Fixing bugs edition
Related Places:
>>>/agdg/
>>>/vm/
Old Thread:
Fixing bugs edition
Related Places:
>>>/agdg/
>>>/vm/
Old Thread:
Other urls found in this thread:
twitter.com
twitter.com
twitter.com
easings.net
drive.google.com
play.google.com
youtube.com
youtube.com
youtu.be
docs.unrealengine.com
youtube.com
howtomakeanrpg.com
dinodeck.com
irrlicht3d.org
tsun.itch.io
youtube.com
twitter.com
Threadly reminder to stop watching porn and clean up your room.
k what am i doing wrong here
i've got forward+right movement (on a 2d plane, so up instead of forwrad) and i want it to rotate towards its velocity
rBody.velocity = transform.up * output[0] + transform.right * output[1]; transform.rotation = Quaternion.Euler(0, 0, Vector2.Angle(Vector2.zero, rBody.velocity));
but the Z in the rotation just ends up being 90
in what axes are you moving? X and Y?
yes
Then the only thing I can think of is that Vector2.Angle gets the angle between two vectors, and not two points, so you shouldn't be using Vector2.zero as one of the parameters.
Try with Mathf.atan2 instead
i'm dumb, i was doing that 10 minutes ago and abandoned it because they were rotated wrong, i apparently forgot to rotate them 90 degrees
Vector2 vel = Vector2.up * output[0] + Vector2.right * output[1]; rBody.velocity = vel; transform.rotation = Quaternion.Euler(0, 0, Mathf.Rad2Deg * Mathf.Atan2(vel.y, vel.x));
...
I guess being original with the style requires an artfag too?
it's been over 100 generations and my networks are still retarded
wat do
make the method that decided how well the generation did better
make them fuck harder and make babies faster
they don't reproduce
the goal is to get the most kills per team
sorting them with individual kills first, and team kills second
ideally they should be trying to surround their enemies while they outnumber them, but instead most of them just stick to the wall, while a few others go to their deaths
each unit knows the position, health and team of every unit
no clue if i'm doing it right or not
FUCK BLENDER, FUCK UNITY, FUCK RIGIFY, FUCKITY FUCK FUCKING FUCK
Just needed to vent there, excuse me.
Im not sure, maybe it just requires some drawing or sketching skills. No idea where to start.
please explain to me how do you think neural networks evolve
How? There is no UV overlap or something like that would do this.
start with what you have
make it a little bit better
keep going until you reach perfection
On a related note, anyone know any recording software that'll work on a craptop?
well i mean, after 20 seconds i kill them and spawn new ones, i mean that they don't reproduce like the birds and the bees
are the new ones parameters based on the old ones?
Reposting webm from last thread
If you have a dGPU, Shadowplay and ReLive are both very light on resources.
plus new very slight progress webm
they're sorted by individual/team kills, and the shittier half of them are killed and replaced with copies of the better half
as you can see, there's not much progress going on
gonna leave it till generation 500, then i'll make drastic changes
you will end up with a bunch of clones of a single dude.
you gotta random a bit of the parameters each time, so that they will randomly change their parameters to better configurations each generation
clones of the 20 better units, not the top one
and i do mutate their weights i think, no idea how it works
That ain't look like shit, but yeah, I see why you might feel that way from stock looking assets/ui. If you wanna start somewhere, I'd say start with some crappy doodles to visualize where ya wanna go with this.
thanks
also have to let them randomly get new weights with new connections or delete old ones randomly
after that, just let them train for ages
Anyone have more stuff like this?
How do you guys record video? Fraps? the webm recorder im using shits itself if i try to record audio, so i think i will go straight to videos instead
I think nodevs like me just have three options, at least these are what im thinking about currently
twitter.com
twitter.com
twitter.com
Find someone that isn't working as a professional, possibly a student, follow them for a while to see what they post and what they're interested in, then ask them if they want to collaborate on something, then show them your stuff
You would be amazed how many artsy people wish they could code to make a game but are completely unable to start doing it
Thanks
how would you come to an agreement over selling it somewhere?
for
Split the costs, honestly the selling part is the last of your worries, first you have to see if progress is being made at all, and if you're compatible to work together as a team.
Once it looks like the project is going somewhere you draft up a contract and you both sign it so you're both protected by it.
With my artfag we're not planning to complete anything yet and we're probably just make a few vertical slices and see where that gets us, but that would've been the plan if we were making a full game.
Sounds like a pretty good idea, is that how you met yours? I could imagine though that projects break apart easily once you disagree on something, like all of the joint /agdg/ projects in the past. How do you manage conflicts? Just one guy that makes every decision, a set in stone art direction and vision from the start?
Actually I met mine by complete chance, he started following me on twitter because I posted about design and game stuff, since he also seemed like a cool dude I followed him back and the vast majority of his posts were under those hashtags.
Ended up sending him a message, we wrote down ideas for like 10 hours and by the end we had a concept and some art to go with it.
Turns out the dude lives like 2 hours away from my city and he's a massive anti-sjw too, what are the chances
We haven't really had any conflicts mostly because we seem to want the exact same stuff so far. When I proposed a character and a gimmick he liked he went to design the character and came back to me with the exact thing I was thinking of. Like, 1:1 what was in my brain, it was some spooky shit.
But the usual deal is that he's in complete charge of the character design and I'm in charge of the game design, if we want to do something weird or change stuff we inform each other and discuss what could be the best option.
I had really good luck finding someone I clicked with, so don't expect that to happen on the first try, but you still might as well try.
Fucking hell.
Welp I found the cause, there was 2 meshes there for some reason which caused this mess, I managed to fix it.
Communication is key to collaboration. Provide and take feedback on each others' work, even if you're making something completely separate.
When there's a conflict, you both have to make a decision to either be an asshole and insist, or adapt/compromise. Comprimise is the best, because chances are, you only agree with your opinion because its yours. Daily reminder that Star Wars was saved by its editing, and the Metal Gear Solid trilogy was saved by its co-writer. Both of these IPs turned into dogshit when their directors were given too much creative control, and god knows how many potentially good works were ruined by making this mistake the first time.
If it's a political conflict, you should either break it off clean before you get too involved, or agree to put aside your differences and keep it apolitical. Again, communication is key, and discussing the differences can keep you from thinking of them like an idiot (even if they are)
t. programmer and internet friend of artfag who is almost complete opposite of me
Toaster just fried itself
A Turkish man crawled out
Rip
I returned to the earlier project for some (re)design work, just to make sure I'll never forget
I finally got some vacation time so I'm taking things slow and chill.
But perhaps I'm putting too much autism into this? it's fun tho
That wasn't my image
Yeah pretty much this.
A year ago I was supposed to start working with a few people I knew ages ago on a game, I was the only one with any programming experience, the only artfag was an interface designer.
These two cunts were convinced they had to make a 2D mobile game to start with because "3D is too hard" and "all the money is in mobile now", no amount of explaining that without a good idea 2D was boring and overdone and that it's impossible to sell on the mobile market changed their ideas. At some point I just said fuck it and left them to do their thing, and what do you know they've gone nowhere since then.
It's half communication and half finding someone who isn't retarded
lel
Sounds like the dream, well done user
top kek i copied that stagger animation frame by frame from DaS3, but it still looks like epileptic limbo dancing, what do
I think in DaS, the stagger movement starts fast and slows down (EaseOut curve):
easings.net
Also, the enemies in DaS don't attack as fast so it doesn't reset so quickly. Quickly looped animations usually look pretty goofy.
