I wouldn't mind going to GDC but mostly out of hobbyist's interest. Networking for fun at most. If I happen to win (lol) - I will totally drive over for a few days. Great excuse to get the wife to see SF, she has never been to California.
So I'm at the point in my game where I needed to start implementing some audio stuff. Now, my original plan was to take this week off-ish and start building the Markov chains that will be controlling the music logic when I came back to it, but it turns out I've got stuff that lines up way better if I pushed to finish more this week, and then took next week off.
I decided, well, since I have a week and this stuff should only take a week, I'll build the synths in Wwise, and connect them to the towers in game, so that once I do have the Markov chains, it'll be easier to test and implement without needing to spend some week in the future doing the shitty grunt work that is dumb.
As a placeholder, when a tower now picks its note, instead of the game generating music using some sophisticated logic, in a way which will hopefully evoke the styles of 80s and 90s arcade machines..... well now it just picks some notes randomly for everyone.
What's the rule with power-of-two textures in UE? I'm going to be handling a bunch of sprites and, as they are not pixel art, their resolution makes it very impractical to try and fit them onto POT spritesheets. What are the side effects of non-POT (sprite) textures? Is it still more beneficiel to pad textures up to POT, even though it's a lot of wasted space?
I saw some people on forums mentioning POT doesn't matter for sprites, but no one linked to any sources.
Last time I worked with a 3d engine on a low level enough for POT to matter it was because mobile processors' OpenGL couldn't handle NPOT textures, though if not being able to make a mobile build is my only problem then I'm happy with that.
Today Amazon has both announced and released a new, free game engine, Lumberyard, which offers deep integration with its Amazon Web Services server infrastructure to empower online play, and also with Twitch, its video game-focused streaming service.
That's right -- the engine, including its full source code, is completely free to download and use to make PC and console games. Amazon will not charge any kind of royalty or subscription fee.
Lumberyard is powerful and full-featured enough to develop triple-A, current-gen console games (and the company has signed official tools deals with Microsoft and Sony, so you can immediately build games for the Xbox One and PlayStation 4 with it.) Mobile support is coming down the road.
Monetization for Lumberyard will come strictly through the use of Amazon Web Services' cloud computing. If you use the engine for your game, you're permitted to roll your own server tech, but if you're using a third-party provider, it has to be Amazon.
What's the rule with power-of-two textures in UE? I'm going to be handling a bunch of sprites and, as they are not pixel art, their resolution makes it very impractical to try and fit them onto POT spritesheets. What are the side effects of non-POT (sprite) textures? Is it still more beneficiel to pad textures up to POT, even though it's a lot of wasted space?
I saw some people on forums mentioning POT doesn't matter for sprites, but no one linked to any sources.
Last time I worked with a 3d engine on a low level enough for POT to matter it was because mobile processors' OpenGL couldn't handle NPOT textures, though if not being able to make a mobile build is my only problem then I'm happy with that.
The two things you lose are texture compression and mip mapping. Mips are irrelevant if you're not planning on changing the distance from the camera to the sprites (or changing the size of your orthographic camera, if you're using one) during gameplay. Texture compression is a bigger loss, but your sprites shouldn't be memory intensive and DXT butchers anything high-contrast or colorful so maybe it's for the best
I was under the impression that UE4 had a sprite atlasing tool built in tho
I believe it does (I need to look into just how powerful they are, unfortunately UE's own tutorials for Paper2d were made in its beta in 2014...), the problem is more that after a certain size sticking sprites into a texture becomes impractical. When one animation loop ends up broken into multiple atlases you're in for funtown.
However, knowing exactly what's affected by POT/NPOT is very useful, thanks!
Today Amazon has both announced and released a new, free game engine, Lumberyard, which offers deep integration with its Amazon Web Services server infrastructure to empower online play, and also with Twitch, its video game-focused streaming service.
That's right -- the engine, including its full source code, is completely free to download and use to make PC and console games. Amazon will not charge any kind of royalty or subscription fee.
