|
I'm looking for some artist folks to take a crack at a new color scheme for the map analyzer (original post).
Update: Another release, 1.4.6 (unofficial) with new colors and new modes!
Skip to the 4th page for instructions.
Update: another unofficial release with more color bands for the influence heat map.
1.4.5 (Unofficial) See posts below for details of 1.4.5 changes over 1.4.4.
What's the problem? A few new things are coming in the next release that require some nice color schemes.
1. Openness levels
The openness levels are probably one of the most helpful outputs of the map analyzer besides the distance measurements. It shows, graphically, how "open" a point of the map is. In the next official release of the map analyzer I'm going to add another level of gradient to this output so a human can more easily detect the differences in openness.
This image shows the current official release 1.4.3 on the left, and what the next release will show on the right. The color scheme on the right NEEDS SOME LOVE!
2. Influence Heat Map
This is a new feature for the next release. What it shows you, for any given point on the map, is which starting location exerts more influence over that point. Right now I'm doing it as a bunch of gradients to a middle color, then the same gradients reversed to the other base. There is a visual balance to that so users can tell where the influence differences are. For instance, in Scrap Station you can see that the influence over the watchtower is even, but the green band of influence of the 12 o'clock base extends past the cliff between its natural and its third base. I think this tells you 12 can (slightly) better protect its third base, but balance discussions aside, the influence heat map needs a nice color scheme!
So how can you help?
Download this unofficial release of the map analyzer. It has already been set up to analyze Blistering Sands and Scrap Station (included in the download) so just unzip it, double-click the sc2mapanalyzer.exe file and it will produce images in the output folder.
Cool, now how do you play around with the colors? Open the text file "colors.txt" and see I've written instructions on defining colors, and given some descriptions of the colors.
For this community experiment I've moved the relevant colors to the top of the file.
- Colors opennessLow, Mid1, Mid2, High are the endpoints of gradients for the openness levels. Low is the color of a cell next to a boundary, and High is the color of a cell in a wide open space.
- Influence colors have two parts. "influence1" and "influence2" are the color of text related to Start Location 1 and Start Location 2 that are being compared by the analyzer (right now they are set to blue and yellow). Then there are "influenceMid1, 2, ... 7" which are the endpoints of the gradients in the influence heat map. You'll notice color 1 equals color 7, 2=6, and 3=5 because the gradients are reflected, color 4 is the color that shows areas of equal influence on the map. In the initial color scheme for Scrap Station shown, the color4 for equal influence is red.
- Colors terrainEvel0, 1, ... are the colors of the cliff levels. The terrain is rendered in the background of all other output images. I chose black for chasms and greys for playable elevations, so colored information will show up nicely. I think a talented graphic artist will know how to adjust these to better blend with the colored information of different output images.
Once you've fiddled with the colors, double-click sc2mapanalyzer.exe again and it will generate the images with new colors.
That's it! If you have questions or cool color schemes to show, please post back. I suppose we can do a community vote about which scheme to use for the official release, but nothing is stopping us from including a bunch of color schemes to choose from.
|
I'm not sure if I like the new openness colors in your image. They look... washed out. I guess that's why you're asking for help. Though, I like the heat map colors. The 'cooler' colors are the less contested areas and the 'warmer' colors are in the more contested areas. It makes sense.
I'm not a graphic artist but I'll probably download it later tonight and mess with the colors and see what I can do.
Keep up the good work.
|
I've had a little fiddled with the colors, and I think maybe a red vs blue scheme might work for the Influence map (see spoiler).
+ Show Spoiler +
One thing I ran into was that the height levels kinda ruin the colors you choose. Like in your example image. The colors get a washed out look from the greys between the tiles. I was thinking maybe you could visualise height by making the tiles darker in an other way. Maybe like this:
![[image loading]](http://imgur.com/jZniT.jpg)
(Zoomed in) + Show Spoiler +
Now I have no idea if this is at all possibly, or if you even like the idea. But it might be something to think about 
PS, looove the Map Analyzer!
|
On October 21 2010 03:48 BoomStevo wrote:
I'm not sure if I like the new openness colors in your image. They look... washed out. I guess that's why you're asking for help.
Yep, the new openness colors are gross. I added a new pink-ish value beyond blue, and I compressed the whole openness scale so it is more clear to a human where the differences are. The new code is what I want but the colors need some love.
On October 21 2010 05:23 burningDog wrote:One thing I ran into was that the height levels kinda ruin the colors you choose. Like in your example image. The colors get a washed out look from the greys between the tiles. I was thinking maybe you could visualise height by making the tiles darker in an other way. Maybe like this:
I agree, and the washed-out colors were actually me attempting to "tighten" the grey values as close as possible, but still be able to distinguish cliff levels, and have the various gradients for openness render nicely. My first try failed, and so why not ask the talented TL community for help?
Currently I have no internal support in the map analyzer for loading in an image to blit into the output image, so I will hand code a version of your proposal with dithered shades to see how it comes out.
EDIT: oh yeah, burning dog, about the influence heat map--my first attempt had it like you showed, with one color at one base gradually becoming another color at the other base. The problem with this is that if I look at two parts of the map that should correspond, like one base's natural versus the other natural, its hard to compare them. Is this blue as blue as that red is red? So I switched it where the colors are the same at the extreme ends, and there is a central color where influence is even. This shows us something cool in the Scrap Station image--you can see very clearly a difference between the bases by looking at that cliff between the natural and the third base, the green band of influence is the same value but it appears in a different place because the map is asymmetric.
|
|
|
Yummy! Post your color definitions, too!
|
On October 21 2010 08:06 dimfish wrote:
Yummy! Post your color definitions, too!