Have you looked into either Evolutionary Algorithms, or Particle Swarm Optimization? There's a lot of cool shit out there regarding Computational Intelligence other than just Artificial Neural Networks.
i've hardly looked at anything
right now i'm trying to centralize things
only 4 nets, but each controls 10 units
if anyone has ideas on what else to do, here's the project
someone said i should let it make its own weights, but i have no idea how it even does them right now
drive.google.com
Timing's a good point. I need more than one stagger animation, the same thing repeated is always bad, same goes for sounds and so on. Then there's also the impact source direction, the animation should reflect that, so 4 different ones at least, maybe 8. I think i should just go for a simple flinch that is added on top of the current pose, but i can't for the life of mine figure out how to do that in UEs anim graph.
If you can communicate well you can determine if someone is a retard before entering the partnership
See you'd think that, but that's not true at all.
One of the two people I talked about in that post I had worked with for maybe 5 years as a duo of webdesigner/graphic designer and we did some amazing shit together
What is it you want out of the NN? What is the purpose?
Haven't devved in almost a month.
Anyways I found a 250 mb .zip on my desktop called "9192-pixar-130-library.zip", some kind of shit to do with art textures, maybe you guys might enjoy it if you google it and needed it, okay bye
I apologize from the bottom of my heart if I'm breaking any rules here, but….p-please play my game. Here are some promo codes to get it free.
If those don't work, I'll post more.
play store link: play.google.com
Trailer: youtube.com
Cheers, and may God smile upon you all.
>>>/christian/ is over there, only the godless lurk here
Its fine to post your game on this thread. It's normal for people who have been working on their game in here to post their game on the actual board if it ever gets done, but you haven't posted any progress on this game here so that doesn't seem to be you. There are no "rules" about it, that I really know about, you'll just get called out for being a shill, and your thread will probably get deleted.
Forgot my name
Forgot where I came from
I died getting here
Make me gold, I'm waiting.
Speak for yourself, heathen
FWIW, here's a video of what I'm working on right now.
youtube.com
I'm just debugging the networking software right now. Not particularly exciting or pretty at the moment.
that looks like LSD
Oops, bad link.
Controls are kind of difficult. How about a virtual joystick that rotates the ball in screenspace, and dragging anywhere else rotates the camera?
I miss old /r9k/.
What I really want to do is use input from the gyroscope so it's more like VR. Unfortunately my potato doesn't have one.
Also, a sphere has the same surface normals regardless of its orientation, so the reflection angle would stay the same.
I mean rotate the point where the "bullet" comes out.
O i c. I will think about that. It could also be fun to just destroy stuff with a laser.
Anybody know what I would have to do to port my Blender model's skeleton for Unreal (or Unity at least)? I just realized that the way it's setup (for animation) is not compatible with any realtime inverse kinematics functions since there are lots of helper bones that are only used for animation. I've seen some videos but they are kind of old and was wandering if there were more up to date solutions. Also because I want to keep my animations.
There's a tutorial somewhere in the official Unity channel about what unity recognizes as a human skeleton but I cannot for the life of me find it right now
I think it was in the docs too, it had a knight as an example
I think I got it, is this what you need?
test
all the replies in the thread disappeared
you can literally just drag and drop into unity, unreal however is a bitch and needs to be fed .xml files hope you version of blender and unreal is up to date. and if you want to use the default rig in unreal or want to reuse animations between skeletons. you "can" via retargeting
docs.unrealengine.com
I'm going to make an MMO
Why must something as simple as a rope be hard to emulate. I just want something similar to how ropes worked in the Source engine for my game in unreal. But it's one thing after another with the stupid physics constraint parameters. It's apparently super easy to use for everything BUT when changing stuff in blueprints. I just want a dirt simple rope for a dirt simple grapple hook.
some gameplay of some mediocre project I've been procrastinating with.
Don't you hate the feeling that something that you've been "working" on for a year or so is effectively just a couple of hours' worth of work?
Thanks the problem was more Blender based since I just had to change the parent hierarchy with the same skeleton. I kept all the same bones. Just parented it differently. On that note do you guys have good tutorials for UE4 to not make bones so fuckign twitchy? It could have been me that picked a bad tutorial.
Switch Target with Component 1 or 2?
Heard your waifu gives killer footjobs.
I always think that after I've done something once, the next time I'll be able to do it much faster and/or better. Or maybe I don't have to change it at all and can just re-use it for future projects.
Tried that and no dice. I'm thinking that maybe I'm doing it backwards, and that the constraint component ought to be handled from the projectile's side, given that the origin of the constraint seems to be where the Player was when the constraint was activated. So if I instead have the projectile handle things on its side instead of the Player, it ought to act as the center of the constraint's radius (at least in theory). But I need to go to bed, I have work tomorrow.
I know there's tutorials on jewtube, but damn near all of them are either showing off what someone made without an explanation, or an advert for something you have to purchase and that is meant for far more complicated uses than what I need; though I did find one that went over how the constraint works in engine, but he didn't go over how to affect it properly with blueprints.
Everyone is bad at stuff they do for the first time. So building the exact same project would probably be prety easy for you right now.
what workflow do you use to get the character into unreal with ik somewhat working?
Well I'm using Blender and I just made a 2 bone IK for each leg. The position in which the foot is moved to is always the trace directly down from foot sole impact position. What I want to fiddle with is maybe the alpha.
If trace has no hit, alpha is 0 (in air so dont use 2 bone IK)
If tracer hit and player is moving fast, alpha is 1-proportional to distance above foot (so run animation doesn't drag feet).
If tracer hit and player is moving slow, alpha is some constant like 0.65. I'll post the nodes in the morning. I double checked my animations and locations of the knees and it all checks out fine so it just has to be the alpha.
it's not a matter of being inexperienced or not knowing how to do shit, it's just that I have these extreme bursts of productivity spread out over several months, and then nearly dead radio silence. I at least used to be able to give myself the excuse that I was improving on anatomy/animation, but I've just been a lazy shit 'til now. Hell, this is actually a remake of an embarrassingly worse project
I made you a shitty diagram that might help. This is how I did rope physics in an old project I've since abandoned.
Attempt number two for the terrain
Will probably redo it again in blender
doing some work on topology for proper deformations (works for low to mid poly).
here's proper elbow deformation topology. It properly deforms for the crease (front), and the bump (back).
On the side picture, the forearm is the right side, while the upper arm is the left side.
I want to make a game. I know it will turn out really terrible, but i know how to make one. Should i do it? Should i not?
For proper butt crease/upper thigh to hip crease it's easiest + most anatomically correct to just do many edge loops (if you don't mind a few hundred extra polys, personally, i do not).
It's also possible to something like attached pics (hips.00X), but I find it less visually pleasing/less anatomically correct (especially for the gluteus muscle groups); though if you want lower poly it's one of the best options.
For knees it's best to do a "knee pad" like design for the bump (circle quad loop around knee cap), and do the same as the elbow "front" picture (but I do multiple of the middle quads that collapse, with only two triangles in total, around said collapsible middle quads) for the back of the knee deformations.
For the shoulder I found that it works to just follow the muscle contours (areas that retain their shape, while other loops deform around them).
A good guide is here:
youtube.com
For the fingers just use the same as the elbow. Attached is a picture of additional approaches + what happens with bad topology.
How did you go about creating that level, just start off and go along whatever comes to your mind in the editor, or did you plan ahead with a sketch of the architecture?
...
I'm curious what /agdg/ thinks of this:
howtomakeanrpg.com
Stumbled upon it a few days ago and have been eye-fucking it ever since. Really tempted to buy, the only thing holding me back is I'm not familiar with Lua and don't know how useful it is, still I'm betting I can take the concepts and translate them in to C# for my purposes. Any thoughts?
...