Lumberyard is powerful and full-featured enough to develop triple-A, current-gen console games (and the company has signed official tools deals with Microsoft and Sony, so you can immediately build games for the Xbox One and PlayStation 4 with it.) Mobile support is coming down the road.
Monetization for Lumberyard will come strictly through the use of Amazon Web Services' cloud computing. If you use the engine for your game, you're permitted to roll your own server tech, but if you're using a third-party provider, it has to be Amazon.
Kind of neat. It has potential, the fact that it's completely free for single player and LAN development is solid. The most intriguing thing to me is how AWS and Twitch can be used in the engine. My friend sent me this announcement earlier today and I've been mulling over ideas on how to appropriately use Twitch chat as a gameplay mechanic to positively add to the game experience for the player that isn't just a derivative of "Twitch plays X".
My friend sent me this announcement earlier today and I've been mulling over ideas on how to appropriately use Twitch chat as a gameplay mechanic to positively add to the game experience for the player that isn't just a derivative of "Twitch plays X".
In Party Hard the chat votes on the next event and in openRCT2 it names the guests after viewers.
Looks like I'm due for another overhaul of the fire rate logic for the towers. I realized I probably want the ability to have different rhythms in the music lines, which then means that the towers have to be able to fire at different rates based on the array of the notes they'll be playing.
I really should've been planning ahead more when I literally just redid it all.
I hope there's more to the Twitch chat stuff than "hey we have an IRC library".
IRC and direct accest to the twitch json api would allow for fun stuff in game.
Also popping up everywhere Amazon had some fun with the TOS as well:
57.10 Acceptable Use; Safety-Critical Systems. Your use of the Lumberyard Materials must comply with the AWS Acceptable Use Policy. The Lumberyard Materials are not intended for use with life-critical or safety-critical systems, such as use in operation of medical equipment, automated transportation systems, autonomous vehicles, aircraft or air traffic control, nuclear facilities, manned spacecraft, or military use in connection with live combat. However, this restriction will not apply in the event of the occurrence (certified by the United States Centers for Disease Control or successor body) of a widespread viral infection transmitted via bites or contact with bodily fluids that causes human corpses to reanimate and seek to consume living human flesh, blood, brain or nerve tissue and is likely to result in the fall of organized civilization.
Did a quick look through the API. All it looks to allow right now is in game interaction from twitch chat parsing. So you could have the game interact any number of ways from parsing chat. Should be fun stuff with that. Also there is a small portion of the api that lets you setup direct connect to/join streamers game with commands as well which could be nifty. It says it lets you join the game from a link in twitch chat too. Like steam invites.
I have an old, battered copy of the first one and it is a wonderful book. Both for the history it represents and for some still quite relevant lessons about designing a game.
RiemannLives on
Attacked by tweeeeeeees!
+2
MachwingIt looks like a harmless old computer, doesn't it?Left in this cave to rot ... or to flower!Registered Userregular
From what I've gathered, Lumberyard is just CryEngine with a new face. I.e. tools are probably still abysmal, which might be why they aren't talking about them anywhere.
Finally got avatar loading to work. So you can make some custom avatars and then load one into your game. The worst part IMO was having to dynamically create material instances and bind them to the right meshes. But it works! Since they're Unreal materials you should be able to do weird stuff like make people with metallic skin but I haven't looked into that too much yet.
Phew. Okay, it took almost two evenings of messing about, but I finally have a working pipeline to get stuff from my animation software into UE without messing up frame pivots, getting texture artifacts or alpha fringes. It requires two trips through Photoshop (once to batch-resize all the frames and again to fix the spritesheets' alpha so I don't get bleed), but hey, it works.
Next up, making a better sprite material, the default unlit one does terrible things to edges.
From what I've gathered, Lumberyard is just CryEngine with a new face. I.e. tools are probably still abysmal, which might be why they aren't talking about them anywhere.
Yeah I saw that pop up on Facebook, went "What? Amazon is doing a new engine?! Awesome! Oh...CryEngine. Let me know when Star Citizen finally releases."
Does anyone have any significant experience with CryEngine? I'm curious about its pros and cons; not because I want to use it, it's just been used for some really neat stuff in the past, but barely anything these days.
