Yeah, that is also a good idea. Also you might have to force an enemy into a lava pit or something. How are you thinking about enemy AI in your 3D platformer?
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
Indie Game Development |OT 2018| Come, help, learn, show and tell!
- Thread starter Popstar
- Start date
- Status
Yeah, that is also a good idea. Also you might have to force an enemy into a lava pit or something. How are you thinking about enemy AI in your 3D platformer?
Yeah that's the sort of thing!
Enemies are a bit of an aside for CnW, none of them are particularly challenging on their own, but combined with the environment they can be. For instance in the western level you have to platform through a canyon area whilst bandits fire their six shooters at you, the challenge is dodging the bullets and taking out the cowboys without being shot out of the air. There's also a boss who's invulnerable and you must lead him to fall into a pit.
Last edited:
Thanks, fair suggestion. You know programmers, everything should be done with code :)
I mean, doing it that way, you solve the problem that you probably shouldn't (?) be able to click something thats being removed anyway, and you can tinker with the animations like making it shrink in size while its fading, or move position, or whatever and not have to handle any edge cases that might crop up, because as soon as you say "I'm hiding this now" it stops responding to any further inputs
Thank you both :-D I'm not sure how feasible it'd be for any of you to come here next time around, but if you are, I'll be happy to show you around town!
Okay, try this on for size:
I don't know either, but I very much appreciate the invite!
That appears to be better, but I'd put them side by side and go with your gut. Also, I missed the round of comments when you first posted that trailer, but it looks great!
Yeah, that's much better already! :DOkay, try this on for size:
I don't know either, but I very much appreciate the invite!
Is there anywhere I can learn about best design practices for common game elements in Unity? Or just discussions surrounding different approaches? Admittedly I should probably go through their tutorials on the official website.
I ask because I was playing around and wanted to make a wall that the character couldn't move through. My current approach is to just throw trigger around the wall and push the player back on enter...but I also found I could get results by throwing rigid bodies on everything and setting their mass accordingly. I also thought of a few other ways to tackle the problem, the issue is that outside of game feel implications (i.e. rapid bouncing off the wall vs a hard stop) I can't figure out how to determine if an approach is a good one.
I ask because I was playing around and wanted to make a wall that the character couldn't move through. My current approach is to just throw trigger around the wall and push the player back on enter...but I also found I could get results by throwing rigid bodies on everything and setting their mass accordingly. I also thought of a few other ways to tackle the problem, the issue is that outside of game feel implications (i.e. rapid bouncing off the wall vs a hard stop) I can't figure out how to determine if an approach is a good one.
Finding the actually best approaches is rather difficult imo. It also depends on what kind of scenario you're talking about. Unity's tutorials are helpful and will teach you a lot but, and it's a big but, most of their materials are generally aimed at complete newcomers and getting them up to speed fast and easy. That usually means that correct and efficient methods are ignored and must either be found through trial and error or extensive googling. Unity is built around being both newb friendly AND being adaptable through more hardcore coding. Different scenarios can clash at different places on that spectrum.
As for your example, rigidbodies and colliders are usually the way to go. Unity likes it when its physics engine is involved.
edit: never mind. I came up with a solution. :D
original post said:unity mecanim peeps
is there a good way to change how mecanim normalizes values in a blend tree? I have two animations where bones are at equivalent positions and scales, but a different position/scale than the default pose for the model.
These are exported straight from the animating software:
And this is what happens when I blend with a blend tree in mecanim at around half speed:
as you can see, the ball appears to grow in size and the character lowers in height. My best guess is that this is some flawed normalizing implementation because the bones don't have their default size/position in the animations, but I don't know what the best way to get around it is. Like the ball is at 75% scale in both animations, so maybe halfway between them is ~86% for some reason? I can't remember the math for this. I would think internally it'd be Lerp(a, b, n) but it almost seems like it gets the square root of the right value.
fwiw animator.linearVelocityBlending seems to be used to correct this for velocity blending, but it doesn't fix transform blending.
Last edited:
I flicked between the two on the fly and it definitely felt better! Thanks for the kind words on the trailer too!
Excellent news!
edit: never mind. I came up with a solution. :D
I love this, both the physics and effects (does it pick up speed down-hill, could lead to some interesting platforming challenges!?)!
It's reminding me of the thought bubble/levitation ball from Psychonauts
shhhhh!!!! I'm not a theif
I haven't explored all of the gameplay implications yet. I'll probably do some more of that next weekend.
edit: never mind. I came up with a solution. :D
How did you go about fixing this? for future reference! :)
Also, I never knew I wanted to roll on a ball until this. What is the plan for it? can it kill things like road kill?
We've been contacting universities, especially games courses, to arrange play testing sessions. They seem really up for it, as the students get some cool experience, feel part of the game, and get to ask you questions about actual game dev.
