if artists could provide optimized proxy images at smaller sizes (for stars, icons, whatever), and the game would use whichever version is appropriate, or scale the nearest size. Mac os X icons use this method, and it allows greater clarity at small sizes, and greater detail at big sizes than any scaling algorithm could provide.
I don't think this is worth doing for stars... There's not a lot of important or interesting high spatial-frequency internal structure for a star graphic that would be lost by algorithmic blurring / mip-mapping whereas MacOSX icons have lots of little details that really need to be redrawn differently at different zooms...
You are right it's not worth doing for the detail in a star, but i'm interested in the star's saturation and brightness.
You'll notice while zoomed at .25x, it's hard to pick out the color of any but the large red stars. At 8x zoom the colors are almost over-saturated and garish. I don't want the 8x stars to look like Christmas lights. For the human eye to easily distinguish 4 yellow pixels from 4 white ones, against black, they must be very yellow. A much milder yellow is easy to distinguish when you have hundreds of pixels together.

I can cope without this feature, but it would certainly be beneficial for stars (and other aspects of the galaxy map.)

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#67 Post by eleazar »

elezar, earlier in this thread wrote: recently noticed that FO's "normal" star types roughly correspond to the basic spectral types of main sequence stars. The continum that goes from big, blue, bright, hot stars —to— small, red, dim, cold stars. The color indicates the FO type name.
Blue = O, B
White = A
Yellow = F, G
Orange = K
Red = M: particularly Red Giants /Super Giants

I've experimented with adding this range of size and brightness to the star types and it looks promising. The fact that it's more realistic is not on it's own compelling, but in this case a dash of realitiy makes the stars more usable and more interesting.
More and more, using red giants rather than red dwarfs bothers me. I'm not really concerned with the science aspect, because even if realism was important, there are many types of stars, and we can't (sensibly) include them all. Star color is primarily important for 1) identifying stars, and 2) determining the probability of various planets.

We have a nice continuum from large, whitish-blue -to- small, dim, red stars, except the red stars are giants rather than dwarfs. There would be no reason for this, except the V.2 req. by some chance happens to use the word "large" in describing the reds.

With large red stars it's not clear which end of the continuum they belong on, the red end, or the big end. Sure the player could learn, but why should they have to? I cannot discern any counter reason to prefer giants to dwarves here.

Image
altered sequence.

So unless someone can provide a good reason, my next iteration of star graphics will include red dwarves, rather than red giants. I wouldn't take so much trouble explaining such a minor point, except that this technically counters a single word of the v.2 reqs.

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#68 Post by eleazar »

I actually have an idea for black holes that might look good if that had halos just like the stars. I'm not certain it would work, but rather than make a special case of black holes graphics on the galaxy map, (as i suggested earlier in this thread) it might be better to render them as the stars are.

If an artist comes up with a good way to use halos, he need simply add the appropriate images, otherwise the game will display the black-holes without halos, if there are no black-hole haloes to be found.

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

#69 Post by Geoff the Medio »

eleazar wrote:I'm not really concerned with the science aspect, because even if realism was important, there are many types of stars, and we can't (sensibly) include them all.
There's not that many, really... Brown Dwarf, Red Dwarf, Orange, Yellow, White, Blue Giant, Red (Super)Giant, Black Holes. White Dwarf and Neutron can be reasonably combined. Black Dwarfs can be ignored, as can the weirder theoretical stuff between Neutrons and Black Holes.
We have a nice continuum from large, whitish-blue -to- small, dim, red stars, except the red stars are giants rather than dwarfs. There would be no reason for this, except the V.2 req. by some chance happens to use the word "large" in describing the reds.
Red Giant stars are interesting to have, as they are a predicted later stage of development for a Sun-like star. They're also pretty cool, due to just being ginormous stars, wheras Red Dwarfs are sort of background filler. Back in the day with my illustrated astronomy book, I destinctly recall reading about Betelgeuse and the to-scale drawing of it next to Earth's sun. I don't remember any Red Dwarfs, though...
With large red stars it's not clear which end of the continuum they belong on, the red end, or the big end. Sure the player could learn, but why should they have to? I cannot discern any counter reason to prefer giants to dwarves here.
It never occurred to me that there was such a continuum... likely because there wasn't one with both main sequence and off-main sequence stars in the mix. So... I'm not sure it's really necessary to force there to be one. For both Red Dwarfs and Red Giants, presumably there is a lower chance of terran planets. And if you make the image for red stars very large, then it will presumably be clear that they are giants, not dwarfs...
I'm not certain it would work, but rather than make a special case of black holes graphics on the galaxy map, (as i suggested earlier in this thread) it might be better to render them as the stars are.
Black Holes currently are rendered just like other stars. If you put an appropriately-named halo image, it will be rendered beneath the black hole...
Image

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#70 Post by eleazar »

