CaddieOpen in CaddieDownload for Mac

How to make a good platforming character (Developing 6)

Shared by Caddie Library · Updated
Game Maker's Toolkit14:49Open on YouTube
0:00Intro0:55Character Controller Goals1:34Tuning the Character2:59Additional Features4:56Juice and Animations5:52More Platforms6:56Throwing the Magnet7:49Magnet Juice8:45Accessibility11:01Getting Feedback11:48Burning Out13:26Conclusion
Chapters show their main points

The bottom line

A good platforming character comes from a simple base controller tuned for snappy response, forgiving input windows and juice, with accessibility built in at the right moment. Getting that feel took almost two months, so it deserves time and a break before moving on.

Key takeaways

  • A small, simple controller gives the best base to tune. 1:34
    • Mark picked Shinjingi's controller over tarodev's and Mix and Jam's because it is a very small amount of code he can put his own spin on.
    • It is a box with a few numbers: acceleration and top speed for snappy movement.
  • Two gravities make a jump feel responsive. 2:35
    • Low gravity on the way up feels floaty, and high gravity on the way down snaps the player back to the ground.
    • Variable jump height applies the heavy gravity the moment the player lets go of the button.
  • Forgiving input rules stop the game from punishing small timing errors. 3:06
    • A jump buffer keeps a jump request alive for a few frames, so pressing just before landing still works.
    • Coyote time allows a jump for a few frames after leaving a platform, and a capsule collider rounds off head bumps and edge bonks.
  • Juice and instant animation make the character feel alive. 4:56
    • Dust particles, a running tilt, screen shake and a magnet swing add charm.
    • Anticipation frames are cut and animations interrupt each other instantly, so the robot pops up the moment jump is pressed.
  • Accessibility is best added once the code is close to final. 8:45
    • Too early and the code keeps being thrown out, and too late it is a nightmare to retrofit.
    • Options include remapping, toggle aim, snap aim, turning off particles and shake, and tintable magnets with plus and minus symbols.

I did want to try a couple other things.

Mark, Game Maker's Toolkit

Chapter summaries

0:00 Intro

The MVP's robot felt bad to play, so the character needed a full rebuild.

  • Character feel was the biggest thing to fix 0:11
    • The robot felt slow and floaty.
    • Landing had an annoying hitch.
    • The jump was way too big.
    • The throw was overly complex.

0:55 Character Controller Goals

Four goals set before opening Unity cover feel, aiming, charm and accessibility.

  • Each goal targets one weakness of the MVP 0:55
    • Tune the physics so moving feels good.
    • Aim with a single button.
    • Add charm to robot and magnet.
    • Add rebindable controls and colourblind options.
    • The work started in yet another blank Unity project.
    • Goals one and two are ticked off later, at 6:48 and 7:43.

1:34 Tuning the Character

Low gravity up, high gravity down makes jumps snappy
Releasing the button early switches straight to heavy gravity, which is also how variable jump height works.

A borrowed controller secures the basics, and long tweaking of its numbers finds the feel.

  • Starting from a downloaded controller gets the basics right 1:40
    • Tarodev's controller feels very good for platformers.
    • Mix and Jam's is inspired by Celeste.
    • The MVP used a different controller.
    • Tuning took quite a long time in a simple blocky environment.

2:59 Additional Features

Coyote time and jump buffer forgive presses a few frames off
A plain 'are you on the ground?' check punishes near misses that land a few frames early or late.

Four small fixes from the Celeste movement research remove timing and collision annoyances.

  • Each extra stops the game punishing near misses 3:01
    • Jump buffer and coyote time work as opposites.
    • Variable jump height has been in games since Super Mario Bros.
    • A plain ground check ignores a jump pressed a split second before landing.
    • Coyote time stops players being punished for being a few pixels too late.

4:56 Juice and Animations

Cut the anticipation so the jump starts the instant you press
Squash-and-follow timing adds cartoon charm without delaying the response, though this is a choice for this game, not a rule.

Cuter art and carefully sequenced animation give the robot charm without hurting responsiveness.

  • The redesigned robot is cartoonier and still instant 5:06
    • Effects were borrowed from tarodev's controller.
    • It is still an orange one-wheeled robot, just cuter.
    • On a jump the screen pops up first and the wheel follows a second later.
    • Cutting anticipation is not a hard and fast rule, only what this game needs.

5:52 More Platforms

Slopes suit the wheel, while moving platforms needed community help.

  • Slopes were easy and moving platforms were a nightmare 5:58
    • Slopes add verticality without constant jumping.
    • GuitarKid's fix adds the platform's velocity to the player's.
    • The controller doesn't gel with the Unity physics system.
    • The help came from the GMTK Discord.
    • Mark then made the character jump along with the platform.