I would say you can probably get your hands on all that information and comparable assets for free if you do a little research, but if it has good ratings, things like that can be well worth it if only for the sunk cost fallacy that will keep you motivated because you spent money on it.
It is the preferred scripting language for C++ engines, and it sees a lot of use in the industry, it is useful to learn.
I've already made about 4 different finished projects and I've been dabbling with code for quite a while. I'm just curious about the fine points of specifically JRPGs (I've mostly been making very arcady games more about progressing levels than exploration and story). I can probably work most of it out, but I also want an efficient work flow so I'm not driving myself crazy with unneeded complications.
If Lua's good then I'll probably check it out just to learn Lua. I've been wanting to pick up a second language lately before I get too deep in to C# and get stubborn with how I see things.
This.
that thing will not give more insight than actually playing two RPGS while thinking about those subjects
Just get gmod and fuck about with lua, if all you want is to learn it. It's $10 and it's been around for ages. Then get wiremod for it, and you'll have everything you need and more.
Thanks user, I'll look into that option. The only problem with that is if I want to attach the grapple to a physics object, but I'll worry about that if/when my game gets to that point.
a % add force based off of comparative weight? so 50/50 gets evenly pulled and 25/75 for weight differences of 1 to 3?
That could work. The grapple is being used in tandem with a recoil gun I made that launches the player opposite where they aim based on how long they held the charge for. I got the recoil rifle done and working exactly how I wanted, now I just need to get the grapple to work decent enough. Afterward I'll see how to make it so both can work together to pull physics objects (I would guess a comparison of momentum would be the way to go given it equals mass times velocity).
This is all made easier by the fact I changed player movement from kinematic Character Movement to plain old physics-based movement just so I wouldn't have to tear my hair out getting shit to work the way I want. Right now I have it so the player is literally a camera mounted on a sphere, where movement keys alter its torque and mouse alters the camera's rotation (it works well enough).
>Room clean enough (will clean up on Friday because I'm too busy right now)
>Have about two weeks to finish two books then I'll read the Gulag Archipelago, which I just ordered
Progress on my game has come to a halt, because I have an exam next Tuesday and I'm spending most of my days learning. I'm doing work for my game in an indirect sense though.
I don't like politics, but understanding some of it is crucial to the writing of my game, so currently I'm reading political stuff. Finished The Prince some time ago and am in the process of reading Rules For Radicals, which I thought would be its opposite, but that isn't the right way to put it. Both Machiavelli and Alinsky are political realists and those two books are quite similar. It's just that Machiavelli wrote about how rulers are supposed to act in order to keep power, whereas Alinsky wrote about how the people are supposed to act in order to take it. Interesting stuff.
Almost done with Speebot. The only thing remaining is a new UI for the level editor and testing.
fucking hell forgot to embeb
So did that one game where you'd play a white haired chick in sci-fi setting get canceled because the dev blew a hissy fit over Nier or what?
Also what about that one game which seemed to become a low-poly/textured dungeon-crawler where you run around all medieval wit a kawaii uguu anime-grill in (at that time) white pantsu&bra onry?
Nofap is a fucking stupid meme dreamt up by a pseudoscientific faggot who is probably also a Muslim. If your reward circuits are so fried you can't jerk off without losing all self control, then be my fucking guest, but you can take your special snowflake mental illness and shove it way up your asshole if you think I give a shit about your pathetic struggle with your defective brain, or more hilarious yet, that you think everyone is as weak minded as you and should join in on your crazy little practice of self denial.
tl:dr go fuck a goat with your fellow muslims on >>>/nofap/
...
Post your start of project/current state videos.
In retrospective, I haven't really come a long way., but that just means I haven't been working hard enough.
You can also fuck around with Love2D a bit if you want to learn some Lua.
It took me only half a year to come up with an idea that maybe I should inspect how standard assets character controllers are made instead of endlessly fucking around with rigidbody without success.
Looking good. I was thinking about Speebot recently, wondering when you'll post some new info. How long before the release?
I haven't been posting much because I've been mostly working on bonus content, which is supposed to be a secret reward for the players that 100% the game. Release is probably in September, unless too many microbugs pop up. There will be a new demo before then.
Man this looks so good. The opening animation reminded me of Kula World. It's really impressive to think back to how you started posting progress here and how far you've come. In a way it shows me how much further I could have progressed, if I were consistent rather than burning myself out and than taking extended breaks.
Breddy good. I don't read much, but ought to read Gulag Archipelago since I see that recommended a lot.
I was actually wondering if there were any decent books written about the American civil war, though. The topic interests me and is sorta related to my dev as well.
I didn't mention nofap at all, just pornography. And it's funny you keep bringing up muslims as if it's a counter point, when they often have absolutely no self-control.
By any means, keep justifying your addiction.
finally I figured how to make fast moving enemies. turned out to be lot simpler than I thought.
now I finally can add jackals in my game
you can get that for free here dinodeck.com
getting into an engine that you have just one tutorial for also doesnt seem like a good idea, even if you can learn something from it.
id still recommend you do it if you want to do it and the alternative is doing nothing gamedev related.
i also think nofap is nonsense, but not jerking off to porn really feels better after a while.
i hope messing around with lewd models of vidya characters in blender all day doesnt count
...
hmm, apparently that introduced a new bug. oh well. at least it's nothing game breaking.
still have no idea what i'm doing
added a save/load function for the current best net
changed the shape of the units
units can only attack whatever's infront of them
they can't strafe, only move forward/backward and rotate
You're trying to create basic AI from natural selection, right? How are you mutating the behaviour of the units each generation?
nets.Sort(); for (int i = 0; i < generalSize / 2; i++) { nets[i] = new NeuralNetwork(nets[i + (generalSize / 2)]); nets[i].Mutate(); nets[i + (generalSize / 2)] = new NeuralNetwork(nets[i + (generalSize / 2)]); //too lazy to write a reset neuron matrix values method....so just going to make a deepcopy lol }
only half of the new nets get mutaded
public void Mutate() { for (int i = 0; i < weights.Length; i++) { for (int j = 0; j < weights[i].Length; j++) { for (int k = 0; k < weights[i][j].Length; k++) { float weight = weights[i][j][k]; //mutate weight value float randomNumber = UnityEngine.Random.Range(0f, 100f); if (randomNumber
I wanna bully my codemonkey
Hey, you were the one who got offended by the original post. If you're not addicted, good on you.
This looks neat. Is it meant for an RTS game or something like that?
i just wanted to see if it would end up something like vid related
I don't think it will ever end up like that because in that scenario there are a lot of factors at play that you're probably not simulating.
And probably a million other things i can't think of. Your video isn't the optimal way of abstract blocks fighting each other, it's army battalions.
The weights are what control the likelihood of a specific action occurring, I'm guessing? This may be too simple to get any meaningful results, especially if the actions are simply "turn, move forward, etc."
Think of how a typical AI state machine works: there is a priority stack of available actions that execute based on given if statements. You could provide a pool of if statements and a pool of actions, and randomly match them together, and then randomly sort their priorities on the stack.
The if statements could include stuff like: enemy within X dist, num allies in X dist is > Y, dist to nearest ally > X, etc.
Simply put to have any meaningful behaviour you'll need your units to be aware of their surroundings, not just have them execute actions blind and hope for the best.
When I use Spatial as my root node my character falls through the floor, but when I use the Area node the collision works and I don't know why.
Should I just keep using Area as my root node? Will I run into any problems down the road if I do it this way?
Huh, weird. It might be because Spatial doesn't create a physics state for its children or something. iirc I use WorldEnvironment for the root node of a scene.