Sorry, here you go:
opennessLow = 0x0000FF
opennessMid1 = 0x00FF00
opennessMid2 = 0xFFFF00
opennessHigh = 0xFF0000
influence1 = 0x00FFFF
influence2 = 0xFF00FF
influenceMid1 = 0x0000FF
influenceMid2 = 0x00FF00
influenceMid3 = 0xFFFF00
influenceMid4 = 0xFF0000
influenceMid5 = 0xFFFF00
influenceMid6 = 0x00FF00
influenceMid7 = 0x0000FF
They're pretty easy colors.
I followed your example on the influence colors, but I went with something a little different for the openness.
|
Interesting, nice and simple. I know you are an analyzer veteran--can you throw a map with lots of cliff changes and all the elevations to see how the colors come out? Steppes of War I think is good for that, just drop a copy in the unofficial release and it grabs all maps in there
|
|
|
hmmm im thinking stevo's summary is quite good
personally i'd invert the heatmaps such that red was the base and dark blue was the even territory [or even purple; more colours in the gradient would make the colours a little easier to distinguish]
|
although burningDogs colors look quite good I think BoomStevos simple colors are perfect for the task. while I feel it is probably too colorfull and one could try to go with less color and more light and dark contrasts it does a really good job to show differences in height.
i think BoomStevos picture looks good, too, because it is so similar to heat-scan visualization - an apearance one is used to.
i want to point out that it could help to have different task of the analyzer displayed in different colors. blue-green-yellow for openness and orange-red-violet for influence. do i ave to try this out to make it valid or could somebody pick that idea up and try around? it could help to have more distinctive pictures.
|
Excellent stuff so far. I've updated the OP with a link to a new unofficial release, 1.4.5, with some changes.
1. There are 11 colors to use for the influence heat map, up from 7.
2. The influence heat map has a non-uniform scaling for the color gradients. The colors at one end (near a start location) only spread over a small radius, while the colors representing equal influence were spread all over the center of the map, so my strategy is to widen the influence range of colors near the start location and tighten colors in the center. If this sounds like jibberish, ignore it! 
3. I am very much liking BoomStevo's openness colors (and notice he changed destructible rocks to a nice grey because his openness=0 is blue, all good stuff). I did a preliminary adaptation of his influence map that looks a bit disgusting, but I think we should all try something more like this, where colors next to each other are very different. If you put influence1=red and influence2=orange then they are so close in the output image that you don't get very much information! Look at the asymmetries that are revealed when you pick colors that are very different, so you can really see lots of bands of colors:
+ Show Spoiler [Influence, more colors, harsh transiti…] +Look at the yellow outside the natural, 12 o'clock is (slightly) more influential over the bottom of its ramp. Look at the teal band, 12 o'clock CLEARLY has an edge over its third base. Look at the equal influence, it shows equal near the watchtower but the gold base is a hair more to 12 o'clock's side. Look at teal/nasty grey: 2 'oclock has a little more influence down the big ramp.
Can we find a prettier set of harsher transitions for influence colors??
|
How about using less colors for the infuence map, but repeating them? In the pictures in the spoiler tag I used Cyan, Magenta and Yellow as the only 3 colors and I think it shows regions pretty well. But maybe you lose some other type of info in doing it this way...
+ Show Spoiler +
|
On October 22 2010 03:48 burningDog wrote:How about using less colors for the infuence map, but repeating them? In the pictures in the spoiler tag I used Cyan, Magenta and Yellow as the only 3 colors and I think it shows regions pretty well. But maybe you lose some other type of info in doing it this way... + Show Spoiler +
Boom, baby! This looks great!!!!
Anyone wanna back me up that BoomStevo's openness scheme is awesome and burningDog's repeated and delicious influence scheme should be on a hot fudge sundae?
I'm still open to more ideas, but for now I'm going to change the official release colors to these. Beautiful, and great work everybody.
Oh yeah, everyone including BoomStevo and burningDog--post the color definitions for me and anyone else who wants to play with what you come up with!
|
I was messing with trying to use light/dark colors next to each other instead of contrasting colors. I was trying to keep with the cold to warm areas theme and still have it look nice.
+ Show Spoiler [Images] +
+ Show Spoiler [Color Definitions] +influenceMid1 = 0x00FFFF
influenceMid2 = 0x000040
influenceMid3 = 0x00FF00
influenceMid4 = 0x400040
influenceMid5 = 0xFF0000
influenceMid6 = 0xFFFF00
influenceMid7 = 0xFF0000
influenceMid8 = 0x400040
influenceMid9 = 0x00FF00
influenceMid10 = 0x000040
influenceMid11 = 0x00FFFF
burningDog's scheme make the bands/contrast easier to see though.
|
Oh right. Here are the numbers (I put BoomStevo's openness scheme colors in there aswell):
+ Show Spoiler +opennessLow = 0x0000FF
opennessMid1 = 0x00FF00
opennessMid2 = 0xFFFF00
opennessHigh = 0xFF0000
influence1 = 0xffffff
influence2 = 0xffde00
influenceMid1 = 0xfff200
influenceMid2 = 0x00aeef
influenceMid3 = 0xec008c
influenceMid4 = 0xfff200
influenceMid5 = 0x00aeef
influenceMid6 = 0xec008c
influenceMid7 = 0x00aeef
influenceMid8 = 0xfff200
influenceMid9 = 0xec008c
influenceMid10 = 0x00aeef
influenceMid11 = 0xfff200
defaultTxtFg = 1.0, 1.0, 1.0
defaultTxtBg = 0.0, 0.0, 0.0
footerTxtFg = defaultTxtFg
footerTxtBg = defaultTxtBg
terrainElev0 = 0x000000
terrainElev1 = 0x000000
terrainElev2 = 0x212121
terrainElev3 = 0x424242
pathingBlocked = 0.7, 0.0, 0.0
pathingClear = 0.0, 0.7, 0.0
regMinerals = 0x53d5f2
HYMinerals = 0xf0c20b
regGeyser = 0x5ec117
destruct = 0x424242
LoSB = 0x424242
mainChoke = 0xffffff
watchTower = 0xffffff
watchTowerRadius = 0xffffff
startLocation = 0xffffff
baseClass-semiIsland = 0xffffff
baseClass-island = 0xffffff
baseClass-natural = 0xffffff
baseClass-third = 0xffffff
spaceInMain = 0xffffff
shortestPathGround = 0xfff200
shortestPathCWalk = 0x00aeef
shortestPathAir = 0xec008c
|
|
|
Wow, nevermind, I'll let you guys keep playing with it a while, there is a lot of stuff on my todo list for the next official release!