6:56 Throwing the Magnet

Throwing now runs on one button, and mouse support was the hard part.

  • One button handles pickup, aim and throw 7:14
    • Press to pick up, hold to aim, release to throw.
    • Pressing once throws quickly.
    • Pickup, trajectory arc and metal snapping came over from the MVP.
    • A mouse and an analogue stick are similar but not identical, so supporting both was a real challenge.

7:49 Magnet Juice

Extra animation and effects on the throw give the robot and the magnet charm and personality.

  • The throw is animated so it feels like a real chuck. 7:54
    • The robot swings the magnet behind itself before it throws.
    • The magnet gets different facial expressions.
    • Screen shake and a particle burst on a polarity change add more impact.
    • Sound effects are wired up but muted, because they are placeholders and sound design is a future problem.
  • Clearer magnetism visuals finish the goal of charm. 8:10
    • Dan Root of Video Game Animation Study suggested big chevron arrows pointing towards the panel.
    • The robot and magnet now feel like proper little video game characters.
    • That meets the third goal.

8:45 Accessibility

Add accessibility once code is final, well before release
Building options while the code is stable but young avoids both wasted work and painful retrofits, though it still costs time and extra code.

Accessibility options are worth the effort once the code is nearly final, even though they are slow and messy to build.

  • Colour options are built by tinting black and white magnets in code. 9:41
    • The game is all about red and blue, so colourblind support was needed in case.
    • The same chosen colour is applied to every panel in the game.
    • After a Twitter suggestion, a plus and a minus symbol now show where the magnet will sail to.
    • Red and blue are fairly distinguishable with most forms of colourblindness.
  • Accessibility is hard and costs code, but it was worth it. 10:21
    • It is time consuming and adds a lot of extra code.
    • One option was dropped because it added too much junk to an already messy, oversized script.
    • That meets the fourth goal.

11:01 Getting Feedback

The finished controller feels good, and a creator of Celeste confirmed it.

  • An expert playtest validated the feel. 11:23
    • Noel Berry, one of the creators of Celeste, played the demo and said it feels pretty good.
    • In many genres the main character is the backbone of the game, so it must feel good to move.
    • Patreon backers of GMTK get a short demo room.

11:48 Burning Out

One character took twice as long as a whole game
MVP: 10-level game30 days
One character controller60 days
The complete 10-level MVP took 30 days, while the character controller alone took almost two months of full days coding and bug fixing.

A long sprint of work calls for a break, so rest comes before the next step.

  • One character cost far more time than a whole game. 11:57
    • The 10 level MVP took 30 days, while this one character in a test room took almost two months.
    • That meant full days of coding, problem solving and bug fixing.
    • Players in the GMTK Discord sent suggestions, bug fixes and caught oversights.
  • Work remains, but burnout means stepping away. 12:29
    • The camera is a placeholder, there is no death and respawn, and magnetism is still broken in places.
    • Good game dev means watching your productivity, energy and mood.
    • Time away should give perspective on the next step.

13:26 Conclusion

The series pauses for a month or so, with a Unity basics tutorial promised later.

  • The devlog won the Unity award for best devlog 2021. 13:26
    • Jonas Tyroller finally finished Will You Snail after four years of development.
    • A basics of Unity tutorial will come later as a thank you.
    • No Unity for a good month or so.

In the growth library

Mark Brown, creator of Game Maker's Toolkit, developing his first game. Shows how to turn a slow, floaty platformer character into one that feels snappy and charming, using two gravities, input forgiveness, juice, and early accessibility work. The worked example is Mind Over Magnet's robot, rebuilt over two months after the MVP's character felt bad.

More from Game Maker's Toolkit · The whole library, and ask it anything

Transcript147 lines
Intro 0:00
0:00

Hi, my name's Mark and I am making a video game about magnets. So in the previous episode of this series I looked at the feedback on my MVP, and if there's one thing I really need to improve, it's the feel of the character. This little robot guy just feels slow and floaty, there's that annoying hitch whenever you land, the jump is way too big, and the throw is overly complex. So I've spent the last month and a half completely overhauling the character and I'm pretty pleased with what I've come up with. So please sit back and let me show you what I've made.

Character Controller Goals 0:55
0:55

Okay, so before I jumped into Unity I set myself four key goals with this character controller. Goal number one is just to make it feel really good to move around the environment, really tune those physics just right. Goal two was to simplify the aiming system down to a single button. Goal three was to add character, charm, and personality to both the robot and the magnet. And goal four was to add in accessibility options like rebindable controls and colourblind options.