We're doing 2D AI for our game. Ours is a stealth game, so the main considerations revolve around that. I'm not involved directly with the AI code, so can't help you myself, but Joe Wintergreen of Impromptu is doing our AI, and he has some really interesting tweets on AI.
Also, our designer Chris has done a few dev streams on our AI, so hopefully these might help, or corroborate your own conclusions: Stream 1, Stream 2, Stream 3, Stream 4
I hope some of that may help in some way.
Last edited:
I see, thanks! Guess I'll just continue the Google process and see what works and what doesn't.
Huh, interesting. I thought it would have been better to manually check which side a wall is on and then stop any changes to the transform in that direction. I'll run with the physics engine and see how it goes then. I guess it's not a big deal when it's still in a 2D space and there's not much complexity going on in-game.
Great thanks I will check out the streams! No stealth in my game but I can really see that you can do a lot of interesting things with stealth and AI. A lot of things that can be improved in games when it comes to creating fun stealth.
Colliders are 'cheap', rigidbodies are 'expensive'; if you're using built in Unity physics, only put rigidbodies on things that are going to move, anything that is not going to move ever should be tagged as static - it lets the backend take big shortcuts in physics calculations.
It also prefers simple colliders (box / sphere) for its calculations.
Yeah, this for sure.
Feel free to ask here. Unity Answers can be pretty good too. But while many answers are fine, they can also be from 2011 and then you really need to keep a skeptic mind and analyse the answer.
I'm loathe to call it fixed because it's pretty messy under the hood, and this implementation wouldn't work for any blend tree with more than two motions. I changed the blend mode from 1D to direct, and I supplied the first motion with a parameter whose value stays at 1. Then the parameter that used to blend between the two motions just controls the second motion in the tree. I made sure to uncheck normalize here, which gives me the poses I want but it's too slow (I don't know why). So I added yet another parameter whose value is equal to one plus the value of the previous parameter, and I multiplied the speed of the whole blend tree (on the base layer) by that speed.
I got a Unity raycast question. In our new project, we're using raycasts to determine if a new room in the room generator can fit without clipping through other rooms already created. Thing is, the code is based on my first prototype rooms, which had outerwalls with vertices and polygons etc. But now, all rooms are one-sided so there's only visible walls from within them, to save on stuff to draw, and to be able to see into the room through the editor while working.
Problem is, our raycasts apparently doesn't understand that they have hit something if they penetrate an object from the wrong side so to speak. Is there another type of raycast for this or something? It seems to be that the penetration still occurs, at least visually, so the lack of a backface shouldn't be a concernm or so I thought. All objects are two-sided when it comes to shadows if it matters.
Problem is, our raycasts apparently doesn't understand that they have hit something if they penetrate an object from the wrong side so to speak. Is there another type of raycast for this or something? It seems to be that the penetration still occurs, at least visually, so the lack of a backface shouldn't be a concernm or so I thought. All objects are two-sided when it comes to shadows if it matters.
There is RaycastAll. It will give you an array of hits. Maybe check the direction of raycast vs hit point to see if you hit the face or back. Something like that.
Thanks, I've sent it along to the group's programmer. You wouldn't happen to know of a good/better method of randomly generating an environment based on doorways between different prefabs, would you? :) I suspect send out 5 raycasts in different directions for each and every room, upwards of 30 rooms, isn't the best of ideas from a performance standpoint.
Raycasts do not hit backfaces by design, so I don't think this setup is going to work. Is it possible to use thin box colliders for the walls rather than mesh colliders? This would get around this issue and also likely be more performant.
Yeah, I was going to question how come rays aren't hitting the colliders, and didn't think about mesh colliders - you're probably right that thats what hes using
Raycasts are pretty cheap, a fixed length ray during a 'build rooms' script shouldn't be any real problem
Oh okay, good to know. I suppose I could add a bunch of colliders around each room, though it would require quite many, even for a single room if you want it to look similar to the primary shape. I also have abut 30 unique rooms right now, with an additional 40 duplicates in differet directions lol.
I see, well we'll see what we'll do then!
Very admirable! Looks like you have light at the end of the tunnel...I know what it's like having something on and off the boil for a long time. It must be very satisfying to have something that' finally getting close to release. Best of luck with it!
Can you flip the normals before raycasting the room from the outside?
Cool, thanks for the help. Appreciate it :)
I think to give a good response, I would need to know more about your game. For now I'm going to assume some things based off your previous post, which included a picture. So it's a side scroller, with high, med, and low parry. It looks like you can't jump.
A game with that selection of controls seems like it's less about really smart enemies and more about enemy patterns. You want to player to be able to recognize the attacks, and figure out how and when they can attack.
