0 Favourites

[Solved] pixel-perfect platform engine using loops

  • EDIT. SOLVED thanks to Magistross and retrodude and newt:

    .capx

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/jumpy3.capx

    playable here:

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/index.html

    ____________________________

    Hi everyone!

    I'm Christina. Long time lurker, first time posting.

    I've been disappointed with the lack of pixel-perfect controls in Construct2, since I have my sights on making a simple metroidvania as practice, and I really need to save time by using small pixelart sprites.

    This (http://69.24.73.172/scirra/forum/viewtopic.php?f=8&t=1095) custom platform engine by Kayin is awesome , but when translated into C2, it suffers from the 1-pixel inaccuracy problem (http://www.scirra.com/forum/platform-object-sinks-into-solid-ground-bug_topic57004.html)

    I managed to find an old trustworthy MMF2 method by Damian (http://damiansgames.com/post/6923285368/making-a-platformer-in-multimedia-fusion-2-or-games) and successfully ported it into C2.

    I had to make the player sprite's bounding box one pixel smaller on each side (player sprite is 25x25px, player bounding box is 23x23px), but other than that it works great. I'm sure I could clean it up a little, especially those "direction" variables.

    If anyone wants to take a look, the project is here

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/index.html

    and .capx is here

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/test%20movement%20with%20actual%20loop.capx

    So!

    1st Question: is there any way to implement slopes in an engine like this? I mean small, 1-pixel steps, that the player sprite can climb, like Kayin's. Check the .capx , that small pixel-high bump on the floor is enough to completely stop the player sprite - which is good, it means pixel-perfect collisions are working. I tried using a slopeDetector sprite, but it kept thinking the walls were slopes. Perhaps I'm missing something.

    2nd question: does anyone know why this engine won't play nice with dt? I tried introducing 60*dt instead of 1px steps and movement immediately becomes erratic. Jump strength becomes random as well. Does using pixel-perfect collision loops come at a price?

    Glad to be here :)

  • Hi everyone!

    I'm Christina. Long time lurker, first time posting.

    I've been disappointed with the lack of pixel-perfect controls in Construct2, since I have my sights on making a simple metroidvania as practice, and I really need to save time by using small pixelart sprites.

    This (http://69.24.73.172/scirra/forum/viewtopic.php?f=8&t=1095) custom platform engine by Kayin is awesome , but when translated into C2, it suffers from the 1-pixel inaccuracy problem (http://www.scirra.com/forum/platform-object-sinks-into-solid-ground-bug_topic57004.html)

    I managed to find an old trustworthy MMF2 method by Damian (http://damiansgames.com/post/6923285368/making-a-platformer-in-multimedia-fusion-2-or-games) and successfully ported it into C2.

    I had to make the player sprite's bounding box one pixel smaller on each side (player sprite is 25x25px, player bounding box is 23x23px), but other than that it works great. I'm sure I could clean it up a little, especially those "direction" variables.

    If anyone wants to take a look, the project is here

    https://dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/index.html

    and .capx is here

    https://dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/test%20movement%20with%20actual%20loop.capx

    So!

    1st Question: is there any way to implement slopes in an engine like this? I mean small, 1-pixel steps, that the player sprite can climb, like Kayin's. Check the .capx , that small pixel-high bump on the floor is enough to completely stop the player sprite - which is good, it means pixel-perfect collisions are working. I tried using a slopeDetector sprite, but it kept thinking the walls were slopes. Perhaps I'm missing something.

    2nd question: does anyone know why this engine won't play nice with dt? I tried introducing 60*dt instead of 1px steps and movement immediately becomes erratic. Jump strength becomes random as well. Does using pixel-perfect collision loops come at a price?

    Glad to be here :)

    Edit: I Foregot to welcome you to the forums <img src="smileys/smiley17.gif" border="0" align="middle" />

    But anyways, I'm not sure about your second problem, but is this what your looking for in your first problem?

    Link:

    dropbox.com/s/0fwfwbe8f5jhoew/jumpy..capx

    If so then I can explain how this works

  • Hi retrodude, thanks for the welcome.

