Just be warned that this will only produce a ‘valid’ approximation for about 0 < x < 2.
Edit: Since icode^2 / p[/icode] and icode^2[/icode] should both be calculated per light (correct, or am I thinking of this variable wrong?), this optimization matters less than the optimization of the equation [icode]y=s/(x^2+t)[/icode] where [icode]s=(plnp)^2[/icode] and [icode]t=(plnp)^2/p[/icode].
Ps: also not gonna use latex sorry not sorry
I taught myself how to rig last night. Fully functional base rig + IK rig. Now, time to rig our giant robots.
Genius coder I forgot this was only happening once per light :
, i’ll fix that. The divisions cannot be optimised easily as that equation cannot be simplified much further than it already is however by precalculating p and lnp then using them when neccessary it can be made much faster.
I did nothing and I´m proud of it, the only problem is that I always think “I´ll go on working on my project tomorrow” and the next day I think “I´ll go on working on my project tomorrow” and the next day I…
Still proud of not doing anything ;D
Had some fun Saturday while at work, on my break, tweaking settings using a new SF/X tool I wrote. I was working on a bird tweet sound (thinking the sounds were quiet and wouldn’t carry), and the store Manager poked her head into the break room. She said she thought a bird had gotten stuck in the attic or something. Success!
It took a few more days to track down some bugs in the system pertaining to using the Java file-choosing code (used in save/load of xml and export of wav). I found out, for example, that I didn’t handle the case where one “cancels” the action and it was leading to the app hanging. Finally posted in WIP thread today. Fingers crossed, am hoping it is stable now.
tinkered around again with my action system, made it now finally possible to correctly chain them, leading to the first math ‘unit’ test:
@Test
public void mathScriptTest() {
final MathAction m = new MathAction();
final GameContext context = GameContext.getContext();
context.setNumber("a", 9f);
context.setNumber("b", 8f);
m.parseFromScriptString("MATH res = (5 + MATH(MATH(a/2)*b))");
assertTrue(m.getResult() == 41f);
}
As this is an action I could also execute its perform method, the result would then be set to a number with the name res.
It looks a bit odd that the string contains several math elements, but there could be other nested function calls - I currently don’t rely on the operation sign alone…
Will update my projects thread once I actually include this in a new version with new more exciting examples 
I sold all my gun skins in Counter Strike for just over $1000
Turned out to be a profitable investment.
Thumbs up!
Rigged the robot last night. We still have some tweaking to make it perfect, but I’m getting excited about animating it for some really fun shenanigans.
I have only one question:
And how many money you spend on keys to unlock them?
Depends on how long it took , a KWH of power is an extremely cheap amount of power only a few cents , most computers run at less than a KWH often at a third of that so I would assume he would have made atleast $500 profit with power costs included. This of course makes it a better short term investment than bit coin
To day I implemented normal mapping for my multi-textured terrain. Still have a lot of improving to do, but just happy to have something to show for it.
Edit: replaced picture after fixing the normal mapping fragment shader. Now the proper normal maps are applied to the right textures. Before they where all being mixed together.
Wrote a ~working version of Breakout in JavaScript, I think I’m going to be playing around with js a bit
Your collision handling is slightly wrong, for example when coming from the top of the blocks it will delete the block but won’t bounce/rebound off at all it will just fall through.
I know, I know… But, meh, good enough. :
I might be a bit late, but precomputing (p ln p)² in the range needed should speed up the function, then another way of speeding up your function might be linear approximation, in fact after drawing your function fx§=(p ln p)² / (x² + (p ln p)² / p ), we can clearly see that after a set p , the function can be seen as a linear function. Example with f100 from p = 17, f100 looks pretty close to g§=p*1,0331625-7.819.
Finished a single week project, one day late. It parses the Vulkan XML registry and asciidoc documentation and generates the LWJGL 3 Vulkan bindings. It was painful (tons of details to cover, bugs in the registry, bugs in asciidoctor, etc), but I’m hoping it’ll save a lot of time (and my sanity) in the future. The Vulkan javadoc is also now much more complete and consistent.
Fun fact: The source code is 3 Kotlin files that compile to 105 Java classes.
Added bitmatrix fonts!
All of the characters in the sentence (the quick brown fox jumps over the lazy dog) are represented as a 64-bit long value
and drawn to the screen by extracting the bits!
EDIT:
Fixed the X
I have been playing with Box2D vehicles for a few hours and decided it would be neat to overlay some 3D buildings like GTA 1. I am using a 2D camera for the floor objects although there are only Box2D debug lines at the moment and then a Perspective camera for the 3D models.
Need to find the correct way to translate 2D to 3D co ordinates, currently I worked out the differences by hand but must be a neater way to do this.
http://imgur.com/gallery/fmTYV
Is it really necessary to use 3D in that case? I mean, as it is now, a simpler way (imo) would be to just draw 2D quads with edges that go towards the center of the frame to simulate perspective
Always willing to try different approaches, will have a search but any links etc would be great thanks for the advice.