If I had to guess I'd bet it's due to the way collisions are reported at runtime. If your rigidbody isn't reporting collisions with the staticbody, try to increase the "contact reported" property in the inspector. I think collisions work when Area is the root node because Area takes care of the contact reporting itself.
Collision works in WorldEnvironment thanks for the tip.
I changed contact reported to 100 and still nothing.
Weird, I recreated your scene and I have working collisions with Spatial as the root node. Something must be different, but I'm not sure what it would be.
I had general ideas for what I wanted it to be. Things like the bridge extending from the building to a vertically offset tower separating the grass into a courtyard and the simple red gothic style roof were ideas I wanted but they weren't substantial until I booted up blender.
It helped that I have an idea of the location of this area in reference to the greater world it inhabits
Using a different computer so different ID
By the way, fair warning for Godot 3D: in v2 it's not finished. Apparently when v3 stable drops it will have a finished 3D toolset but right now certain things just don't work because they're not implemented. Most notably collisions between two meshes you've imported (ie: not engine generated primitives) simply don't work. A primitive will collide with a custom mesh, but two custom meshes won't ever collide with each other.
Any particular reason for using an unsupervised learning model, as opposed to a supervised one? Also, what kind of neural network model is it (back-propagating, recursive, feedforward, etc.)? It seems you are using a genetic algorithm for assigning weights, have you considered using a Particle Swarm Optimization algorithm instead to see how it fares?
God fucking dammit, I feel like I got rused. Why is it up in this menu? What is this symbol supposed to represent anyway? Why isn't it in the settings tab?
Wow, after doing shit in OpenGL and SDL I can say I really respect engine devs now. But, this is probably the most coding I've ever done.
What did you do ?
current inputs for the net are:
for every unit on the map - positionX, positionY, health, team, rotationZ
outputs are velocity and rotation
so technically they know everything
i am gonna try to add/remove weights though, maybe that will change things
i think you're under the impression that i've actually studied what i'm making and have any idea how it works
Nofap isn't nonsense. When you combine it with the proper ingredients, it's a doorway to a world you've always thought was possible, but never dare speak about for fear of embarassment.
I always feel like I'm out of my depth and don't know what I'm doing and that my implementation is fucking shit and I there's probably a better way to do it that would be easier I'm just too stupid to figure it out.
Does this feeling ever go away and are there any guidelines by which I could judge my implementation of something to see if it's good or bad?
Is this your first game? Then don't worry about it. Try to make a finished, playable game.
I figured out what I was doing wrong for my rope constraint mechanic, but I've fucked with my blueprints so much in trying to figure that out that I don't know how I had things before that sorta worked; now I'm stuck with a complicated mess that doesn't work at all, I've got the variable I needed but I don't have the system to stick it in anymore. Fuck, back to square one.
I barely understand how they work, but the concept interests me greatly. Are you just muckin about with neural nets for the sake of it, or do you have an end goal in mind? I ask because NN's are generally used for either pattern recognition or prediction, but your apparent application looks like neither, so I was just curious.
both, goal is to see if they'll form anything close to a formation
I got my simple grappling hook to finally work, but I lose all momentum mid-swing as soon as I am directly below the grappled point. I'll probably have to add a force that'll act as momentum.
Look up genetic algorithms and how to make them. It'd probably work better for what you want, but you'd need to understand bitwise manipulation, and you're gunna need to experiment with how to properly encode your genomes. That pdf I shared earlier goes over that topic pretty well, too.
i'm too lazy to read it all
anyways, nets can now add or remove a neuron during mutations
so they start with 1 neuron, chance for new neuron is 15%, chance to lose a neuron is 10%
they gain fitness for moving/rotating/attacking/killing
now my question is how do i handle a variable number of inputs - i don't want every unit to know exactly where every other unit is, that would kill performance if i had more nets
i only want them to know about the units in a certain range from them, but how do i handle that? i can add new neurons for input, but the weight change is a lot slower than the new units in range/out of range
Yeah, that requires a good degree of pain tolerance. Especially if you look at what modern engines are capable of:
And then these are multiplied:
It's the primary reason for me using engines rather than making my own. I want to make a game and it would take ages for me to build an engine that would end up worse than Unity/Unreal anyway. I'm a dude, not an army of programmers.
It wasn't even that for me. I stopped watching porn for purely pragmatic reasons.
So I just dropped it.
By which metric? Performance? Fun?
idk
I guess I'm just looking for some arbitrary guidelines to follow.
What I'm trying to do now is a game like mount and blade but with text battles.
picrelated is the start of how a battle would play out. The AI would also make the same decisions as the player with their army. So it would play out in many ways. The player charges cavalry? Then the AI does, or doesn't. If he does they meet and there's a morale check to see if one of them swerves aside and the other is then able to continue on and get to the skirmishers, or if they both hold then they meet in the middle and fight. Or maybe the player doesn't but the AI does and so on. Units would have upkeep, stats like strength and defence and morale, and you can get companions like in M&B and can assign them as commanders to different types of units and provide buffs based on their own stats.
There's an object obj_combat_menu which creates four instances of obj_button in a for loop. It assigns each button a variable called button and increments it each time. So the first button is 1, then 2 etc. Then when you click a button it returns that variable to obj_combat_menu which has a switch in the step event and will make certain things happen depending on what button is. To display what should be displayed in the draw event everything is written down and contained in if statements which checks a variable called draw, which is changed in the switch for button. So button == 4, then go to switch case 4, and draw is changed to 5 and maybe this or that script is called or whatever, then break.
Gamemaker can only return one thing from a script so I've put all the relevant things in a array global.stats
global.stats[0] s the total of men in the players army, then 1 to 6 is how many of each unit. So [1] is skirmishers, [2] is archers and so forth.
7 to 13 is the same for the army for whatever enemy the player is fighting.
14 to 27 is stats for the player himself such as strength, and also the same thing for the enemy commander.
Then after that are stats to keep track of how many men are killed during the battle to be displayed to the player.
It just seems like a clusterfuck since I'm going to have like 40 or 50 or more things that could happen so the switch is going to go that high too, then this array which contains all the stats for the player and the enemy army all together.
Seriously, you sound like a fucking scientologist. Go back to your cult.
So no drawings?
Use a version control system like git for that kind of shit, i always tear up my project as well when things don't work, but if you can look at the diff or just revert, it's all fine.
I think engine devving is a good exercise, but only viable for just like making game if your game is really simple and if you know from the start what exactly you need from the engine.
That architecture anxiety feeling goes away after a while, it's just experience. Don't bog yourself down with tutorials of other beginners trying to adopt their shitty practices, just think about what would make sense and iterate. Fundamental best practices like a component-entity system are also worth looking at.
Thinking on making my first game, a small little game to get the hang off it, i have tons of programming-knowledge from before but it's been a while i used it.
What game-engine should i use? Gamemaker, Unity or Unreal?
Python -> Godot
C# -> Unity
C++ -> Unreal
AIDS -> Gamemaker
do you already know, what kind of game you want to make after learning gamedev? if not just do what said.
There's also Atomic Game Engine which is apparently a fork of Urho3D with a scene editor, so it's pretty similar to Godot in terms of being easy for newfriends. You can use C++, C#, and JavaScript evidently. I've been interested in checking it out for a while now but haven't gotten the chance.
There wasn't any sketches of the architecture. If that is what you are asking. I'll use this as an excuse to talk about how I'm making levels in general.
Here's the general workflow I'm trying to follow for world design.
In this case this area is supposed to be fairly wide and open with lots of room for movement. This plays off of one of the parts of gameplay relating to combat. This is first to ensure that gameplay comes first.
How the 3d space I want to create can fit into the world I want. In this case the area is not a castle because I think castles are cool. Instead the area is a castle because castles can allow for large open spaces in the form of courtyards and that fits with the overall theme I'm going for.