Fellow Unity devs, anyone have any experience working with top down 2D local avoidance steering? I am having difficulties finding a solution for my game. I have looked at options I can buy and none I've found perfectly fit what I want to achieve. Further other than some art and sound effects that I can't make myself, I really really want to build this entirely myself if possible.
The requirements are that I need to move my ships via physics because it gives the right look and feel. I don't mind if ships bounce around a little if they bump into things but stuff can't be getting stuck all the time. Basically what I need is a system that says "whoa, there is a thing! we should avoid slamming into that, go here to do so" and then once it gets within X dist to that point it hands it back over to my regular nav system (which works great but doesn't try to avoid anything around it). I could use an A* style pathfinding system but i need local avoidance more than pathfinding, since it's space and most of it is empty. The stuff I need ships to avoid will mostly be moving - other ships, asteroids, ect. So that is the most important thing.
I currently sort of have a solution that I might be able to tweak into working how I want. Basically I have it detect a thing to avoid, then two detectors "scan" it by aiming a raycast2d at it then rotating to the right or left until it no-longer hits the object, then it records those points to find the width of the object from the ship's perspective. It next compares the angles between the two points from the ship to figure out which one is the quicker option and overrides the normal nav system to steer the ship towards that point. The point itself is aware of the object the ship wants to avoid and makes sure it's X distance away from the object, to give good clearance away from it's edge. Once the ship gets close to the point it considers the object avoided and hands it back to the regular nav system. It has issues though, like getting confused if there are too many things near it so it will kind of fly around in circles for abit sometimes. It also currently doesn't adjust it's detection window further or closer based on speed though that wouldn't be hard to fix. It's reasonably fast but not instant.
I have another solution in mind that I'm going to build and test that might be way simpler and achieve mostly the same thing, so I will post about that later on.
Additionally one idea I had was to implement the best system I can achieve and set up the collision layers so that the ships themselves cannot actually collide with each other and the environment so that ships don't ever get stuck. So most of the time the illusion would remain intact but if the system should fail stuff won't get stuck.
Steam Profile | My Art | NID: DarkMecha (SW-4787-9571-8977) | PSN: DarkMecha
I dunno why anyone would be excited about a "new game engine" from the folks who brought you "no previous engine" anyway
but it's a free CryEngine! Think of all the third party companies that have made successful games on it--the most recent Sonic Game! Star Citizen!
MechWarrior Online is still trucking along, although it has to be noted that the MWO thread generally considers PGI to be bad at programming ("look at all the other companies who were able to do great things with CryEngine!"). And I often hear good things about State of Decay in the Steam thread. Other than that...
So my other solution shows some promise. It's definitely got some issues to work out but is faster and simpler than the previous one.
Basically I'm pulsing a raycast2d X units ahead, if it sees something it gets the distance to the object then sets the position of a gameobject that I've taken to calling an avoidance probe with a sphere collider in trigger mode x times the size of the ship's own (to give plenty of clearance, currently using 4x) forward of the ship based on the distance value gained from the detection and tells the probe what gameobject to look for. It then rotates the probe around the ship until it's collider is no longer overlapping the target gameobject's collider, at which point it tells the avoidance ai that it's got a solution for it, set's the coordinates of the steering point to where it is then resets back to the ship. The ship steers to the point and once it gets within 5 units hands control back over to the regular nav ai.
It's actually not bad so far. I setup a little obstacle course for it and while it always favors right turns currently, it's not gotten stuck at all! (yet...)
Steam Profile | My Art | NID: DarkMecha (SW-4787-9571-8977) | PSN: DarkMecha
In Unreal's BP I was working with a function and trying to figure out the difference between using the function inputs directly via Get, or by setting them as intermediate local variables (I didn't feel like having connections running back to the input pins every single time).
Turns out, even though you can Get an input from anywhere within it, it won't actually hold the input value for whatever reason, you need to either pull it from the input pin directly or Set a local var and then Get that one. Kind of weird if you're used to coding, as you expect function parameters to be defined within in without having to copy them into a separate variable.