I think a good test of whether the color bands show the best info is if you see the matching bands clearly (like a strip of yellow for one base and a strip on the other end, easy to compare them) for Blistering Sands and Scrap Station, then we're good. Blistering is a map where the players are at opposite ends and the rest of the map forms a wide line between them, like tennis players with a net in the middle. This is one common pattern influence should give good info for. Scrap Station is a reflection symmetry map where the players have a path to get to each other, and there is a "far side" of the map from the shortest route. This is another common pattern (think of what the map looks like when players spawn close to each other on Metalopolis) that the influence should look good on, too.
And if we just use 2 spawn maps then you don't get all the spawn pairs of a four player, since we're just in test mode anyway.
So yeah, keep posting as you tweak the schemes, but already they're MUCH nicer!! And I'll keep taking requests for alternate code updates (even MORE color bands?).
|
playing around with it a bit ending with yellow-green-blue-blue.
not as much contrast as BoomStevos.
+ Show Spoiler +
+ Show Spoiler +
opennessLow = 0xFFCC00
opennessMid1 = 0x33CC00
opennessMid2 = 0x3399FF
opennessHigh = 0x0000CC
|
Steppes openness map with the same colors as the influence map:
+ Show Spoiler +![[image loading]](http://imgur.com/ZmW6e.gif) opennessLow = 0x00aeef opennessMid1 = 0xec008c opennessMid2 = 0xfff200 opennessHigh = 0xFFFFFF
|
|
|
|
|
|
|
|
|
My theory on the openness colors were that the opennessHigh color was a warning color and the rest were 'safe' colors. So you would take a quick look at the image and if you see a lot of red, it would stand out because you probably don't want everything to be extremely open. You want most paths to be green to yellow and contrast isn't as important since you don't need as much precision.
With burningDog's openness colors, there is definitely more contrast and you can better see the changes in openness but I think the meaning of the colors is lost because the magenta color stands out to me, but then it is used on the middle openness that shows up the most. Instead of warning us when something becomes too open.
|
|
|
Yeah, exactly what Barrin said. Openness is not good or bad. You really want the various bands to stand out though, so if your map has different openness in areas that are supposed to be balanced, then it should stand out in the image.
One of my goals for openness is to detect that the choke between the natural and third bases on Scrap Station is tighten for the lower spawn and more open at the top. I'd like for this difference to be noticeable just from looking at an openness image. Many players found this by experimenting in the game, but the map analyzer should nail stuff like this in a snap.
|
|
|
On October 22 2010 07:42 Barrin wrote:
This is intrinsically connected to the # of color variables that shift over a specified area of space. In other words, I really think you're the one with the proper tools to change this.
But for us, it seems to me that in order to do this is it extremely important to make sure that opennessMid1 and opennessMid2 should contrast the most. opennessLow and opennessMid1 should also contrast a lot, depending on the map really.
If you open up constants.txt you can change the variable "maxOpennessRender" or something like that to set the scale!
So it goes like this: when a map cell is directly adjacent to a boundary is has openness zero, it always renders the openness minimum color, say red (I think it is currently red?)
Let's say you made a completely empty map that was 64x64. I think the current opennessMax constant is 14 or 15, so all the cells in the middle of your map that are farther than 15 from the edge will all be the opennessMax color. So if you make the opennessRenderMax constant 32, then the openness gradient would stretch to the center of your 64x64 map.
So yes, you can stretch or squish the scale as you like, but if you want more levels you gotta tell me and I'll code it in, and create the extra colors like for influence.
|
|
|
Maybe the maxOpennessRender should be set to 9 (or maybe 12 + one extra level). Somehow that feels like it makes sense to me, because alot of buidings are 3x3.
I'm done messing arround with the analyzer for today, but maybe it can help someone (if it actually does make any sense ).
|
|
|
Barrin, what your saying is true for a hallway that is exactly 20 cells wide (10 cells of good openness colors in each direction). But the analyzer actually does a really good job of calculating when a point is like 7.35 cells away from a boundary, whatever the angle is to the closest thing. So actually lots of cells on the map are colored in the gradient with in-between values.
Still, if it looks good, it looks good. and YES YES YES we need a standard, like the current version has become a standard, so when someone looks at the openness image of a new map they can put it in perspective compared to maps they know.
Any takers for even more openness colors, or are the current levels with a new max of 10/12 or some such looking good?
Any suggestions what the benchmark maps are for openness? What maps show extremes in tight and open in the most critical areas (main choke, natural + surroundings)?
|
|
|
|
|
barrin:
the problem in picking colors with strong contrast is that the actual space of the map gets separated into several stripes. this is not very intuitiv since one could get the impression of a narrow space.
i like the concept of sticking to 'neutral' colors. did the same. no comments on my try (first page)? probably more contrast needed?
|
|
|
Barrin, nice analysis. I'm tinkering with new color schemes, too. Can everyone post their latest versions (if newer than what's posted already?) and I'll collect them into a nice post with lots of styles side-by-side. Maybe not time for choosing a final version just yet, but I want to collect everyone's input up to this point.
