Spring RTS Engine

bumping

someone shoudl sticky

why?

why not unsticky all the tutorials we have stickied now?

this one is nicheier i guess but i still get people asking me shit

we need a master section or at least someone to FIX THE WIKI so images can be posted to it and restore the old images so the old tutorials work.

i immediately run four sequential gaussian blurs of high power

You may already know this, but the mathematics of the gaussian blur dictate that multiple applications of the blur filter yield exactly the same results as a single application of the filter with a higher blur radius. Of course you'd have to either experiment or know how to calculate the bigger radius to get the same results.

Just thought I'd mention that since I thought it was odd that you were running several blurs in a row. Is there a maximum blur radius that was constraining you?

some of those tips actually hurt you if you are doing pro with tiled builds.

Many things like blurs etc should use a preprocess pipeline and then have the final pipeline using the exported altered heightfield. For example, items with ! have special considerations when using a tiled build which is what you would use when doing real textural building.

There are many many many things I could write about worldmachine and it's usage as I have been drilling the fuck out of it's potential

http://smoth.net/foram/download/file.ph ... &mode=view

Thing is worldmachine is very good but it also does not have much suggestions on how to do any sort of optimized pipelines and the forum is a bit sparse as far as discussion.

some things like flippers, blurs and erosion should be handled prudently. Also a good bit of what makes wm2 impressive is the textural capabilities that have not really been explored. I have a great deal of experiments going but if you guys want try and get good heightfields out of world machine and not actually do any texture generation you can pretty much ignore me :\

RogerN wrote:

i immediately run four sequential gaussian blurs of high power

You may already know this, but the mathematics of the gaussian blur dictate that multiple applications of the blur filter yield exactly the same results as a single application of the filter with a higher blur radius. Of course you'd have to either experiment or know how to calculate the bigger radius to get the same results.

Just thought I'd mention that since I thought it was odd that you were running several blurs in a row. Is there a maximum blur radius that was constraining you?

No...IIRC I was having trouble with large radius blur killing the tops of my hills, and low radius were too steep so I just did a bunch of sequential low radius. I don't think i was able to just pick out a radius and have it work.


Anyway, I really don't think it's not mathematically the same....

Here's three jpgs - one with no blur, one with 1x gaussian blur radius 16, and the last with 4x gaussian blur radius 4.



Of course different blur algorithms are going to have different results, but just going by logic it seems like the blur's inherent gradient curve will be altered simply due to sequential blurs vs one single big blur. That is unless theres some kinda golden ratio shit going on with gaussian blur curve and it always ends up the same, but I don't know the equation for the ratio of radius : number of applications. Its obviously not radius1 * applications1 = radius2 * applications2

smoth wrote:some of those tips actually hurt you if you are doing pro with tiled builds.

Many things like blurs etc should use a preprocess pipeline and then have the final pipeline using the exported altered heightfield. For example, items with ! have special considerations when using a tiled build which is what you would use when doing real textural building.

There are many many many things I could write about worldmachine and it's usage as I have been drilling the fuck out of it's potential

http://smoth.net/foram/download/file.ph ... &mode=view

Thing is worldmachine is very good but it also does not have much suggestions on how to do any sort of optimized pipelines and the forum is a bit sparse as far as discussion.

some things like flippers, blurs and erosion should be handled prudently. Also a good bit of what makes wm2 impressive is the textural capabilities that have not really been explored. I have a great deal of experiments going but if you guys want try and get good heightfields out of world machine and not actually do any texture generation you can pretty much ignore me :\

By all means add to the tut/modify it/whatever. I only have experience with WM1 so this tut is primarily for it. I should also add the part of how to export and compile with 16 bit goodness, since its a lot of hoops to jump through, but the reward of doing so is immeasurable. Actually it is measurable, by a factor of 256 ;p

WM is IMO the best heightmap software available so it deserves springtut lovin.

definitely a great heightfield tool

I still don't see why you need a tool for heightmaps. I'm more interested in the texturing half of WM.

because wm2 allows for great heightfield control. The texturing is handy and my work in the tool has been fruitful but there are some issues that I have not been able to get around, trying to get some attention from stephen or nikita on it.

neddiedrow wrote:I still don't see why you need a tool for heightmaps.

smoth wrote: http://smoth.net/foram/download/file.ph ... &mode=view

you need a tool to make a heightmap be shaped like that

I admit, that is beyond my handiwork, but I'm not interested in realism for anything I don't intend to merely reconstruct from satellite/aerial photography.

http://en.wikipedia.org/wiki/Gaussian_blur:

Applying multiple, successive gaussian blurs to an image has the same effect as applying a single, larger gaussian blur, whose radius is the square root of the sum of the squares of the blur radii that were actually applied. For example, applying successive gaussian blurs with radii of 6 and 8 gives the same results as applying a single gaussian blur of radius 10, since sqrt(6^2 + 8^2) = 10. Because of this relationship, processing time cannot be saved by simulating a gaussian blur with successive, smaller blurs ÔÇö the time required will be at least as great as performing the single large blur.

There you have the relation

So in your case, hunterw, 1 blur application of size 8 would be same as 4 times size 4.
( sqrt(4^2*4) = 8 )

personally i'm not terribly impressed by "hey look i can erode the shit out of my heightmap"

yeap worldmachine has 1 setting, ultramax erosion...cant make any good gameplay maps w/ it, might as w

http://spring.jobjol.nl/show_file.php?id=670

The real power of WM doesnt lie in erosion, imo it lies in layout mode and the whole paradigm shift of procedurality.
I can just define basic layout, and can tweak it to my liking very easily:

Beherith wrote:paradigm shift of procedurality.

this

btw acid water / lava for the win. highly gameplay relevant - define your map's borders instead of another big square to play on.

Tobi wrote: ( sqrt(4^2*4) = 8 )

yeah i wouldnt have figured that equation out on my own

heeh

ralphie wrote:personally i'm not terribly impressed by "hey look i can erode the shit out of my heightmap"

who said that?