In Unreal's BP I was working with a function and trying to figure out the difference between using the function inputs directly via Get, or by setting them as intermediate local variables (I didn't feel like having connections running back to the input pins every single time).
Turns out, even though you can Get an input from anywhere within it, it won't actually hold the input value for whatever reason, you need to either pull it from the input pin directly or Set a local var and then Get that one. Kind of weird if you're used to coding, as you expect function parameters to be defined within in without having to copy them into a separate variable.
Yes I'd like to not have to stretch long wires within functions too.
Posts
2) Today I implemented UV distortion and flow mapping in a 13 year old game engineeeeeee
Yo @Machwing! You at GDC?
Unreal Engine 4 Developers Community.
I'm working on a cute little video game! Here's a link for you.
Me too! Let's at least high five once
Unreal Engine 4 Developers Community.
I'm working on a cute little video game! Here's a link for you.
So I'm at the point in my game where I needed to start implementing some audio stuff. Now, my original plan was to take this week off-ish and start building the Markov chains that will be controlling the music logic when I came back to it, but it turns out I've got stuff that lines up way better if I pushed to finish more this week, and then took next week off.
I decided, well, since I have a week and this stuff should only take a week, I'll build the synths in Wwise, and connect them to the towers in game, so that once I do have the Markov chains, it'll be easier to test and implement without needing to spend some week in the future doing the shitty grunt work that is dumb.
As a placeholder, when a tower now picks its note, instead of the game generating music using some sophisticated logic, in a way which will hopefully evoke the styles of 80s and 90s arcade machines..... well now it just picks some notes randomly for everyone.
The result is..
uh....
It sure makes sound now!
Took the liberty of grabbing some screens:
Unreal Engine 4 Developers Community.
I'm working on a cute little video game! Here's a link for you.
Unreal Engine 4 Developers Community.
I'm working on a cute little video game! Here's a link for you.
Hell, I've been clamoring for VR support in 3D apps since I first put on my DK2.
Unreal Engine 4 Developers Community.
I'm working on a cute little video game! Here's a link for you.
I saw some people on forums mentioning POT doesn't matter for sprites, but no one linked to any sources.
Last time I worked with a 3d engine on a low level enough for POT to matter it was because mobile processors' OpenGL couldn't handle NPOT textures, though if not being able to make a mobile build is my only problem then I'm happy with that.
https://aws.amazon.com/lumberyard/
http://www.gamasutra.com/view/news/265425/Amazon_launches_new_free_highquality_game_engine_Lumberyard.php
The two things you lose are texture compression and mip mapping. Mips are irrelevant if you're not planning on changing the distance from the camera to the sprites (or changing the size of your orthographic camera, if you're using one) during gameplay. Texture compression is a bigger loss, but your sprites shouldn't be memory intensive and DXT butchers anything high-contrast or colorful so maybe it's for the best
I was under the impression that UE4 had a sprite atlasing tool built in tho
However, knowing exactly what's affected by POT/NPOT is very useful, thanks!
what engine yo
Kind of neat. It has potential, the fact that it's completely free for single player and LAN development is solid. The most intriguing thing to me is how AWS and Twitch can be used in the engine. My friend sent me this announcement earlier today and I've been mulling over ideas on how to appropriately use Twitch chat as a gameplay mechanic to positively add to the game experience for the player that isn't just a derivative of "Twitch plays X".
In Party Hard the chat votes on the next event and in openRCT2 it names the guests after viewers.
Looks like I'm due for another overhaul of the fire rate logic for the towers. I realized I probably want the ability to have different rhythms in the music lines, which then means that the towers have to be able to fire at different rates based on the array of the notes they'll be playing.
I really should've been planning ahead more when I literally just redid it all.
IRC and direct accest to the twitch json api would allow for fun stuff in game.
Also popping up everywhere Amazon had some fun with the TOS as well:
Write Your Own Adventure Programs for your Microcomputer
and
Write Your Own Fantasy Games for your Microcomputer
I have an old, battered copy of the first one and it is a wonderful book. Both for the history it represents and for some still quite relevant lessons about designing a game.