Single Pattern:
An enemy with a single pattern of attack is very easy. Up attack, Low attack, and wait. Then repeat after the wait. This allows the player to block up, block down, then attack. I imagine this is a very rewarding play loop.
Multi-Pattern:
This creates a much harder enemy and I would limit them to 3 patterns. It creates a Simon says situation where the player needs to memorize the pattern and differentiate between them to find the opening.
Up Attack, Low Attack, Up Attack, Wait
Up Attack, Low Attack, Low Attack, Low Attack, Wait
Middle Attack, Middle Attack, Up Attack, Low Attack, Wait
Unblockable:
Giving some enemies unblockable attacks could change up the pace. Give all of these attacks a charge up period. It gives the player a chance to go ham on the enemy, but it's at the risk of not getting out of range and dodging the attack.
Health Based:
Adding or removing patterns based on the enemies hp would add more surprise to important encounters.
Ranged Attacks:
Force the player to dodge attacks while approaching the enemy.
Mixing & Matching:
An easy enemy with a single pattern backed up by a Ranged attacker, would make for a much harder challenge, making your attack windows much smaller.
I don't know all of your design goals though. So it's hard to figure out exactly what your enemies should do.
I think to give a good response, I would need to know more about your game. For now I'm going to assume some things based off your previous post, which included a picture. So it's a side scroller, with high, med, and low parry. It looks like you can't jump.
A game with that selection of controls seems like it's less about really smart enemies and more about enemy patterns. You want to player to be able to recognize the attacks, and figure out how and when they can attack.
Single Pattern:
An enemy with a single pattern of attack is very easy. Up attack, Low attack, and wait. Then repeat after the wait. This allows the player to block up, block down, then attack. I imagine this is a very rewarding play loop.
Multi-Pattern:
This creates a much harder enemy and I would limit them to 3 patterns. It creates a Simon says situation where the player needs to memorize the pattern and differentiate between them to find the opening.
Up Attack, Low Attack, Up Attack, Wait
Up Attack, Low Attack, Low Attack, Low Attack, Wait
Middle Attack, Middle Attack, Up Attack, Low Attack, Wait
Unblockable:
Giving some enemies unblockable attacks could change up the pace. Give all of these attacks a charge up period. It gives the player a chance to go ham on the enemy, but it's at the risk of not getting out of range and dodging the attack.
Health Based:
Adding or removing patterns based on the enemies hp would add more surprise to important encounters.
Ranged Attacks:
Force the player to dodge attacks while approaching the enemy.
Mixing & Matching:
An easy enemy with a single pattern backed up by a Ranged attacker, would make for a much harder challenge, making your attack windows much smaller.
I don't know all of your design goals though. So it's hard to figure out exactly what your enemies should do.
Thanks for all the great input. Yeah the game is a lot about recognizing patterns. Most of the enemies you described will be in the game I also have enemies that stagger and enemies that does not stagger.
Your idea about two enemies with different attack patterns seems really cool. I am not sure I can implement that but I might try.
When I was thinking about enemy AI I was also thinking about other behaviours except attack patterns that could make the AI more interesting. I do not plan to have a living world in the sense that the enemies will be active all the time I rather plan to have them activate when the player comes close (similar to Dark Souls) which I guess limits what you can do with the AI.
Not sure, the rooms are made within Unity with ProBuilder, and there is a button for inverting the normals while in the editor, but I don't know if that's accessible at run-time. But considering we now have an indoors environment and not a single window, I think outside walls will be coming down the road.
Just a reminder: Theme voting is open for rounds 1-3 on ldjam.com, and Ludum Dare is this weekend. I'm hoping to finally participate again after skipping a few.
Unsure about what tools the player has at their disposal for navigating environments but I'm a huge proponent of building enemies alongside the environment they work in.
Enemies can pose a danger not just through their active behaviours, but also in how the player approaches them. They can pose a threat merely through placement in the environment and line of sight.
Howdy All,
Been lurking about this community for a few years and I figured it was about time to share. Also I have started and stopped so many side projects/games that I have lost count. So in order to change things up I wanted to share early, and maybe motivate my self to see this one through.
Anyway this is a UE4 project using Paper2D for all of the sprite and flipbook animations. I am shooting for a top down puzzle game, in a similar vein as eggerland or lolo games. All of the art is just a first pass, as a programmer by trade and don't have a ton experience with making art assets. Hopefully all of this will start looking better in time. The yellow walls and blue floor will not be sticking around...
Been lurking about this community for a few years and I figured it was about time to share. Also I have started and stopped so many side projects/games that I have lost count. So in order to change things up I wanted to share early, and maybe motivate my self to see this one through.
