Post What's on Your Mind

[ZenVirZan]

SDL doesnt let you use src rects that are negative or out of bounds, so i spent today making a camera, then editing the drawing system to automatically correct out of bounds src -> dest conversions, keeping it correct across any framebuffer size (logic is done at 240x136, framebuffer can be any resolution, and display is 960x544)

EDIT: turns out if the srcrect is larger than the source image in both width and height everything breaks
busy procrastinating fixing the camera cus i cant figure out whats wrong for the life of me, but i also cant be bothered implementing animated autotiles and stuff hmmm

 

[Fayte]

Got my 3 promo cloud and 27 packs.. really wanted to pull a foil Terra, but i didn’t.

 

[Mega Flare]

Got my 3 promo cloud and 27 packs.. really wanted to pull a foil Terra, but i didn’t.

LOL bought three kits? Canada isn't getting pre release. We getting a post release lol

 

[Fayte]

Got my 3 promo cloud and 27 packs.. really wanted to pull a foil Terra, but i didn’t.

LOL bought three kits? Canada isn't getting pre release. We getting a post release lol

Yeah i just bought 3 kits.. got my playset out the way. That sucks you guys got stiffed i thought they fixed that issue of not everyone getting stuff at the same time.

 

[Mega Flare]

Yeah. At least we will still get it

 

[Ellie]

Sometimes I wish functions like that would just default to a size that does work if you put negative in.

But at the same time sometimes those defaults are the source of much confusion and annoyance too, especially when my code shouldn't really be producing negative space in the first place.

 

[Xilef]

If you're going to try to convince us you're a legit user maybe don't necro-post a thread and with a complete, quoted product listing title - Walmart.com's page titles probably aren't the best source of information for what westerners enjoy.

 

[ZenVirZan]

EDIT: turns out if the srcrect is larger than the source image in both width and height everything breaks

haha even though i made literally everything in my graphics calculations a double and cast at the end, my scaling multiplier for src.x -> dest.x & dest.width reductions was somehow an int

zen what the fuck

 

[Xilef]

i made literally everything in my graphics calculations a double and cast at the end

Hold on I'm just going to stab out my eyes.

There's moments I find it difficult to justify using a float for graphics (which is the basic unit for modern GPUs), I've never found a justification for using double.

The only times I use doubles are for compile-time constants or values, in which case I use the double just to get good precision on my floats or fixed-point integers.

 

[Glitchfinder]

i made literally everything in my graphics calculations a double and cast at the end

Hold on I'm just going to stab out my eyes.

There's moments I find it difficult to justify using a float for graphics (which is the basic unit for modern GPUs), I've never found a justification for using double.

The only times I use doubles are for compile-time constants or values, in which case I use the double just to get good precision on my floats or fixed-point integers.

I don't know if it's still true, since Microsoft has redone a lot of the graphics backend post buyout, but Minecraft's rendering system was very poorly conceived, and everything was rendered relative to 0, 0, 0. This had a minimal effect on normal play, but it became noticeable if you did something like use a mod to teleport a few million blocks from center (approaching the old far lands, which no longer exist). You would start seeing issues caused by coordinates not being specific enough, like so:

 

[Xilef]

I don't know if it's still true, since Microsoft has redone a lot of the graphics backend post buyout, but Minecraft's rendering system was very poorly conceived, and everything was rendered relative to 0, 0, 0. This had a minimal effect on normal play, but it became noticeable if you did something like use a mod to teleport a few million blocks from center (approaching the old far lands, which no longer exist). You would start seeing issues caused by coordinates not being specific enough, like so:

This was fixed a while ago. However, you shouldn't just use doubles because you made a crappy rendering system - the Minecraft example is at extreme distances where players are unlikely to be, using doubles to solve an issues that 0.01% of the player-base go out of their way to experience is silly, just wastes performance for the other 99.99%.

 

[Glitchfinder]

I don't know if it's still true, since Microsoft has redone a lot of the graphics backend post buyout, but Minecraft's rendering system was very poorly conceived, and everything was rendered relative to 0, 0, 0. This had a minimal effect on normal play, but it became noticeable if you did something like use a mod to teleport a few million blocks from center (approaching the old far lands, which no longer exist). You would start seeing issues caused by coordinates not being specific enough, like so:

This was fixed a while ago. However, you shouldn't just use doubles because you made a crappy rendering system - the Minecraft example is at extreme distances where players are unlikely to be, using doubles to solve an issues that 0.01% of the player-base go out of their way to experience is silly, just wastes performance for the other 99.99%.

Actually, a lot of Minecraft coordinates were being processed in doubles at the time, which is why I brought it up. I never said it was a good use case, just that it was a use case.

 

[Xilef]

Gawd Minecraft is badly implemented...

 

[Glitchfinder]

Gawd Minecraft is badly implemented...

Having helped implement colored lighting mods multiple times, you have no idea. It went years with half the lighting/rendering code active, making calculations, and then defaulted. So many wasted processing cycles. The source had so much evidence of notch attempting to implement features, not being able to get them to work the way he wanted, and then just defaulting the results of calculations to something that "worked" without actually commenting the calculations.

 

[Fayte]

Bout to buy another 3 kits

 

[Xilef]

Having helped implement colored lighting mods multiple times, you have no idea. It went years with half the lighting/rendering code active, making calculations, and then defaulted. So many wasted processing cycles. The source had so much evidence of notch attempting to implement features, not being able to get them to work the way he wanted, and then just defaulting the results of calculations to something that "worked" without actually commenting the calculations.

I've also gone through the renderer code a few times, I do have an idea and it really stresses me out to think about.

The way transparent surfaces are handled is my trigger.

 

[aphadeon]

Having helped implement colored lighting mods multiple times, you have no idea. It went years with half the lighting/rendering code active, making calculations, and then defaulted. So many wasted processing cycles. The source had so much evidence of notch attempting to implement features, not being able to get them to work the way he wanted, and then just defaulting the results of calculations to something that "worked" without actually commenting the calculations.

I've also gone through the renderer code a few times, I do have an idea and it really stresses me out to think about.

The way transparent surfaces are handled is my trigger.

Out of curiosity, /is/ there a better way to handle transparent surfaces in GL, besides z-sorting and alpha blending? I'd certainly hope so, but I haven't found it yet. Assuming I want to be able to stack an arbitrary number of them.

 

[ZenVirZan]

i made literally everything in my graphics calculations a double and cast at the end

Hold on I'm just going to stab out my eyes.

There's moments I find it difficult to justify using a float for graphics (which is the basic unit for modern GPUs), I've never found a justification for using double.

The only times I use doubles are for compile-time constants or values, in which case I use the double just to get good precision on my floats or fixed-point integers.

fair enough, ive swapped everything over to floats instead.

also do you happen to know how inefficient it is passing functions as arguments via std::function? its only being called a few times per frame (under 5) but im wondering how bad this practice is (this is my first c++ project)

 

[Byron]

This programming assignment is going to kill me. The deadline is midnight tonight. I've still gotta produce a test plan and write a report about if/else statements and case statements and how other languages handle them.

The main programming is done, there is a few small kinks that to be honest aren't necessary for me to iron out. Oh hey. It's 5 am and I have class at 9:30 whelp...

I'm going to be so dead in class...

 

[ZenVirZan]

thats rough man
anything we can help out with?