The actual layout of the area. This comes after thematic design because I still want it to feel like a castle rather than a game with the skin of a castle (ala doom levels that look like levels and not like real areas). I would say its normally a sin to put thematic design after level design but I think I'll get away with it because conceptualization of the gameplay is first. I made a low poly basic mesh to map out ideas. Pic related is an early export, couldnt find the original model. Also pic related is an old sketch for a level very similar to this for when this project was isometric pixelshit. mp4 related
Realistically gameplay should be tested in the previous stage to make sure your idea isnt shit, I skipped that process so that I could see what it would look like when polished. This is where you do research on shit that looks vaguely like what you want in real life and piece it together. I'm using the same colors as in the pixelshit project which I stole from an user on /agdg/ making some 2d city builder about a year ago before he got cancer from trying to animate cars in pixels or something like that and abandoning it.
Of course this is a general workflow and the process should be much more messy, jumping to and from different stages, to make sure it works at all parts.
Thank you, sounds like a very solid workflow. Also your pixelshit project looks pretty polished, what made you change your mind? How does the game work now?
ooga booga, fucking unity
Moar !
Welp I am almost done with my ore mine now, I just need to texture it and get it animated later, I really procrastinated a lot with that one though I am lately not so motivated anymore doing any modding and modelling stuff. Also is there a way to fix the damn Belt somehow? It really has a ugly warping effect going in the middle.
Could you just make the belt flat and bake AO into the texture?
Checked it out, seems like a solid engine and a decent alternative to unity.
Once I'm done torturing myself with pic related I think I might try it.
Something like this?
finally got the shitty net viewer to work
not that i'm gonna get any interesting results, last attempt reached generation 800 and they were still more or less moving at random
i simplified it so now a net knows only the positions/rotations/health of its own units, and the same for the closest enemy, and how many units are in the enemy's sight, since i couldn't figure out any other way to handle variable inputs
I'm really into making a kind of a third person stealth/action game with lots of levels and freedom, but that's the end game kinda. I have to start somewhere first if i'm a total newbie and diving headfirst into making a real game would be stupid.
im not telling you to make your dream game first, but picking and learning an engine, that lets you do what you want later is probably better than having to learn a new one, when you want to make the game you are aiming for.
With that in mind godot is probably not a good idea since 3d is still lacking there.
I might be biased, because i picked it myself, but ue4 seems like a good idea, since it has a lot of stuff already set up for typical fps/tps games. Downsides compared to the other realistic alternative unity are the worse asset store, 5% revenue share and the smaller community, so there is less chance that someone already asked your question and got an answer.
Nice thanks for the answer!
UE4 is owned by the asian jews of Tencent, who are directly responsible for the worsening condition of the vidya industry. Someone can post the infograph, I don't have it.
There are plenty of FOSS 3D game engines out there that there's really no excuse to use something owned by kikes.
fug we need an agdg wiki for these questions, this is the nth time I've typed this out
Follow this user's advice
I.e. utilize an engine that supports your preferred language.
Beyond that it should be noted that each engine has their own initial advantages, and each of the engine's developers had a different design goal.
It was designed for the unreal series, and has had expanded capabilities utilizing that template; while having artists + quick prototyping in mind (blueprints for C++ and shaders).
So, it's made by people who actually played their games +
people who used their engine to make games.
It was designed to be a generic engine, and was made for you to build up from that.
Unity has a large community, really good documentation, tutorials galore, and it's a lot easier to quickly get into the "nitty gritty" of the engine compared to UE (i.e. a lot slower/less documented).
So, things like hand writing shaders with the automated version or manually going through each shader stages like vert/geom/tess/hull/frag shaders, C# has jit compilation so can recompile while you're playing your game, etc etc.
When it comes to stuff like EULAs then it's up to you, and what you'd prefer.
cmon man how many games were made in those engines? I would like to use foss, but there is just no way they even come close. The only hope is maybe godot, but you you will probably be waiting for godot 3.0 for a while.
And in this case at least their interest directly align with mine. The better the tools they provide the more money they can make.
dissadvantages:
A number of popular indie games were made in Monogame, including Apptheon, Bastion, Transistor, Salt & Sanctuary, and Stardew Valley. Do not assume just because an engine is FOSS that it's "just not as good" as a proprietary engine.
Moreover, I can't even imagine what "exclusive features" AAA engines have that a one-man-army developer could possibly need.
Was able to salvage my brother's laptop. Hoping this can help me learn to code because then I can be on a couch or other comfy thing instead of my desk. But once I got it I noticed some horrifying details:
Check the thermal paste and clean the fan, should probably be enough to fix that.
Fair warning, if you plan on setting up guhnoo/eunuchs you're probably going to have to deal with UEFI bullshit, which was a pain in the dick for me when I configured the bootloader.
This seems more pragmatic than Urho3D. They don't seem to be going full voxel cones and untested tech.
Ops.
im not saying its worse because its foss. if they were just slightly worse i would accept that, just for having complete freedom and not having to pay anything.
as for exclusive features: how about 3d that looks like it belongs in this decade.
PBR isn't exactly memetech, it's pretty standard triple-A shit these days don't talk shit about voxel cones
I'm just saying, they got a solid base down before going for futuretech.
Yeah, to be honest I have no idea why they've made such a high-quality renderer when basic 3D classes remain unfinished. I won't even be using PBR in my Godot 3D project so that's particularly frustrating.
I think in a not so far away future, engines will become like distros, made of all these FOSS middlewares. Even perhaps the gameplay scripting itself.
I want to use Godot but holy shit dynamically typed scripted languages are fucking cancer
should I just wait for Godot 3.0 so I can use a real language?
If that'll help you, sure. That's what I'm doing. I'm just using Unity to learn and refine skills in C# then I'm going to move to Godot once it gets a stable build with C# because fuck giving Unity money. Then I might move onto Unreal after making some small-scale Godot projects because some of Unreal's tech would be a great fit for games I want to do in the future
Sexy loli wife feet and some v1 eye and neck rotations.
Man this is some pixel-ass pixelshit.
I like it
what are you making?
Do you see me trying to "convert" people to this nonexistent religion? No, just corecting the record.
I don't know yet.
I'm between projects, trying to figure out what I want to make next.
The RPG Maker community has a whole ton of fantastic artists drawing tilesets and sprites, I got one and put it in a different engine to see how it looked in first-person.
It was tiny, so I made them 2x. It was still too tiny, so I made them 4x. Now I look like a little tiny kid in a pixel world.
/agdg/ help
WELL THEN MAYBE YOU SHOULD HAVE FUCKING SAID SO IN THE DOCS
The documentation for Godot is total garbage. I can't remember what I was looking at in the docs, but I remember at the end of one of the pages it literally says to contact some dude on irc for help. The API isn't much better, I'd estimate that only 1/4th of the functions are documented.
honestly i just want a low poly first-person game made in this style now
Oh god what the fuck. I have a decent understanding of the 3D toolset for v2 but I suppose the current docs are only for v3, so I probably couldn't do much to contribute to the docs. They had some event like a year ago where they asked the community to focus on improving the docs but obviously this didn't result in the docs being even close to complete.
Moar?
pls develop someone who'll love me and wont be lying
Machines cannot love, so any professment of such will be a lie.
user, if it was possible with current technology I would have done it years ago.
The neurochemical events which cause love are understood. But what are feelings? They're feedback provided by neurotransmitters, of course, but why is it experienced that way? Could it be reproduced in a machine?
Now were you looking at the "latest" version docs or the "stable" version docs? Sometimes if you google something and go to the docs, it will take you to the latest documentation rather than the stable.