The Heroes of the Storm / Starcraft 2 / Warcraft 3 engine
They'll still be using this thing when I'm dead
Unreal Engine 4 Developers Community.
I'm working on a cute little video game! Here's a link for you.
Next up, making a better sprite material, the default unlit one does terrible things to edges.
Yeah I saw that pop up on Facebook, went "What? Amazon is doing a new engine?! Awesome! Oh...CryEngine. Let me know when Star Citizen finally releases."
but it's a free CryEngine! Think of all the third party companies that have made successful games on it--the most recent Sonic Game! Star Citizen!
The requirements are that I need to move my ships via physics because it gives the right look and feel. I don't mind if ships bounce around a little if they bump into things but stuff can't be getting stuck all the time. Basically what I need is a system that says "whoa, there is a thing! we should avoid slamming into that, go here to do so" and then once it gets within X dist to that point it hands it back over to my regular nav system (which works great but doesn't try to avoid anything around it). I could use an A* style pathfinding system but i need local avoidance more than pathfinding, since it's space and most of it is empty. The stuff I need ships to avoid will mostly be moving - other ships, asteroids, ect. So that is the most important thing.
I currently sort of have a solution that I might be able to tweak into working how I want. Basically I have it detect a thing to avoid, then two detectors "scan" it by aiming a raycast2d at it then rotating to the right or left until it no-longer hits the object, then it records those points to find the width of the object from the ship's perspective. It next compares the angles between the two points from the ship to figure out which one is the quicker option and overrides the normal nav system to steer the ship towards that point. The point itself is aware of the object the ship wants to avoid and makes sure it's X distance away from the object, to give good clearance away from it's edge. Once the ship gets close to the point it considers the object avoided and hands it back to the regular nav system. It has issues though, like getting confused if there are too many things near it so it will kind of fly around in circles for abit sometimes. It also currently doesn't adjust it's detection window further or closer based on speed though that wouldn't be hard to fix. It's reasonably fast but not instant.
I have another solution in mind that I'm going to build and test that might be way simpler and achieve mostly the same thing, so I will post about that later on.
Additionally one idea I had was to implement the best system I can achieve and set up the collision layers so that the ships themselves cannot actually collide with each other and the environment so that ships don't ever get stuck. So most of the time the illusion would remain intact but if the system should fail stuff won't get stuck.
MechWarrior Online is still trucking along, although it has to be noted that the MWO thread generally considers PGI to be bad at programming ("look at all the other companies who were able to do great things with CryEngine!"). And I often hear good things about State of Decay in the Steam thread. Other than that...
https://en.wikipedia.org/wiki/List_of_CryEngine_games
...yikes. Never even heard of most of these.
Basically I'm pulsing a raycast2d X units ahead, if it sees something it gets the distance to the object then sets the position of a gameobject that I've taken to calling an avoidance probe with a sphere collider in trigger mode x times the size of the ship's own (to give plenty of clearance, currently using 4x) forward of the ship based on the distance value gained from the detection and tells the probe what gameobject to look for. It then rotates the probe around the ship until it's collider is no longer overlapping the target gameobject's collider, at which point it tells the avoidance ai that it's got a solution for it, set's the coordinates of the steering point to where it is then resets back to the ship. The ship steers to the point and once it gets within 5 units hands control back over to the regular nav ai.
It's actually not bad so far. I setup a little obstacle course for it and while it always favors right turns currently, it's not gotten stuck at all! (yet...)
I made a game where you fly a single rocket and try to cause as much damage to the enemy base as possible.
Steam: Handkor
Turns out, even though you can Get an input from anywhere within it, it won't actually hold the input value for whatever reason, you need to either pull it from the input pin directly or Set a local var and then Get that one. Kind of weird if you're used to coding, as you expect function parameters to be defined within in without having to copy them into a separate variable.
ITT we learn the true identity of DenverCoder9
Yes I'd like to not have to stretch long wires within functions too.
Steam: Handkor