    Yes, I'm trying to make "player" climb that big step just by going to the right.

    Hmmm just had an idea.. Maybe I can make a high bump detector sprite and a low bump detector sprite... If the high *and* low bump detectors are overlapping an obstacle, then don't let "player" move. But if only the low bump detector sprite overlaps an obstacle, then assume it's a step and stand on it. So "player" can tell the difference between a wall and a step. I'll test this!

    What was your idea?

    Edit: oh I see! You made a smaller step, on frame 2 of the platform animation. OK, how does this help? Do a check of if FramePlaying=2 -> set player.Y -=1 ?

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/jumpy2.capx

    It worked!!! This is so cool, thank you retrodude!!

  • Hi retrodude, thanks for the welcome.

    Yes, I'm trying to make "player" climb that big step just by going to the right.

    Hmmm just had an idea.. Maybe I can make a high bump detector sprite and a low bump detector sprite... If the high *and* low bump detectors are overlapping an obstacle, then don't let "player" move. But if only the low bump detector sprite overlaps an obstacle, then assume it's a step and stand on it. So "player" can tell the difference between a wall and a step. I'll test this!

    What was your idea?

    Edit: oh I see! You made a smaller step, on frame 2 of the platform animation. OK, how does this help? Do a check of if FramePlaying=2 -> set player.X -=1 ?

    https://dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/jumpy2.capx

    It worked!!! This is so cool, thank you retrodude!!

    Np <img src="smileys/smiley17.gif" border="0" align="middle" /> but I think I must have read something wrong because I didn't realize that you wanted the player climb the step when pressing right. However it seems you pretty much solved your own problems so nice job <img src="smileys/smiley2.gif" border="0" align="middle" />

  • don't sell yourself short, you pretty much solved this one! And it needs just one event!

    if player is overlapping platform

    and FramePlaying=/=0 -> set player.Y -=1

    I think this is quite a robust engine already. Now if we could figure why dt is so unpredictable with this one...

  • Hmm.

    This is very weird.

    Can anyone confirm this?

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/test%20movement%20with%20actual%20loop.capx

    I've set the (25x25pixel) player's bounding box to be 23x23, providing 1 pixel of padding (otherwise player is stuck) The code works great, all collisions are pixel perfect.

    If I change lines 12-15 to take dt into account,

    (for example if I change 'Set Xdirection to 1' into 'Set Xdirection to 60*dt'), collisions become very erratic. There's a random pixel-wide gap between 'player' and platforms/walls. And if I reset the 'player' bounding box to 25x25, 'player' doesn't move at all, possibly because the condition for standing still is satisfied (i.e. don't make a move if moving will result in overlapping a platform)

    Does anyone know the inner workings of this? I suspect it's the collision detection, but I may be missing something in code.

    I can't stress this enough: the code runs *perfectly* when I leave dt out of it. Ashley? Someone? No rush, I'm just obsessed with finding out if it can be helped. If not, does anyone know if it is a logical compromise to leave dt out of a game? Would it be game-breaking?

    Thanks in advance

  • I should also mention, the code only works correctly for odd pixel sizes. For example 25 works, 26 doesn't

  • My guess would be that the hotspot is different on an odd sized box.

    There is no pixel perfect solution for an even sized poly because there is no one center point.

    Again that's just a guess, I haven't checked out the capx.

  • newt: You're probably right. Thanks! I don't mind basing an engine on odd size tiles at all, I was just puzzled

  • If you want to use dt, don't change line 12-15. Change the lines that are affecting speed, namely lines 10 and 11. Also, change the loop condition "xLoopIndex <> 0" to "abs(xLoopIndex) > 1" (same thing for the y loop).