Tuning the Character 1:34
1:33

So I opened up yet another blank Unity project and started working on the character. Though I did go and get a character controller off the internet to make sure I got the basics right. Now there's a bunch of really good ones online: there's this one which I used for the MVP. This one from tarodev feels very good if you're making a platformer. And Mix & Jam made one entirely inspired by Celeste. But I actually went with one from a much smaller youtube channel called Shinjingi. I like it because it's actually a very small amount of code which means I can use it for the absolute basics, but put my own spin on it. So Shinjingi’s character controller is just a box that moves around and there's a few numbers to play with.

2:21

For movement there's acceleration and top speed, so I can make the character get going immediately for that very fast, responsive, snappy feel. And for jumping this character actually has something really crucial: two types of gravity. You've got one for going up, which should feel a bit low and floaty - and another for coming back down, which should be very high gravity and snaps you back to the ground. That makes it feel amazing.

2:50

And so I just fiddled around with these numbers for quite a long time until I had it feeling nice to move around this simple blocky environment. After that I set about adding in

Additional Features 2:59
3:01

a few extra features that I documented in my video on Celeste's movement system. So first up: jump buffer. Basically, whenever you're coding a jump you first ask if the player is currently on the ground, and if yes, let them jump. But this means if you press jump a split second before you land, the game ignores your command and it feels crappy. So instead when the player hits the button the game should keep requesting a jump for a few frames and then you'll do a jump if you hit the ground within that window.

3:37

Next up, coyote time: the opposite of the jump buffer. Basically if you jump just after you leave a platform, again the jump won't register because you're not on the ground. So we remember that the player has been on a platform for a few frames after they leave it, and allow them to jump during that time. Again, this feels better than punishing the player for being a few pixels too late with their jump. We also need variable jump height. This has been in game since Super Mario Bros.: you know, you press the button for a short hop, or hold the button down for a big springy leap. To achieve this I just check if the player has let go of the jump button and throw on that heavy downward gravity, like, immediately.

4:22

And finally, it's never fun to bump your head when you get a little too close to a platform or bonk off the edge of a platform because you're not quite high enough. This was easily fixed by changing the character's collider from a box to a capsule. Now it rounds things easily. And at this point I feel really happy with how the character controls. It just feels good to move around an environment. Especially when compared to the robot from the MVP. But I can make it feel even better with some juice.

Juice and Animations 4:56
4:56

So I went back to that tarodev controller and, uh, stole some of the excellent effects like dust particles when you run, jump, and land. And this really lovely little tilt when you're running. I also changed the pixel art. It’s still an orange robot with a single wheel and a face on a screen but it just looks more cute and cartoony now. To really sell the cartoon effect I added some animations, so when the robot jumps his screen pops up first… and then the wheel joins a second later. I found that when working with video game animation you really have to cut out the anticipation part of the animation to make things feel more responsive. So the robot just like immediately pops up when you hit jump. Likewise, animations need to instantly interrupt each other to account for changes in player input.

5:45

That's not a hard and fast rule for every type of game, but it's definitely what this one needs.

More Platforms 5:52
5:52

Oh, before I finish, now that the robot works nicely on simple basic blocky platforms I did want to try a couple other things. So… slopes! They feel really nice with the wheel and it means I can add more height and verticality to levels without having the player jump constantly all the time. And moving platforms were a nightmare. Because of the way this character controller is coded, it doesn't gel perfectly with the Unity physics system so I was basically stuck. Luckily I've got the GMTK Discord which is full of super smart people who know Unity much better than me. User GuitarKid really sorted this one out with some black magic that essentially tracks the velocity of the moving platform and adds it to the player's velocity. But, you know, way more complex than that. For what it's worth though, I figured out how to make the character be able to jump along with the platform. So hooray for me.

6:48

At this point it feels like I have completed the first goal to make a character feel good to control. So next up it's on to throwing the magnet.

Throwing the Magnet 6:56
6:58

Luckily, a lot of this code worked pretty nicely in the MVP, like picking up nearby objects, aiming with a trajectory arc, and magnetically snapping to metallic surfaces. And so I could bring a lot of that over without rewriting it from scratch. Though I did want to, like I say, redo the controls. I wanted to put everything on one button so now you can press the button to pick up the magnet, and then hold it down to enter aim mode, and release to throw it. Or just press it once to throw it quickly. I also added in mouse support. And this was a little tricky: a mouse and a controller analogue stick are similar, but they're not identical.