It'll be something like this:
Openness Schemes
-BoomStevo's dark blue cold to red warm, colorful, I think blends well with other colors in the image
-Samro225am's yellow/green/blue is stylish, but maybe needs some color reordering to achieve a little more contrast?
-burningDog's wicked evil blue-magenta-yellow-white, which is also stylish, and for some reason the blue edges really bring out the map outline which gives the user an immediate frame of reference. I think the inner colors could use more contrast? Maybe just tweak the yellow so the in-between orange color pops out more?
-Barrin's grey-green-blue, again a stylish set, my only concern is that it generates a lot of grey which blends with the terrain, BUT there have been proposals to help the terrain stand out more...
Influence Schemes
-BoomStevo's rainbow
-BurningDog's Cyan, Magenta, Yellow, which I think reads really, really well in terms of being able to pick out the bands, and with a minimum of contrast
-Barrin's high-contrast, information-is-most-valuable scheme
I'll use either burningdog's or Barrin's terrain elevation adjustments for the round-up. burngingDog flattened the lowest elevation and the 1st playable to zero to keep the average color dark, clever idea:
terrainElev0 = 0x000000
terrainElev1 = 0x000000
terrainElev2 = 0x212121
terrainElev3 = 0x424242
and Barrin's is similar, but not as tight a range: 0, .2, .4, .6
If you have input on which of these terrain color sets works better for you, give a shout!
Oh yeah, and burningDog, I didn't forget about your idea to do terrain as dithered points, but I think the initial round of schemes showed it could be done much better than my initial attempts. Your new way does require some coding, so I'll put it to you: are you still interested in seeing a version of that or do you think the color schemes coming out have solved the problem? If you still want it I'll get it for ya.
Don't forget about burningDog's suggestion to find a slick version of a color scheme you're already using: colorpicker.com
|
The dark elevation scheme Barrin and I used looks the best, but it can be hard to see wat level your looking at. I mean, we all know maps like steppes so it's easy there, but on new maps it's pretty hard to tell.
But at the same time I kinda don't like my dithered example anymore either, it's to noisey
Maybe something like this. Where there's a 4 pix square at the center of each tile.
lvl1 black, lvl2 grey and lvl 3 white. (or maybe lvl2 = no square in the middel)
![[image loading]](http://imgur.com/9uxR5.gif)
But before you start the coding, it's probably best to hear what other people think about it.
So, what do you people think about the height map representation?
|
Mmm, that could be cool, I'll vote for a square on low and high, nothing on mid, and the exact coloring can of course be in the colors.txt so it's easily played with. And then you want the crosses that are currently the terrain color to always be black, to really contrast whatever color information is on the map cells
|
+ Show Spoiler +
slightly more contrast while keeping it easy to look at, stylish - it ended up pretty much like the old version :D the main difference is that the yellow and lime green work better together in my opinion than the orange and green in the original version by dimfish, because the reddish orange contrasted a lot. darker blue and violet are easy picks, so the colors form a gradient from light to dark as well as from yellow to blue (and violet).
i like burningDog's suggestion a lot but i am not sure if the difference is strong enough between levels. when the crosses remain black throughout all cliff heights 4/5 of the height indexing remains the same. might be an idea to only use crosses and go from black to grey to white.
|
|
|
|
|
Barrin, my thoughts
Openness, the information is crisp, so I think your general strat of dark, light, light, dark and whatever opennessRenderMax you picked are very good (12?) but I honestly do not like the inner green and red.
And for the influence, I think your post on page 2 is better. White as the equal influence shows better, and the colors close to the basest gave much clearer bands for comparison.
I think burningDog's influence from early on really, really, shows great bands, its so easy to look at a ramp or a space and compare it to the opposite end. Maybe you could try working your black/white into his color scheme and see how it goes?
On October 22 2010 03:48 burningDog wrote:How about using less colors for the infuence map, but repeating them? In the pictures in the spoiler tag I used Cyan, Magenta and Yellow as the only 3 colors and I think it shows regions pretty well. But maybe you lose some other type of info in doing it this way... + Show Spoiler +
Actually, now that I think about it, maybe its something about cyan/yellow/magenta none of which has a pure RGB value in it that makes those gradients pick up really well... Who knows color theory?
|
|
|
|
|
On October 24 2010 06:24 Barrin wrote:
@samro yes it does look better but I really reccomend a scheme like this
opennessLow = dark color
opennessMid1 = light color
opennessMid2 = light color
opennessHigh = dark color
right now you have
opennessLow = light(er) color
opennessMid1 = light color
opennessMid2 = dark color
opennessHigh = dark color
i really think that opennessHigh should contrast a lot with opennessMid2 so it's very obvious which places are the most open.
i was aware of it. i also pointed out in my last post that i thought it was a good idea to have a color gradient as well as a lightness gradient.
having read your suggestion i tried around a bit in order with dark-light-light-dark and it works well. at the same time a narrowed the color palette with a dark yellowish green.
when opennessLow and opennessMid1 have a similar color the image signifies quite well were space will feel small.
+ Show Spoiler +
|
|
|
On my colors
All constants unchanged, float opennessRenderMax = 14.0
You can try it:
opennessLow = 0x56630D
opennessMid1 = 0x6CE362
opennessMid2 = 0x33AAFF
opennessHigh = 0x7800C2
I think what works so well is the small nuances. the dark green actually is very yellow or warm, the light green has a tendency into cool blue. I'll try to use an even warmer color for opennessLow.