Not that the stable docs are too much better, see light baking.
This is my progress in about the past month, on the Sigma II engine. I Hope you guys find some of the stuff I explain useful.
machines cannot lie either so check mate
wouldn't /CURRENT/ technology just be a £500 priced bit of cardboard with "love is a social construct" written on it by female sweatshop technicians?
...
it's a feminine dick so how could it be gay???
my friend from church watches traps and he is not gay, he is so not gay that he
Is anybody else annoyed here at the different naming conventions used in C++ libraries ? Sometimes you get FunctionsNamedLikeThis() and other times it's likeThis() and yet most if not all of C++ standard library is like_that(). It makes my code look messy.
That's nodev talk
Shut up you're clearly not intelligent enough for perfectly flawless languages such as C++
i'm glad godot is all snake case, fuckin' superior
I spent 8 hours fixing eyeball following rotation in Unreal. Whatever you do, don't nest your modify bone rotations unless you want to run around plugging in values until rotations work. I ended up having to hard bake vertex positions for the left eyeball for this to work. I was modifying head rotation and then on top of that modifying eyeball rotations. And even though it was world space rotation space, adding or subtracting angles fucks up roll/pitch values. Patience is bitter but the after taste is sweet.
Waifu AI is nowhere near done but this is one of her behaviors where she just follows you around. Still need to add facial expressions and some other fun stuff.
UE4 forces you to use Title case C++ variable names because it converts them to actual phrases for you in editor when you compile.
The old project wasnt much of a game and more of an experiment. I never thought I'd finish it in that state, I just got carried away with polishing it. The game as it is now looks like its going to be a goofy action rpg with focus on an open interconnected world and lots of types of movement. I'm looking forward to trying to implement pole vaulting.
never ever
Get cracking! Where is the mouth with which to smile lovingly at you for creating her?
Do you have a head tilt function planned?
I see what you did there.
Was it your intention to make it seem as if the environment is more realistic than the girl? Now, to me it sort of looks like a attempt to indicate that the player is part of real world while girl is anime character in a real world.
It could be pretty funny if you went down this route. I can imagine a scene where a deranged son that has a habit of bringing his body pillow to a family dinner table, one day brings home an actual anime girl that was depicted on the pillow.
How did you do that? I have tried using a VInterp like in pic related, but it results in a jitter. And does your version work with pathfinding?
You can still use whatever naming convention you want, it also converts theRedBanana or the_red_banana into a phrase.
When you turn around your head at a great angle, you kind of automatically tilt it, so there's
more than just memes as an implementation reason.
You don't deserve any help if you call yourself lazy faggot
wow, i guess babies should just give up on learning how to walk or talk, i mean if they don't know how, why bother
Your waifu is shit.
Or is it?
Hey, I have a problem with some tutorial I'm trying to follow here :
irrlicht3d.org
//Bullet globalstatic btDiscreteDynamicsWorld* World;static core::list Objects;//Bullet initbtBroadphaseInterface* BroadPhase = new btAxisSweep3(btVector3(-1000, -1000, -1000), btVector3(1000, 1000, 1000));btDefaultCollisionConfiguration* CollisionConfiguration = new btDefaultCollisionConfiguration();btCollisionDispatcher* Dispatcher = new btCollisionDispatcher(CollisionConfiguration);btSequentialImpulseConstraintSolver* Solver = new btSequentialImpulseConstraintSolver();World = new btDiscreteDynamicsWorld(Dispatcher, BroadPhase, Solver, CollisionConfiguration);
The compiler says the World variable does not name a type yet it has one and when I declare it like that :
btDiscreteDynamicsWorld* World = new btDiscreteDynamicsWorld(Dispatcher, BroadPhase, Solver, CollisionConfiguration);
It works properly, what's wrong ? I have the feeling I shouldn't even give a shit about all the static variables and declare them normaly.
made some more learning changes
now the inputs are the positions of your teammates relative to your position, not the global one. Personally i have no idea what my world coordinates are, so this makes more sense
also attacking also depends on the output now, otherwise i get braindead nets that get fitness simply for being spawned next to someone
and i got the net visualizer to be positioned correctly
if there is no one linking to all outputs you can already discard the network and breed a new one, especially the shoot output
technically all the weights are there, they're just very close to 0 so the alpha channel of the line color is also low
since it might not be exactly 0, i haven't bothered to do what you said
and units can do friendly fire, so shoot output could be 0 or lower when facing a friend, doesn't matter
how do i get rotation of a transorm, relative to my own transform rotation? in jewnity
There are functions for that afaik. Something like Transform.Transform/InverseTransform. I don't quite remember the names.
there's transform.InverseDirection, but that's for vectors, not quaternions
Think about it logically. What is the difference between your two images? You've removed the transform of You so that You is at Zero Transform, and also applied that to the Target. So create a variable of the target's transform and negate Your transform from it. Then query the rotation from that variable.
what, just do quaternion2-quaternion1?
Yup.
I made a meme
this doesn't exactly look like a videogame, user
...
Did you just assume my project's gender?
My project identifies as a videogame, from now on I want you to refer to it with the pronouns game/games/gameself
But it can be used as a seamless texture creator :^)
Thanks for the answer, I will look into it right after I am done texturing my model.
A static global variable in c++ means, "only initialize this variable in this particular .cpp file". The static keyword probably isn't necessary in your case. Make sure you are including the Bullet3D include file which defines the variable types you are trying to use.
This is way of weeding out autists. notice how C++ projects actually do useful things that people want? It's not a mistake.
And constants are CAPS_SNAKE_CASE. What's the problem?
A few new animations and a new bug that makes no sense.
nice
Amazing.
Everyday is a busy day for me, even when I got absolutely nothing to do.
That's true I'll consider it.
Yesi it works with pathfinding. I overrode a UPathFollowingComponent and wrong some stuff in C++. This is the part of a MoveTo task that is in charge of actually applying velocity variables. I just did what you see that is highlighted and made the facing rotation always follow the velocity.
Sounds and Music are on, preparing the Friday Report now
Best vampire
Can you bully her?
The last kick looks kinda bad. The leg moves far too slow for the impact it has on the enemy and the entire movement looks wrong. I'll try to articulate what's wrong with the movement, but it's kinda difficult to put it in words.
When you start such a rotating kick, you don't just twist while raising your leg (try doing that). Your Jesus faces the enemy with his left foot at the front. The first thing he does is twisting his feet inward and rotating his upper body 180Ëš, so that he now faces the opposite direction. The weight distribution between his legs is about 90/10 with the majority of his body resting on the left leg, which is slightly bent. You need to keep this in mind while drawing, because it means that his center of mass needs to be above his left leg. If you want to get an idea of what that looks like you can stand up, bend your left knee while keeping your right one straight and shifting it forward so that the majority of your weight rests on your left leg.
The next thing is that his upper body keeps going while his legs stay on the ground. He twist it so that he looks over his right shoulder back to the enemy. This builds up a lot of tension, which is where the power of those kicks comes from.
Then he releases this tension like a loaded spring. He rotates, not 360Ëš but 180Ëš, and rather than lifting his leg up and keeping it there (which is really fucking hard), he just throws it up once. It's the angular momentum that produces the arc, which means that his leg only reaches an apex for a moment (preferably in your opponent's face) before it falls down again. It also means that his kicking leg leaves the ground quite a bit later than it does in your current animation. See my shitty top-down drawing for the arc. What I'm trying to show is that his leg being high in the air while facing away from him (like in your current animation) doesn't happen.
So yeah:
It's kind of a shit explanation, but the best I can do.
Wanted to revisit that animation anyway, so thank you very much for your detailed post. I'll hopefully make good use of your input tomorrow.
