Better Galaxy-map stars

[eleazar]

Continued from here, because for some reason i can't reply to that topic.
The basic idea is that stars on the galaxy map would look better if they core and halo were separate graphics, the halo being invisible while zoomed out for sharp crispness. EDIT: rejected: There's also the possibility of rotating the halo to make the star "twinkle" on mouseover or while selected. And random combination of star cores and star halos (of the same color of course) multiply the number of unique appearances possible with the same number of graphics.

In the example bellow the halo and core both [edit: start] at 5% of the graphics size. In each successive example the halo is enlarged more than the core, untill finally the core is at 100% of the graphic's size and the halo is at 200%. I think it looks good. For economy the programmer might consider not even displaying the halo graphics at the lowest magnifications.


The largest size in my example is probably unnecessary, and ultimately the core star graphics could probably be shrunk to 64x64, but things can be easily shrunk later.

You can download the yellow test images from here. I think they will work better than breadman's test images because the halos have more transparency, and thus the "beams" will show through better.

[Geoff the Medio]

The different scale factors with zoom is interesting... though I'd like to see it done the reverse way, with more halo and less disc when zoomed out, and more disk and less halo when zoomed in.

My preference is for much smaller solid disc; I'd like pointlike stars with relatively large spikey halos, rather than disc stars with small halos or small discs with large low-transparency circular halos (like many of the stars we have now). So for me, your example discs seem too big, regardless of halo size. Especially when zoomed out, IMO there should be no central disc, and just a point of light with a halo. Even when fully zoomed in, the central disc shouldn't be much bigger than 5 or 8 pixels across...

... Though I'd likely change my mind if we started showing the planets in a system on a map when zoomed in close though, in which case size of star in comparison to planets would need to be evident.

Another alternative to consider is having stars have no halo unless moused over, rather than brightening a bit as they do now. Probably wouldn't work to have only pointlike non-halo then, though...

Also, your pointy rays seem a bit too regularly spaced, thick, and not pointy enough. The look more like wheel spokes than ... whatever the spikey bits are supposed to be (flares either lens or solar...?). I particularly like the spikey bits of yellow1.png and yellow2.png more (though could do without their rather large glow halo over top).

[eleazar]

The different scale factors with zoom is interesting... though I'd like to see it done the reverse way, with more halo and less disc when zoomed out, and more disk and less halo when zoomed in.

Bad idea. The definite shape is needed at small sizes to make the stars distinct from the background. Look at Celestia, i'm doing essentially the same thing with halos.

....So for me, your example discs seem too big, regardless of halo size....

Please take the time to read all the words before rushing to comment. I've already said this:

The largest size in my example is probably unnecessary, and ultimately the core star graphics could probably be shrunk to 64x64, but things can be easily shrunk later.

Another alternative to consider is having stars have no halo unless moused over, rather than brightening a bit as they do now. Probably wouldn't work to have only pointlike non-halo then, though...

There are lots of better ways to indicate mouse-over than the current set-up. As it is now all the stars are dimmed, except when one is moused-over, this muddies up all the stars almost all the time. Counter-rotating halos as Breadmans suggested or an expanding halo would both be better ways to show mouseover. (Though the counter-rotation would be more obvious over time as a means to show selection)

Also, your pointy rays seem a bit too regularly spaced, thick....

Nit-picking. Did you see i called these "test images?" It's a waste of time trying to prefect graphics for a feature that's not yet been coded, by trying to imagine how it will look ingame.

[Geoff the Medio]

...more halo and less disc when zoomed out, and more disk and less halo when zoomed in.

The definite shape is needed at small sizes to make the stars distinct from the background.

Background stars don't have halos, though... So forground stars with (relatively) big halos around tiny discs would still be distinct.

...so for me, your example discs seem too big, regardless of halo size....

Please take the time to read all the words before rushing to comment. I've already said this:

The largest size in my example is probably unnecessary, and ultimately the core star graphics could probably be shrunk to 64x64, but things can be easily shrunk later.

I did read that... A clearer version of my meaning might be that the ratio of halo size to disk size is larger than I'd like, especially for the small sizes.

That said, I might reconsider with your style of star. A significant source of the objection for me with the current ones is their large, low-transparency halo that has a rather noticable edge to it. Your solid star discs might work better, though... Will have to play around with it in-game, I guess...

As it is now all the stars are dimmed, except when one is moused-over, this muddies up all the stars almost all the time.

You've got a bit of a unique perspective on this... Originally, all the stars appeared as the unhighlited ones do now, which I assume is how they were originally intended to look. When highlighting was added, just that one was made brighter, not the rest dimmer.

That said, the equivalent observation that all the stars are too dim is a good one...

[eleazar]

...Will have to play around with it in-game, I guess...

That is the ultimate test.

As it is now all the stars are dimmed, except when one is moused-over, this muddies up all the stars almost all the time.