Anyway this is a UE4 project using Paper2D for all of the sprite and flipbook animations. I am shooting for a top down puzzle game, in a similar vein as eggerland or lolo games. All of the art is just a first pass, as a programmer by trade and don't have a ton experience with making art assets. Hopefully all of this will start looking better in time. The yellow walls and blue floor will not be sticking around...
Has anyone worked on a solo or close-to-solo project before? Sometimes I get a little tired of just playing games and get the spark for wanting to create a game.
I have some unique ideas brewing and sometimes want to make an indie version of the game I've always wanted to play but no one has made yet (for fun and fulfillment rather than profit). I'm a software developer by profession (non-gaming related) and an artist by hobby, so with the exception of music I think I could learn what I need to in order to create one. Working on both programming and asset creation myself would make the game feel really personal, which would be nice.
However, since I don't plan to leave my day job or start a team/company and just work on this in my free time, I'm worried about the scope/effort and whether or not I would lose motivation in the long run. My favorite game of last year was created by a team of two + a composer, but I have no idea how much time they poured into their work. Has anyone that developed a game solo/on a small team or is creating a game as a hobby have any stories to share on their experience?
I have some unique ideas brewing and sometimes want to make an indie version of the game I've always wanted to play but no one has made yet (for fun and fulfillment rather than profit). I'm a software developer by profession (non-gaming related) and an artist by hobby, so with the exception of music I think I could learn what I need to in order to create one. Working on both programming and asset creation myself would make the game feel really personal, which would be nice.
However, since I don't plan to leave my day job or start a team/company and just work on this in my free time, I'm worried about the scope/effort and whether or not I would lose motivation in the long run. My favorite game of last year was created by a team of two + a composer, but I have no idea how much time they poured into their work. Has anyone that developed a game solo/on a small team or is creating a game as a hobby have any stories to share on their experience?
Unity looks like it's getting some serious work done to it now which is exciting, they've hired some decent talent as well from other companies.
Sort of tempted to make the move back to Unity from UE4 (The UE4 compile times are killing me).
Has anyone made anything in Unity with a multiplayer focus?
Has UNet seen any advances recently? I always heard complaints from people about it but was never too involved to understand their gripes.
Sort of tempted to make the move back to Unity from UE4 (The UE4 compile times are killing me).
Has anyone made anything in Unity with a multiplayer focus?
Has UNet seen any advances recently? I always heard complaints from people about it but was never too involved to understand their gripes.
I'm a designer on a hobby multiplayer game. Don't know much about any new stuff though.
I'm a bit of an outlier but figured I'd chime in: I work solo (programming/writing/art/UI/design/music/etc) on my JRPG project while going to university full time, it's slow progress all the time but it's what I truly want to make so I stick with it. Scope is a valid concern, as well as possible motivation loss, but for me this setting I'm working in has literally been my whole life and something I plan to do a lot with over time, so I've never had a doubt I'll be making this at some point and have been poking at it for years without losing interest. My frustration is splitting my free time from life responsibilities and wishing I could deliberate some easy tasks to others to speed certain things up or not have to work for others especially when I'm in the groove, but it is what it is. Especially when I hit a wall where I have to learn a new technique that I'm not familiar with, and it takes more time away than I had hoped to get around, there's no one to have your back to keep the wheels turning so to speak. It just takes incredible discipline and a realistic look at what you hope to accomplish in a set time, especially if you are doing something beyond purely gameplay-focused games. I'm just stubborn above all else, but I believe in what I'm working on so I feel a bit justified in that ;)
Hopefully another solo dev has something more insightful and helpful to share, but that's my take.
I'm developing solo while working 50/60 hours a week at the day job.
I started with people but collaboration can be unreliable unless the other party is as invested as you are. The biggest factor is drive. You'll either be excited to come home and do more work or you really won't. Solo has it's upside as you are only accountable for yourself but it's drawback is that you've got nobody to keep you in check. Remaining focused is difficult, at times, but so far I'm quite happy with where I ended up after almost 3 years of collaboration.
It's not all roses and puppy dog tails but I like it.
My game is largely a solo endeavour. I decided to keep it small for many of the same reasons, I really wanted to retain creative control because my vision was so clear. Along the way however, I have found several incredibly talented people that I work with on occasion. They have really helped clarify where my vision was, but ultimately little has changed as far as my goal goes. Collaboration isn't something I'd advise ruling out, but doing it on your own terms is very much doable.
I would say it's worth pursuing if it's something you really want, and very much possible to fit around a day job and whatever other responsibilities you may have. However, there will be sacrifices made along the way, mostly as far as your social life goes! For time-scale perspective, it's been nearly 7 years for me, and I reckon I've got maybe another year left to go.
If none of that scares you off, I wish you the best of luck!
I work alone on small mobile games. It takes me about 1.5 - 2 years to make a proper game to completion. Make of that what you will.
- Status