On this discussion
You analysis are convincing but I think one could also try with a more playfull attempt. this might be also an answer why so few people participate in this dicussion. there are many people with good colorschemes here, there is a relative high level of discussion going on and everybody seems to put quite some thought into this - it is just not very playfull although very sexy indeed - I just love maps. while I am more an analytic guy I tend to leave that a bit away when trying around with colors, sothat it feels natural, not too logic and stringent. this might be a reaction to your very analytic apporach. I think your thought help everybiody a lot to understand what is going on, but also it might put of some people who are less interested in a very strong effort. do not get me wrong such analysis have to be done. it is just a lot to read for somebody who is not in maps (cartography, not leveldesign) or graphic design etc.
but don't feel bad about pushing this discussion's level up. groups often work quite well in such a size like 5 people because everybody can contribute and actually take the time to look at everbody's effort. if 15 people try to contribute to this thread there might be a lot of post one misses.
On your new colors
your analysis is very strong and I like the outcome. I would probably switch the blue to a very light color sothat it looks white but probably with some green or blue touch. this is my only critiscism to your apporach. having a strong idea yopu end up with very strong and pure colors. probably you now want to look into the very detail of each color: is there a lot of blue in the red? check the hex code for dark purple: #800080. Or is the white a bit blue-ish? does the yellow acually lean a bit into the the green spectrum? then you can configure the colors a bit more and they end up easier to the eye (this does only apply to blue in your map) and blend well while they give information to the expert.
A new analysis on my own stuff
In my last try I did pretty much this:
Y(G)-G-B-V (dark warm color - light cold color - light cold color - dark warm color)
I think with a higher resolution one could see all the details but until then I have to put more yellow into my dark green and more red into my purple. Also I will try to pick a darker blue while hoping that the lightter pruple and the darker blue willl not split but blend well.
edit:
New image
I change the colors and are quite happy with the outcome. while they still blend well the give more information as there is more contrast. big plus is that there is no red used (in my humble opinion). somehow red is a 'very loud' color...
here is the code:
+ Show Spoiler +
opennessLow = 0x544C0C
opennessMid1 = 0x6CE362
opennessMid2 = 0x1873B5
opennessHigh = 0x9400F0
and the image:
+ Show Spoiler +
|
|
|
am I wrong or is there no yellow spike in position 2 on your image?
probably I misunderstand your point, but when you say that there should be spikes (more visible information once you look at the details), then there should be more contrast. when you use very similar tones like red and yellow they look orange anyway. when you then add orange as the tone in between than it looks good, but it is harder to see the information. I mean, we need more spike!
this is the reason why i changed my colors: blending because they are at the right position with the color spectrum (Y-G-B-V) but switching it to dark-light-dark-light (but not too strong) and using different tones (violet/pruple is quite red while green is pretty blue) to separate them enough from another sothat the blueish spike is visible.
i do not want to agument that my colors are supperior, actually i think there is a wide variety of colors that could work well, but as far as i understand in your example there is more info in position 1 and 2 (blistering sands - pretty much the best map for analysing the analysis)
|
|
|
This is sick, I love that you all are getting into this. I agree that Samro's colors look good, show the details, and also they play well with the terrain colors.
So you noticed that the LoS blockers get rendered on Blistering Sands right? Barrin helped a lot with cataloging all the doodads and I'm writing up their coded footprints. In the next release we'll do new color schemes and support every flippin' doodad you can place currently. I'm psyched!
Who else is working on schemes? Is the general consensus that a variation of burningDog's influence and samro's openness are leading the pack? Barrin's of course still tinkering, but I want to hear whether anyone else is playing around.
|
On October 26 2010 06:05 dimfish wrote:This is sick, I love that you all are getting into this. I agree that Samro's colors look good, show the details, and also they play well with the terrain colors. So you noticed that the LoS blockers get rendered on Blistering Sands right? Barrin helped a lot with cataloging all the doodads and I'm writing up their coded footprints. In the next release we'll do new color schemes and support every flippin' doodad you can place currently. I'm psyched! Who else is working on schemes? Is the general consensus that a variation of burningDog's influence and samro's openness are leading the pack? Barrin's of course still tinkering, but I want to hear whether anyone else is playing around.
mea culpa - i did not even try to play around with the influence map although I followed your discussion.
do you think influence and openness should share a style or color scheme? I would definately want to take a shot at the influence colors tomorrow.
i saw the LoSBs in the map but forgot to comment on that: it is great. having all doodads and their collision represented in the mapanalzer is sick XD
|
|
|
burningDog's incluence colors are great for seeing the bands of influence since they contrast so well, but I have a problem with only using 3 colors instead of 6. The problem is because the colors lose meaning and you have the same color representing two separate areas of influence. For example:
+ Show Spoiler [burningDog's Influence] +![[image loading]](http://i.imgur.com/ykklV.gif) Reference from burningDog's post.
Look at the Steppes of War image using burningDog's influence. You will notice that the main and the third are both largely blue, yet the difference in influence is huge. The main has an influence of about 96% yet the third has an influence of 55% yet they appear as the same color of blue. Since the main and the third are actually pretty close, this could add to the confusion. I'd like it if you could find a way to give the same contrast as burningDog's colors yet maintaining separation by using 6 distinct colors.
|
|
|
OK, that was a long read!
really interesting theories being thrown around. And I agree with most of 'm, but they don't seem to always work in practice (as others have pointed out).
Anywho, I've tried to make my influence map with out the repeating colors (at each players side) and came up with this. It's CMY from the center and then I inverted the CMY colors to fill up the last 3 slots.