That looks very promising anonkun, good luck.
Would look better if the fucking popups and tooltips were recorded too, i can't make them show up in the video
oh boy
Good, means you already slept for tomorrow too, so you can dev for a day and half straight :^)
Years ago before I put on a lot of weight I did Capoeira and tried various other martial arts for fun and because I didn't have to pay extra. Although the transitions (going from fighting stance to initiating the kick back to the fighting stance) varied greatly, the basic premise was the same:
I embedded a video where you can see it nicely (although their form leaves a bit to be desired). Disregard all the Capoeira specific parts and just focus on the kick itself and its buildup.
get on my level and wake up at 2p.m.
What'd you make that in? also, good choices in placeholder music
Shaping up to look good. Are you still using that monolithic file of dish definitions?
Yes, but i will make a flexible and moddable script-based system when i "import" the system to Unity
Every other form of content from ingredients to faerie traits is based on jsons and is easily moddable
heheh…
Are you even trying?
bretty cool.
song?
Got some blinking up.
Katamari Damacy - Lonely Rolling Star. I uploaded the song just change it to .mp3
what are the best softwares for creating 2d art like logos ? photoshop still is best or am i missing anything ?
Thanks m8.
...
photoshop really works for anything 2d art related
What you're looking for is probably "vector art", and is generally used for logos.
After I actually tried to use the code I ran into some issues with unset variables, because I'm retarded and didn't call some initialization functions.
My flaming heap of shit I call my codebase is growing.
Is this even a question? Fucking gimp, 100% gimp
Have you thought of uploading your model of shinobu?
It's absolutely fucking perfect and adorable and I want to use it to make pictures of her posing
...
Please reconsider
Why won't you autists just die already
I also thought it was too pointy. This better?
ew fuck no why would I do that to something so pure?
I want to rig her in cute poses with backgrounds and set up my discord bot I wrote to post those pictures
The fact you think I would do something so disgusting hurts me.
Looks better
Likely so.
The texture warping(?) at the tip made it extra noticeable.
have you tried Krita yet?
I like it. I just do texturing work for my UV maps though. Not an artist. Wish I was.
Krita is great and probably inkscape or illustrator
...
Gimp is legitimately better than photoshop now though, even if you neglect the plugins.
What's he going to do with that Egg?
Once you get used to it, it just werks. Krita or inkscape might be better though.
GIMP will never even get close to photoshop.
It's been around for ages and has barely moved forward. Using gimp not only makes your work worse, but it makes you less employable as NOBODY uses GIMP in a professional manner no matter what field they happen to work in.
Gimp is trash.
It might be on-par
That's delusional
So basically you're making an appeal to popularity?
Wow this looks really nice, I'm a dummy and don't like carnival aesthetics but this gives me the same feeling as an old N64 game about a robot that rode around on one wheel as well, Camera looks good mate, keep it up. Would love to see prototype gameplay, your player movement speed seems just right for my motion-oriented brain.
The UI is a bit obtuse but it's perfectly fine for editing.
Imagemagick is the superior tool anyways
what?? Seriously? You believe this? Gimp is low-rent trash. The devs are openly dismissive of their users. It's a painful and obtuse piece of software. Stop wasting your time.
Eat it.
Eggs are delicious.
I love seeing the work you're doing, it motivates me to keep animating my own stuff. I only have images right now, put off in-engine work 'til I have a walking/running/jumping/standing animation group to work with. Here is what I have so far.
You should have used actual arguments.
gimp and photoshop are image manipulation centered, for logos you want something vector centered like inksape or coreldraw
Mother fucker, didn't upload my images.
Hell yes.
Even if it's just placeholder music, Unreal is patrician taste.
It's placeholder until I can get around to figuring out if I can take tracker music apart and use assets for my own music. This is all for an Unreal Gold mod. I want to show what can be done if someone takes the time to review an older engine and make a total conversion of it.
Can't post images and embed at the same time
GIMP can do anything Photoshop can do but better and for free!
D O N ' T L E T T H E D R E A M D I E ~~!
Looks nice. Thanks for the praise too, it motivates me a lot.
It's probably a good idea to post to some development blog outside of this website too, right? I'm terrible at marketing.
You're going to want to post on twitter more than anything.
People have a very short attention span and will not want to read a blog.
I can't be the only one this happens to right? Is this normal?
Her mouth is too red and too small, possibly with too much lip.
Oh dont worry about the lip and mouth color part that's just because that was in Blender not the in engine shaders. I did make the mouth 1.3x bigger though. I can't make it as big as that because anime logic. Her mouth size is inconsistent and cartoony. I'm also using my figurine of her as reference. If anything I would have to make a special separate shape key for when she wants to make a mouth of that proportion because If I just made the normal "mouthOpen" key that size, it would not look very well if I were to use it for small sizes. Also I started toying with outline shaders and added fangs.
On top of what posted, there's also the abysmal speed of development. I needed high bit-depth support four or so years ago. Gimp 2.10 is supposed to deliver that and a whole lot more. It's 2017 now and it's still not here. Compare that to Blender. Blender I can totally see becoming as good as Maya after they rewrite it from the ground up in Blender 3.x.
1000 generations and they're still more or less retarded
i think i made a mistake when i chose to run 4 units per net, should have went with 1 either have a couple of nets on the same team, or duplicate all the neurons and their weights when i want to add another unit per net
and adding fitness just for moving/rotating was a bad idea in the long run, they never stop spinning like retards. all they've learned so far is that attacking enemies is a good thing
...
How long does it take to git gut at Bullet Physics ? I'd like to be able to do robot simulations someday.
The new kick animation is mostly done, first implementation test looks like I'm missing a frame or two in the middle of the arc though. The charged version (= second version) could probably stand to be a bit slower and more impactful as well.
wouldn't it be easier to just use 3d models, animate them and then render/draw over them?
Not for me. I don't know the first thing about 3D modelling, posing, rigging or animating, and I just like traditional animation a lot. Besides, 3D model rotoscoping will give you a somewhat mechanical, lifeless look if you're not careful about it.
gamemaker
Looks a lot better.
How do you resist the temptation to add support for literally everything, and instead just like make game?
I tweaked the jump physics in my engine, but now I want to add support for Mario 64-like jump combos, and make it possible to give the player new jumps like when you equip accessories in Terraria.
You don't resist. Add that shit and see if it's fun.
Because I've never made a proper game before in my life, and I mostly blame it on this kind of stuff. I keep adding features to my engine and making it as flexible as possible just in case I need it at some point, instead of just making the game.
i think i found your problem. If you want to make game making your own engine is probably a bad idea
How come? I'd have already made a game if I had the self control not to make the engine perfect and generalized for all purposes.
Consider having it decrease fitness a bit per death, but make fitness increase more per kill. Honestly, though, a genetic algorithm seems more aligned with what you want, but ultimately it all comes down to how you "encode the genome".
Just make a one use engine, for your game and your game only.
Make two lists, one with the features that you know you need for your game, and another with features you might need for your game. Follow through on the first list. Don't change the
need list until playtesting reveals that a change is needed. Feel free to add and remove from the might need list as much as you want, use it as a mental notebook.
Enginefagging is fine as long as you can stick it out. Go ahead and keep doing what you're doing, it sounds great.
good advice I think. I started doing this (in a text file). I've made so much real progress in finishing the game instead of exploring cool little pieces of rendering tech and shaders all day.
It turns out there were many things I wasn't that good at but I needed to get done - ask for help or just do it. I put them off to do things I was more comfortable with.
10 minutes of planning a day was all it really took.