7:37

And so making something work seamlessly across both is a real challenge. But I got it done and so that's goal two completed: to simplify the aiming system.

Magnet Juice 7:49
7:49

I also wanted to add more animation as part of the throw. I decided to make the robot, like, swing the magnet behind itself so it really feels like you're chucking this thing across the screen. I also gave the magnet different facial expressions. I added some screen shake - that feels good. I also added a particle burst when you change polarity. And I added some nice new visuals for the magnetism system. Thanks to Dan Root from Video Game Animation Study, the YouTube channel, for advice on this. He suggested the big chevron arrows that point towards the panel.

8:21

And I think with that, that's goal three met. Both the robot and the magnet now have charm and personality. They feel like proper little video game characters. Oh and I also wired up all the sound effects but they're just placeholders so I'm keeping them on mute for now. Sound design is a… future problem to figure out.

Accessibility 8:45
8:45

Finally, there was that fourth goal: adding in accessibility. You know, when it comes to putting in accessibility options I feel like you have to pick the right moment. You don't want to do it too soon when it's just a prototype and ideas and code are constantly changing and being thrown out. And you definitely don't want to do it too late, like really close to release, when retrofitting those options into the code would be a nightmare. You want to do it now basically. I’m still pretty early on in development but this is hopefully final game code. So it's the perfect time to implement some options.

9:25

And so I tried to put in as much stuff as I could like remapping controls. An option to toggle aiming on and off, instead of having to hold a button. A low sensitivity snap aim. An option to turn off particles and trails and screen shake and more. Oh, and I also needed to add in colourblind support. This is a game all about red and blue and while those are pretty distinguishable with most forms of colourblindness, I wanted to add in some options just in case. So by making the magnet actually black and white by default I could then apply a colourful tint in code and that way let you put basically whatever colour you want on the magnets, and apply that same colour to all the panels in the game as well.

10:09

I posted this on Twitter and people suggested I should add symbols as well and… okay you can now show symbols as well! A plus and a minus show up which tell you where the magnet will sail to. Now I have to admit accessibility stuff is pretty hard. Like, it's really time consuming and it adds so much extra stuff to your code. There was one option I wanted to have in the game but had to drop it because it just added so much extra junk to an already messy and oversized script. But other than that I've got a lot of stuff in the game and I'm pretty proud of that. Accessibility is really important and I'm pleased to have it in the game now. And that also means that goal four is well and truly ticked.

Getting Feedback 11:01
11:01

So overall I am very happy with this character controller. I mean, especially when I compare it to the disaster that is the MVP character. In a lot of genres the main character is the backbone of the game ,and so it just needs to feel good to move around the world. And with this controller I think it does. Though to validate it I had to do one important thing: I crossed my fingers and sent a demo to Noel Berry, one of the creators of Celeste. And he played it and he said that it feels pretty good, which, you know, made my friggin’ week! And if you want to try it for yourself everyone who backs GMTK on Patreon now has access to a special, super short demo room.

Burning Out 11:48
11:48

But while it does feel very good to get this thing done, I have to admit I am exhausted. Remember, the MVP - a complete, playable, 10 level game took me 30 days to complete. But this one single character moving around a simple test room took almost two months of long hard work. I'm talking full days just coding, problem solving, and bug fixing. Thank you to everyone in the GMTK discord who played various builds of the game and told me suggestions, bug fixes, and other oversights.

12:29

And I mean I'm still not done. The camera is a placeholder, there's no death and respawn, and the magnetism system is still a bit broken in places and needs some more code. But I think ultimately I do need to give myself a bit of a break: this was a long sprint of development. And I mean part of game dev is being aware and cognisant of your productivity and energy and mood. And I have to say I am nearing burnout on this project so I think the smart thing to do would be to put Unity down for a while, focus on some other types of GMTK video, and also think about the project more generally.

13:12

Like, what is the next step for this project, now that the character controller is done? I think some time away from the coal face of making games will give me a much needed perspective on where I should go next.

Conclusion 13:26
13:26

So until then, uh, oh wait! I won the award! I won the Unity award, the best devlog 2021. Thank you so much to everyone who took time to vote for the show over at Unity. I don't really get anything but it's still a good feeling. Also congratulations to all the other amazing nominees. These are great youtube channels you should definitely watch and subscribe to. Special congratulations to Jonas who has finally finished Will You Snail after four long years of development.

13:58

I did say that I would release a basics of Unity tutorial if I won the award and I will keep that promise, but right now I don't think I can look at Unity for a good month or so. But sometime in the future I will release that as a thank you. Until then, thank you so much for watching and I'll see you when I see you. Thanks, bye.