+ Show Spoiler +influenceMid1 = 0x12ff73 influenceMid2 = 0xff510f influenceMid3 = 0x000dff influenceMid4 = 0xfff200 influenceMid5 = 0xed008c influenceMid6 = 0x00aef0 influenceMid7 = 0xed008c influenceMid8 = 0xfff200 influenceMid9 = 0x000dff influenceMid10 = 0xff510f influenceMid11 = 0x12ff73
I don't know if it's more clear in conveying all the information we need, but it strikes me (atleast) as less asceticly pleasing. Which might not be the most important thing ever, but it does count!
Oh, by the way, my CMY(K) values come from photoshop. You just input CMYK values and it turns it into RGB and hex for you.
+ Show Spoiler +
Which brings me to the following RGB based openness map. Just like CMY values RGB values have lots of colors in between them, so the gradients should work pretty well. (added yellow as the forth color)
+ Show Spoiler +opennessLow = 0xfff000 opennessMid1 = 0xff0000 opennessMid2 = 0x0000ff opennessHigh = 0x00ff00
At the end, even though I think more subtle schemes like Samro225am's look beter, we might be beter of using super bright colors that burn the eyes. They just seem to bring out small (1 or 2 tile) differences beter.
Edit: just realized that cmy-green might make a good combo. I'll have to try that later
|
Wait--WHAT? That's crazy!!! No blending? 29 variables?? KU-RAHZY...
Give me a little bit...
|
|
|
Just to note, I was using burningDog's variables and I felt like I was not getting the same results and colors in my images. But then I clicked on the image and looked at the original image and noticed it was actually the same. The shrinking and jpeg compression was making the colors look different.
|
|
|
Okay, okay, okay, MORE FUN STUFF!!! Download 1.4.6 (unofficial) and you'll see a new bunch of colors AND constants for influence.
As it is, there are 3 influence colors to define, and I'm using burningDog's:
influence1 = 0xfff200
influence2 = 0x00aeef
influence3 = 0xec008c
and in the constants.txt you'll see I have these set up:
int influenceHeatMapRepeat = 3
int influenceHeatMapBlendMode = 1
This means, repeat the color scheme 3 times from one base to the center, and the analyzer automatically mirrors the scheme on the other side of the map. Blend mode = 1 looks like what we've been doing already. The above numbers give you:
+ Show Spoiler [repeat=3, blendMode=1] +
So let's have some fun. Change repeat to 9:
+ Show Spoiler [repeat=9, blendMode=1] +
WHAT?!!! That's too many bands of color, really. So let's relax it a bit and do 6 repeats. And while we're at it, switch blendMode to 2, which means "no blend."
+ Show Spoiler [repeat=6, blendMode=2] +
KUH-RAZY.
See what ya'll can do with that.
|
|
|
How about calculating influence without rocks?
|
On October 26 2010 15:20 iGrok wrote:
How about calculating influence without rocks?
The analyzer can do it, it just isn't configurable yet. I added a ticket on the project page at SC2Mapster for your suggestion.
|
|
|
imo there should be two maps generated by the analyzer:
1. many repeats
2. few repeats (because for the general appearance and understanding of an image it dows not help to have the same color in three different spaces signifing different things)
@burningDog and Y-R-B-G:
this is really tricky. a combination of CMYK + RGB but really messed up:
it looks like you took the color spectrum:
(magenta) Red (yellow) Green (cyan) Blue
( i know cyan and blue could also be at tghe left end, but we all know the spectrum is not linear but as a color works like a ring)
and then you made it less blending by switching yellow away from blue and green and green away from yellow. very clever!
I think it works well, is very accurate but I hate to look at it. it hurts my eyes and the strong contrast seperate the space - though not as strong as one of barrins earlier models (grey, bue, green).
conclusion: now use less bold colors and introduce some shades and you will have something that looks good soon 
|
|
|
@Barrin: I'm at work, so I can't fool around in the analyzer right now, but about the CMY + lighter/darker CMY scheme: You lose a lot of contrast because the "new" colors are (for instance) 33, or 50% black or white. That really takes the egde of it I'm afraid. Below I made a quick image of CMY lines with 4 altered blocks in the middle: 33% black, 50% black, 33% white & 50% white. And the contrast just isn't as strong as pure CMY.
+ Show Spoiler +
But this did make me think of something else: every tiles consists of 64 pixels, 20 of which are (more or less) black (the crosses). That black contrasts heavily with the "internal color" of the tile, so it probably reduces the relative contrast of the tiles color and that of the one next to it.
__ __
I did this quick little experiment in photoshop. In all 3 pictures the tiles have the same colors. But if you remove the black crosses that seperate the tiles you can see the slight differences in color beter (I think). But you of course lose the height information, that's where the 3rth picture comes in: thats the black/white dot in the middle "system" I suggested earlier. Because it's only 4 pixels that indicate height in every 64 pixel tile, it doesn't mess with the tile to tile contrast as much.
Atleast, thats what I think
|
okay guys, I have all these suggestions noted. Let me see what I can do.
|
On October 26 2010 23:16 burningDog wrote: __ __ I did this quick little experiment in photoshop. In all 3 pictures the tiles have the same colors. But if you remove the black crosses that seperate the tiles you can see the slight differences in color beter (I think). But you of course lose the height information, that's where the 3rth picture comes in: thats the black/white dot in the middle "system" I suggested earlier. Because it's only 4 pixels that indicate height in every 64 pixel tile, it doesn't mess with the tile to tile contrast as much. Atleast, thats what I think
good idea, one problem though: replacing the crosses with points in the centre of each color square brakes the separation and lets the colors 'mix in the eye'. I'd rather go for one black pixel in each corner of every square.
|
On October 27 2010 02:14 Samro225am wrote:Show nested quote +On October 26 2010 23:16 burningDog wrote: __ __ I did this quick little experiment in photoshop. In all 3 pictures the tiles have the same colors. But if you remove the black crosses that seperate the tiles you can see the slight differences in color beter (I think). But you of course lose the height information, that's where the 3rth picture comes in: thats the black/white dot in the middle "system" I suggested earlier. Because it's only 4 pixels that indicate height in every 64 pixel tile, it doesn't mess with the tile to tile contrast as much. Atleast, thats what I think good idea, one problem though: replacing the crosses with points in the centre of each color square brakes the separation and lets the colors 'mix in the eye'. I'd rather go for one black pixel in each corner of every square.