Personally I just build the base out first then build up from there. Although I never get to the point where I start adding gameplay mechanics.
eh, if it's just for prototyping and temporary sprites, no, not at all. Takes more time than just making sketches
exactly because of what you just wrote in this post you dingleberry
you're wasting your time on shit you don't need
Can energy as currency work in real life?
I want to implement into the world I'm building but I don't think it might make sense.
That is not something that should be in an engine, in fact not even "jump physics" is something an engine should include. That's all game-specific implementations. Engines just include:
The only reason you should include jump combo code is if you're making a game with jump combos right now. The only reason to enginefag is to have a framework tailored exactly to what you like to work with so you make games faster. Not to make something someone might one day find useful.
as a currency no, we're constantly getting energy from somewhere and it can be at different rates, so the price of energy is completely variable
gold is a finite resource, which is why it could replace the current currency
yes, and no
the first "currency" the Sumerians made silver coins of equal size that where not backed by the silver that they where made out of but rather a bushel of grain. this allowed them to pay soldiers and collect taxes from villages that did not have silver mines or any other means of collecting silver without trade ((((merchants))) getting a cut b4 taxes) it also allowed villages that produced a surplus to store wealth effectively and other villages that had a bad year not to starve but to eat their savings. the basis for the coinage having value was the practical value of the grain not the silver. now you see we really have yet to stray to far from money being backed not by gold or silver but rather the continence of the market. and to limit the continence of the market to 1 good would be to have every other good on the marketplace to be dependent on that good. (cough, OIL, cough, cough) as it used to be human labor the means to feed ones self and ones slaves was the means to produce everything, now we have oil, and if you want "energy" witch has always been the backing factor of currency. i ask what format? what you get out of your outlet? or a tank of gas or a meal? are you energy coin guy? the one that made the light up coin? yes if electricity was easily fungible and transferable to the point where something could be made, with less energy than it stores and that amount was worth trading. yes you could have electricity backed currency.
new idea, stamina is money.
sure, just look at the petrodollar. Money is only useful as long as everyone believes that it is backed by something.
I have a float value per tick, a game controller analog stick, and i want to translate that into menu input. So
change selection one to the left
But not
change selection a gorillion to the left because of update per tick
What's the smartest way of doing this?
fun fact: energy as in electricity has highly fluctuating value, which is sometimes negative, when the power grid is close to an overload. You can earn money by offloading electricity in storage, this is called "control energy".
Just build a cooldown. Maybe change the length of that cooldown depending on how far you tilt.
That way you can't tilt rapidly though, right?
If you want to let the player tilt-reset-tilt and always register two inputs for that, just clear the cooldown when the input goes to neutral.
var x = 0if stick.xaxis < 0.1: x-+= deltatimeif stick.xaxis > 0.1: x += deltatimeset_selection(round(x))
This will increment x by 1 every second the control stick is held in a direction.
Yeah that was my solution as well, thing is the tilt float isn't updated per tick, it's a function that only gets called if the stick is tilted. So that would involve some fucked up caching solution that im too retarded to think up
Almost all of the games I'm interested of making would have jumping of some description. I'm not going to rewrite it from scratch every time if I can just make it once and then tweak it for any particular game. The idea is to have commonly needed functions that I can just pick and choose from and tweak for any game I want to make, and jumping is part of my sidescroller controller.
Don't you have a function that you can call and that returns the tilt float? Just call that every frame and handle things accordingly.
If you want the speed of the navigation to be relative to the amount the stick is tilted:
if abs(stick.xaxis) > DEAD_ZONE: return deltatime * stick.xaxis * SPEED_MULT
This eliminates the need for the if statement, assuming stick.xaxis goes from -1 to 1.
This
I've researched a bit and it doesn't look like i have one. The events are handled asynchronously
What game engine are you using?
UE4
In your project settings you add an axis mapping Forward, ranging from 1 to -1 based on axis mapping. You probably already have this.
In your c++ class' initialisation code, you put BindAxis(TEXT("Forward"));
Then in Tick() you can use >GetAxisValue(TEXT("Forward")); to get those values.
Fuck. The amount of shit to do is overwhelming and some of the stuff I can't do. Along with relearning shit I've forgotten.
The only positive is that I've started writing my ideas down. Writing down ideas is really hard, completely different from just thinking of them.
Will there be a gondola skin?
Thanks user!
If I use makehuman would I be considered lazy?
It does negate the entire point of me having to learn how to make a human myself.
The results are going to be pretty fucking bad
Meh fug it then, I guess I will just use animated textures for that instead, I looked for tutorials how to get a conveyor belt animated and its not really a trivial setup (video length = 30 minutes) which is for me personally too long and wouldn't really make that much of a difference when my sprite is going to be scaled at 150x150. And all that because the fucking curve/array modifier has not a damn option to disable screwing up with the mesh, well done blender developers
what kind of shitty tutorial are you following?
if your object is already following the curve
I was looking at this one: youtube.com
i can't say it is bat, but isn't very good either, if you already started watching just bear with it, but know that there are faster methods
Welp I got it working now for real this time, thanks again for the help. I got the belt animation working without the generator as I applied the animation to the flat belt directly, even though I should have used skeletons for that but eh its not a big deal right now. I have to admit I was confused from your first belt post as I didn't really get it
great, now that you know the magic of duplicators everything will become duplicators.
need to fill a whole city with lamp posts? duplicators
want some trash in the ground? duplicators
a whole fucking building can be made with about four or five objects, then those objects mixed to make a few variations of the same building, is amazing the amount of bullshit you can get away with using that same trick
by the way, don't change your creative process to fit into duplicators, instead create as much parts as you need then replace the redundancies and save the unused parts for your new projects
Gamemaker.
with (obj_water_v1) { var s; // Create a var that will be the y value of the object var i = 0; // i is the variable used to move DOWN the grid var x_num = 17; for (s = y; x_num < 48; i += 1) { ds_grid_set(global.world, x + x_num, s + i, 1); if (i == 63) { i = 0; x_num += 1; } } tile_add(bck_water, 1, 1, 64, 64, obj_water_v1.x, obj_water_v1.y, 20); instance_destroy(); }
I have a grid, global.world, which is created using the room width and height. I have some objects which are meant to be water and the movement in the game works by checking the grid. So wherever there is water set the grid to 1, then replace it with a tile, then destroy the object.
Problem is it all works except for the tiles. It only adds the tile in for one object, but still deletes them all.
How do I make this work?
that is because you are adding that tile outside of the loop
yeah i noticed that i knew it all along i was just testing you
nvm
it still doesnt work
just creates the tile for one of them and not the others but the grid shit still works for all
aren't you creating all your tiles in the same space?
what part of this defines position?
This defines position.
I did this and it's working a bit better but not much.
global.world = ds_grid_create((room_width), (room_height));var i;// VERTICAL WATERfor (i = 0; i < instance_number(obj_water_v1); i += 1) { with (instance_find(obj_water_v1, i)) { var s; // Create a var that will be the y value of the object var i = 0; // i is the variable used to move DOWN the grid var x_num = 17; for (s = y; x_num < 48; i += 1) { ds_grid_set(global.world, x + x_num, s + i, 1); if (i == 63) { i = 0; x_num += 1; } } tile_add(bck_water, 1, 1, 64, 64, obj_water_v1.x, obj_water_v1.y, 20); } instance_destroy(instance_find(obj_water_v1, i)) }
It does what it is supposed to do but only for the first and last instances of this object. The other ones it just deletes and for some of them it might change the grid to a 1 but mostly not.
If anybody is interested I updated my ARWM mod to 0.1.10 for Factorio 0.15.26
Check here: >>>/vm/1769
new thread, but the idiot op forgot to link it here