...Originally, all the stars appeared as the unhighlited ones do now, which I assume is how they were originally intended to look. When highlighting was added, just that one was made brighter, not the rest dimmer.

I can't check out the code, but as someone who makes his living looking at things and noticing differences, it's obvious that the star graphics are dimmed (i'd say displayed at about 85% opacity) except when moused-over. Notice that none of the cores are pure white in the galaxy map.

[Geoff the Medio]

...it's obvious that the star graphics are dimmed (i'd say displayed at about 85% opacity) except when moused-over. Notice that none of the cores are pure white in the galaxy map.

Could that be due to partial transparency in the images, causing some blending of white star with black background? I don't have a png editor that can show me the alpha channel to check for myself...

[eleazar]

...it's obvious that the star graphics are dimmed (i'd say displayed at about 85% opacity) except when moused-over. Notice that none of the cores are pure white in the galaxy map.

Could that be due to partial transparency in the images, causing some blending of white star with black background? I don't have a png editor that can show me the alpha channel to check for myself...

Theoretically. I checked and a few of them are slightly transparent in the middle. This is a bad thing in itself, but can't account for the majority of the muddiness i'm seeing.

[Geoff the Medio]

...can't account for the majority of the muddiness i'm seeing.

I took a quick look through SystemIcon.cpp. In the SystemIcon constructor, there is a line:

Code: Select all

AdjustBrightness(m_default_star_color, 0.80);

which I suspect is the source of the "problem". This was added around late Januray of 2006, and seems to be intended to make the stars dimmer so that they can be later brightened by rollover. So it might not be your unique perspective so much as my not noticing that all the stars got dimmer between v0.3 and v0.31.

[tzlaine]

FWIW, AdjustBrightness() affects only the non-alpha channels.

[eleazar]

FWIW, AdjustBrightness() affects only the non-alpha channels.

Hmm, so i wasn't entirely right. I still hold that the decreasing the brightness (via alpha or not) is a non-optimal means of showing non-selection.

Another way to show selection/mouse-over with separate core and halo graphics is to have the halo "throb" i.e. expand and contract rhythmicly. This would save the trouble of needing 2 halo images per star.

[Geoff the Medio]

I think that we can find a better / more obvious way to show mouseover of systems than anything involving changing the appearance of the star itself. Using any of throbbing halo, counter-rotating halo, or brightening seem like they might conflict with other uses for such animations we might have to indicate gamestate.

So how about a mouseover indicator? Something not too big, most likely...
.
It should not conflict with any system selection indicator, ala pd's (intended to be around the system currently shown in the sidepanel, whenever I get around to implementing it):

[eleazar]

...Using any of throbbing halo, counter-rotating halo, or brightening seem like they might conflict with other uses for such animations we might have to indicate gamestate.

I hesitate to ask, "what other uses?"

So how about a mouseover indicator? Something not too big, most likely...

It should not conflict with any system selection indicator, ala pd's (intended to be around the system currently shown in the sidepanel, whenever I get around to implementing it):

Bah, you don't just randomly pick to different styles of reticule for mouse over and selection. Icons indicating similar information should have similar appearances. However it's easy change the graphic once the code supports it.

I also wonder about pd's system selector. It's cool, but stylistically it doesn't seem to have anything to do with the current UI. Is it left over from an abandoned redesign of the UI?

[Geoff the Medio]

I hesitate to ask, "what other uses?"

If we had a star surrounded by a Dysonsphere or Startree, it wouldn't necessarily have a halo. Black holes might look best with slowly spinning accretion discs and no halo. Neutron stars currently do, but might look better without halos.

Bah, you don't just randomly pick to different styles of reticule for mouse over and selection. Icons indicating similar information should have similar appearances. However it's easy change the graphic once the code supports it.

Right, so the point now is whether using any sort of reticle for mouseover is good / bad and why...

I also wonder about pd's system selector. It's cool, but stylistically it doesn't seem to have anything to do with the current UI. Is it left over from an abandoned redesign of the UI?

I think he just made something that could be used now (then), not really worrying about how it fits into the overall scheme. It will almost certainly need replacing at some point.

[Daveybaby]

A good way of giving mouseover feedback is to give an indication of what will happen if the item is selected, i.e. what will happen if the user actually clicks on the object under the mouse. Ergonomics 101.

In this case you would display a fainter/smaller/thinner version of the system selection indicator around the mouseovered system. Then if the player clicks on the system, the indicator 'solidifies' into its full state.

[Geoff the Medio]

Good point Daveybaby. That also fits thematically with the current projected fleet move lines that appear connecting your fleet to wherever you'd order it to go to if you right-clicked, which appear when you mouse over a system while a fleet window is open.

However, I must ask:

... give an indication of what will happen if the item is selected ... Ergonomics 101.

Does your desk chair have a faint picture of you sitting painted on it...?