You're right, and it is important to be able to see the map cells when you want to evaluate buildings at chokes and things like that.
|
@dimfish
i sent you a pm, some of my colors seem to be old
if somebody wants to try them out:
my last colors code was
opennessLow = 0x544C0C
opennessMid1 = 0x6CE362
opennessMid2 = 0x1873B5
opennessHigh = 0x9400F0
|
On October 27 2010 02:42 Samro225am wrote:@dimfish i sent you a pm, some of my colors seem to be old if somebody wants to try them out: my last colors code was opennessLow = 0x544C0C opennessMid1 = 0x6CE362 opennessMid2 = 0x1873B5 opennessHigh = 0x9400F0
my mistake, will fix that for next mini-release
|
|
|
okay i tried some things using only three colors. again there are color schemes with more contrast, but I wanted to achive a look and feel (that does not hurt ) similar to my openness colors.
here we go:
+ Show Spoiler +
|
|
|
Wow! Those unofficial releases are awesome! :D
But, is it possible that this last version have some different colors for different bases? Something like this:
+ Show Spoiler +![[image loading]](http://img27.imageshack.us/img27/8874/scrapstationinfluencemi.png) This is two different images put together, a black and white new release image, and an old version RGB.
|
|
|
On October 28 2010 01:26 Barrin wrote:Show nested quote +Wow! Those unofficial releases are awesome! :D
But, is it possible that this last version have some different colors for different bases? Something like this: yay another person participating in the discussion! ^_^_^ I understand why that's worth trying, but to be honest it lowers contrasting which is the real key here. Actually that might make sense for the non-blend map, but then again the non-blend map helps the more advanced users. The only people this helps IMO is the newer users. Once the new users get good they would want more contrasting anyway. Maybe make a third output to help the newbies? Not sure if it's worth the effort :X While it does contrast less than the most recent color schemes, I'm surprised at how much it actually does contrast. Don't let my tone fool you I'm actually pretty impressed with what you did there. Black is pretty sweet
I agree about contrast, but I was only trying to focus the color difference between the two bases. And using the B&W value over the colored picture was the shortest way to show what I had in mind. Sure I could have used other colors in the old version, sure I could have more iterations in the new version, but it really isn't the point. My point is: putting A and B together to show as much information as possible in a single picture.
|
i think it was already tried and discussed earlier in this thread. the point is that it gives a general overview about how balanced a map is terrainwiese, but solely on a very general level (blue side vs red side. it actually visuallizes what you already see quite well in the numbers of influence percentage at each potential base.
when using more colors and repetitions you can see how the usable space expands from a choke into the open field and how it connects to the opponent's space/movement wave. if you take a look at the various images of Blistering Sands this come clear quickly.
one could decide to use reddisch colors for the one side and blueish for the other, but then one would have to switch colors in the spectrum, e.g. midnight blue - teal - light blue - turqouise etc. so that dark and light or clear and broken colors alternate in the sequence and you get more contrast than color+white or black. but still, it is quite easier and better looking using 3 or 5 distinctive colors.
|
Look, I'm not saying that more constrast is worst.. The opposite, actually. I agree with you that this new version is better for many reasons. My only point is that it could be noob-friendly and pro-friendly. My first image, as I said, was only to say about the mix of images, not how it should actually be done.
What I want is the possibility of changing the iteration colors of
- the middle - to know where the middle of the map is.
- the bases - to see the map territorial balance faster.
+ Show Spoiler +Something like this. The middle is green and one base is red and the other is blue. I know that the dark blue is actually the same iteration as the light red, but it is only to ilustrate an idea. (I've tried doing that with a picture with influenceHeatMapBlendMode = 1 and it becomes too difficult to know what iteration I am looking)
|
On October 28 2010 04:18 k10forgotten wrote:What I want is the possibility of changing the iteration colors of - the middle - to know where the middle of the map is.
- the bases - to see the map territorial balance faster.
Let me see if I get you right: you'd like a separate set of colors for the middle of the map and for either base, like this:
base1color1
base1color2
base1color3
...maybe more
base2color1
base2color2
...more
middleColor1
middleColor2
...more
like that?
|
On October 28 2010 15:03 dimfish wrote:Show nested quote +On October 28 2010 04:18 k10forgotten wrote:What I want is the possibility of changing the iteration colors of - the middle - to know where the middle of the map is.
- the bases - to see the map territorial balance faster.
Let me see if I get you right: you'd like a separate set of colors for the middle of the map and for either base, like this: base1color1 base1color2 base1color3 ...maybe more base2color1 base2color2 ...more middleColor1 middleColor2 ...more like that?
Yep. Although, I think the middle needs only one color, more than that would be confusing.
|
|
|
On October 28 2010 15:58 Barrin wrote:Oh I liked the only red and only blue with black on both a lot better :X Show nested quote +My first image, as I said, was only to say about the mix of images, not how it should actually be done. Yes I guessed as much, and I said what I said assuming that. I really don't think that losing the sweet contrasting of yellow/cyan/magenta on the blend map is worth this TBH.