Geoff the Medio wrote:
eleazar wrote:We have a nice continuum from large, whitish-blue -to- small, dim, red stars, except the red stars are giants rather than dwarfs. There would be no reason for this, except the V.2 req. by some chance happens to use the word "large" in describing the reds.
Red Giant stars are interesting to have, as they are a predicted later stage of development for a Sun-like star. They're also pretty cool, due to just being ginormous stars, wheras Red Dwarfs are sort of background filler. Back in the day with my illustrated astronomy book, I destinctly recall reading about Betelgeuse and the to-scale drawing of it next to Earth's sun. I don't remember any Red Dwarfs, though...
Red dwarfs don't make dramatic diagrams when compared is size with Sol. But FO doesn't do that anyway. The "interestingness" of a Red giant as FO's largest star type, rather than an equally big blue-ish white star is negligible.
Geoff the Medio wrote:
eleazar wrote:With large red stars it's not clear which end of the continuum they belong on, the red end, or the big end. Sure the player could learn, but why should they have to? I cannot discern any counter reason to prefer giants to dwarves here.
It never occurred to me that there was such a continuum... likely because there wasn't one with both main sequence and off-main sequence stars in the mix. So... I'm not sure it's really necessary to force there to be one...
I don't especially care what kind of planets a particular star type would really have, because 1) it's highly hypothetical, and 2) it's less important than making it easy for the player to guess which types of planets a star might hold.
Thus the importance of a continuum.

Looking at the color -to- planet distribution modifier in the universe table, a crude sort of continuum is strongly implied. A planet common at one end, is usually scarce at the other. A planet common in the middle is less frequent at either end. While the continuum effect is imperfectly realized, it's such an obvious and simple way to order things, that it definitely should be enhanced as the universe tables are adjusted. For the sake of players scouring the universe for colonies, each planet type should be most common around 1 (or perhaps 2) star type(s), and increasingly uncommon further away on the continuum.

I can't figure out the math to make this work, but furthermore (if possible) planet types adjacent to each other on the EP wheel should be most common on stars near each other on the Star-continuum.

:arrow: It "makes sense" (i.e. it is a pattern easily grasped and used by a player) that a race will find more of the most hospitable planets under a star of a particular color, and that adjacent colors will be their next most likely location.

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

#71 Post by Geoff the Medio »

eleazar wrote:Red dwarfs don't make dramatic diagrams when compared is size with Sol. But FO doesn't do that anyway.
I'm not sure what you mean by that, but in battles in which the star system is shown, a star that fills up half the diameter of the system would be quite dramatic, compared to other systems in which the star is tiny. We could even have science-bending races-against-time as a star expands during a battle, destroying planets and ships in pseudo-real-time...
The "interestingness" of a Red giant as FO's largest star type, rather than an equally big blue-ish white star is negligible.
The interestingness of Red Giants is not just that they're as big as Blue Giants.

We could also make them bigger.
...Thus the importance of a continuum.
OK, I can see a case for a logical progression between star types and planet types they contain. But, why can't it be done by just by colour? Or, we could make it a loop, like planet types, with Red Giants being the connection between the two ends of the loop.

User avatar
eleazar
Design & Graphics Lead Emeritus
Posts: 3858
Joined: Sat Sep 23, 2006 7:09 pm
Location: USA — midwest

#72 Post by eleazar »

Geoff the Medio wrote:
eleazar wrote:Red dwarfs don't make dramatic diagrams when compared is size with Sol. But FO doesn't do that anyway.
I'm not sure what you mean by that, but in battles in which the star system is shown, a star that fills up half the diameter of the system would be quite dramatic, compared to other systems in which the star is tiny. We could even have science-bending races-against-time as a star expands during a battle, destroying planets and ships in pseudo-real-time...
On the battle map we could indeed increase the size difference, relative to the galaxy map, but not (i believe) to the size you are suggesting. But that's a tangent. And irrelevant unless you have a compelling answer to my next comment...
Geoff the Medio wrote:
eleazar wrote:The "interestingness" of a Red giant as FO's largest star type, rather than an equally big blue-ish white star is negligible.
The interestingness of Red Giants is not just that they're as big as Blue Giants.
Perhaps to you and a handful of astronomy buffs (do even all astronomy buffs have a special fascination with red giants?). But i still fail to see how it big red stars are better than big blue stars to our players in general.
Geoff the Medio wrote:We could also make them bigger.
Not on the galaxy map. My most recent star graphics are pushing (if not exceeding) the sane upper limits of star graphic size. Even if we could we could that doesn't address my last comment.
Geoff the Medio wrote:
eleazar wrote:...Thus the importance of a continuum.
OK, I can see a case for a logical progression between star types and planet types they contain. But, why can't it be done by just by colour? Or, we could make it a loop, like planet types, with Red Giants being the connection between the two ends of the loop.
It could be done with "just color", but not as nearly as well. A person will have a harder time perceiving an assortment of stars as an ordered series, if different criterion (size, color) imply a different order.

:arrow: Making the continuum obvious is not so vital that major game elements should change to accommodate it, but it certainly trumps one person's personal preference for red giants.

It can't be a loop, because (arranged as you propose) there's nothing visually to make people perceive it as a loop. And (secondarily) if i'm not mistaken, one of the rationales for planet distribution varying by star is the temperature of the star, a variable that doesn't lend itself to a loop.

Post Reply