    I did those changes and everything runs fine ! <img src="smileys/smiley1.gif" border="0" align="middle" />

  • What's wrong with the built-in platform behavior? It's been improved a lot over the past few months, and now even works to sub-pixel accuracy (down to 1/16th of a pixel) with full support for slopes, jump-thrus etc. If there's a particular problem with the built-in Platform behavior we definitely want to fix it, because re-doing all that work in events is extremely hard and pointless if we can make the built-in behavior good enough. The Platform behavior is over 1000 lines of javascript and has been very tough to get right, so I expect you'd really struggle if you want to get the same robustness and accuracy as the built-in behavior using events.

  • Construct 3

    Buy Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Buy Now Construct 3 users don't see these ads
  • Magistross thanks for taking the time. Could you share your .capx? Because I tried the same yesterday, and it ruined xSpeed for some reason (-3 going left, +2 going right)and also introduced a 1-pixel gap again. You can see it if you drive 'player' in a corner and keep pushing: it sinks one (or half) a pixel in.

    <img src="https://dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/platform.jpg" border="0" />

    @Ashley

    Thanks for taking the time. So, this is the built-in (and awesome, don't get me wrong) platform movement:

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/C2platformmovement.capx

    When the 'player' bounding box is 25x25px, 'player' doesn't even move.

    When the 'player' bounding box is 23x23, 'player' moves ok, but again, the 1-pixel gap makes its appearance.

    <img src="https://dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/platform.jpg" border="0" />

    And this is my implementation which works without a glitch always, except when introducing dt (although Magistross may be on to something)

    dl.dropbox.com/u/28087823/Construct%20Examples/pixel-perfect%20engine%20with%20loops%20-%20no%20slope/jumpy2.capx

    It does all I want it to do, and takes advantage of C2's inbuilt bounding box and overlap functions, so your code is definitely not going to waste.

    I think this (http://www.scirra.com/FORUM/the-big-timedelta-headache_topic35671_page1.html ) discussion is really really pertinent to the problem, but I can't really parse the ConstructClassic-specific ideas, since I know things are different with C2. I do hope I'm not making a nuisance of myself.

  • christina

    I don't know if this is what you're looking for, but I have been dealing with the platform behavior and metroid/mario/megaman/cobra genre of games to teach some kids, also, I did my own project with all the learned knowledge.

    The equilibrium have been developed to meet a gigantic portion of collision checks and all did with detectors.

    On the first step, I give up of using the platform behavior, but since the behavior was updated even more, I changed the code to take advantage of it.

    So, I can say with certain baggage, the platform behavior is very precisely now, but, you'll need to think on new ways of checking the environmental conditions.

    The only thing I'm working around actually is the vertical speed, because I didn't figured out how to stop the player on a wall after a jump, all my tries finished with player slow sliding the wall. So, I decided to use an invisible sprite with no behavior and set it to meet the player X,Y, and when hit a wall, simple change who is meeting who, and the player stop on the wall =)

    Take a look on the sample (I don't know why, but dropbox is doing some weird problems and you'll need to use another navigator to play the game...)

    Press on my signature to see the system.

    Use the arrow to move, or the mouse.

    A is the jump, if I remember right.

    Take the box with the mouse.

    EDITED:

    I forgot about another problem to deal, because the player is not on the ground, in other words, he is on the air or wall, the falling speed will keep updating, and was necessary to set the speed to 0 while on a wall, or climbing a ceil.

    I don't know if it make sense, but you'll need to deal with the same thing on your study.

    [tube]http://www.youtube.com/watch?v=pJFQftlOdH0[/tube]

  • Here's the capx with dt based movement.

  • christina - can you share examples without third party plugins (SpriteFont)? I don't have third party plugins in my dev copy of C2. Also, your screenshots show perfectly aligned sprites don't they? I don't see any pixel alignment issue in them.

Jump to:
Active Users
There are 1 visitors browsing this topic (0 users and 1 guests)
Similar Topics Posts Views Last Post
Unread hot topic
0 Favourites
Tutorial: Platform School
267 93,734
Megabeard's avatar
Megabeard
Unread hot topic
121 11,755
Artpunk's avatar
Artpunk
Unread hot topic
88 18,483
anhduc3108's avatar
anhduc3108