But understand that doing what I seek won't invalidate what you like. If what I want is done, you'll still have your CMYK contrast. Just put the same colors on both bases. That's it.
|
|
|
On October 29 2010 08:27 Barrin wrote:Show nested quote +But understand that doing what I seek won't invalidate what you like. If what I want is done, you'll still have your CMYK contrast. Just put the same colors on both bases. That's it. I'm confused. I thought that having the two sides different colors was the whole point :X
Right. Let me get this straight...
In the way I was talking this whole time, you can get different colors for different bases... But you would still be capable of doing the CMYK way - what you like - by using the same colors in the bases. My point is the possibility of doing, not the obligation (couldn't find a better word) of doing it.
As I said, I prefer this way because I can see as much information as possible, in the same picture. The overall territorial balance, as well as the instancial territorial influence. That would be noob-friendly because you can see the overall and the middle very fast (since you can set different colors for bases and middle), and it could be pro-friendly because you would still see what you used to see in 1.4.6...
|
|
|
The way it seems, it's like the program begins the iterations in both start points and goes on until it reaches the middle. Since dimfish's not making the program open source, I can only speculate (I'm not saying that he should or not make it open source. Please don't get me wrong). And that's why I made the suggestion - I don't know how hard it is to make it.
Actually, I was thinking it should look more like this:
baseA1 = blue
baseA2 = other blue that doesn't look too much like the first
baseA3 = yet another blue that doesn't look like none of those
middle = green
baseB3 = yet another red that doesn't look like none of those
baseB2 = other red that doesn't look too much like the first
baseB1* = red
*names are just to ilustrate, again - no need to be exactly like this.
For CMYK, you should have like:
baseA1 = cyan
baseA2 = magenta
baseA3 = yellow
middle = orange
baseB3 = yellow
baseB2 = magenta
baseB1 = cyan
In theory (I don't know how he's coding the program), reading an array of variable size from a file (I'm presuming he's coding in Pascal/C (or Object Pascal/C++)) wouldn't be hard (just very annoying, IMO) - the problem is when the person doesn't put all the colors (5 of 203, for example), but again it can be fixed so that the program would exit and/or give a warning (or, if he's using <vector>, no need for stating the number of variables, you could put as many as you want, if you put them in the same way ("influenceX", for example)).
(And I'm not talking about the calculations it'd have to make, nor the complexity of putting this together with the code.)
|
Okay, k10, I read you loud and clear, and yes the map analyzer is open source. The entire source code is in the SVN repository at the SC2Mapster map analyzer project page. And I use a blend of C/C++
Anyway, my day job is flaring up, and I'm in the process of planning my wedding (well, helping) so I kinda have to do maps and map analyzer in bursts these days. I'll do a version that does what you're asking and with some other suggestions as soon as I can. Plus, gotta squeeze in GSL games, too!!
|
Sorry for my ignorance, then. I didn't know it was. I'll check it out later. :D
Best wishes, man!
|
|
|
On November 05 2010 11:33 Barrin wrote:
I think we overloaded him :O
No man, I love doing this but I've got work coming out my ears on one end and I'll be getting married in a few months which eats up my usual spare time. I'll get back on it when I can, believe that
|
dezi
Germany1536 Posts
I like the current colors of Influence. Looks really nifty
|
|
|
|
|
|
|
|
|
|
EPT![[image loading]](http://static.sc2mapster.com/content/images/25/874/newopenness.png)
![[image loading]](http://imgur.com/uRmUq.jpg)
![[image loading]](http://imgur.com/i651V.jpg)
![[image loading]](http://imgur.com/Uh1JF.jpg)
![[image loading]](http://imgur.com/fyIzi.jpg)
![[image loading]](http://imgur.com/iX3sx.jpg)
![[image loading]](http://imgur.com/ojOAy.jpg)
![[image loading]](http://imgur.com/2KKX7.jpg)
![[image loading]](http://imgur.com/oC3sE.jpg)
![[image loading]](http://imgur.com/CmYFn.jpg)
![[image loading]](http://imgur.com/ykklV.gif)
![[image loading]](http://imgur.com/eYBEj.gif)
![[image loading]](http://imgur.com/1saDx.gif)
![[image loading]](http://imgur.com/rV4xe.jpg)
![[image loading]](http://imgur.com/wFB9b.jpg)
![[image loading]](http://imgur.com/0ZEU3.jpg)
![[image loading]](http://imgur.com/0Hcqf.png)
![[image loading]](http://imgur.com/WBrwm.png)
![[image loading]](http://imgur.com/dS4E5.png)
![[image loading]](http://imgur.com/Liuvj.png)
![[image loading]](http://imgur.com/fJUS7.png)
![[image loading]](http://imgur.com/YZWg7.png)
![[image loading]](http://imgur.com/dclPJ.png)
![[image loading]](http://imgur.com/4GDhd.jpg)
![[image loading]](http://imgur.com/9rrjm.jpg)
![[image loading]](http://imgur.com/64XWI.jpg)
![[image loading]](http://imgur.com/XFAbq.jpg)
![[image loading]](http://imgur.com/bjhxR.jpg)
![[image loading]](http://imgur.com/D7BTl.jpg)
![[image loading]](http://imgur.com/PKYdu.gif)
![[image loading]](http://imgur.com/s1ui0.gif)
![[image loading]](http://imgur.com/F9jcZ.gif)
![[image loading]](http://imgur.com/OnyeF.gif)
![[image loading]](http://imgur.com/jF3OB.png)
![[image loading]](http://imgur.com/VRVvF.png)
![[image loading]](http://imgur.com/96zN9.png)
![[image loading]](http://img526.imageshack.us/img526/8874/scrapstationinfluencemi.png)
