The VG Resource

Full Version: The GameDev Lounge 00000011
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Do any of you know Unity? My girlfriend is studying programming in Unity and I've sent her some assets - but the sprite doesn't seem to rest over the dirt tile correctly (it looks like the sprite is floating)

She sent me this video of her trying to play with the hitboxes a little, and this is what happens: https://dl.dropboxusercontent.com/u/3444...65.mp4.mp4 (hello Arisa)

Any suggestions on how to fix this?
I'm haven't done anything 2D specifically, but is the ground collision mask accurate? Is that mask that we see the one that's actually being used for collision? It looks like the sprite's collision mask is what's colliding instead. Does she have gravity enabled to make the player fall onto the ground? Has she changed anything with the way Unity handles collisions (through script).
So, I've an idea for a partially peer to peer massive multi-player networking solution.
You have 4 components, The Server, The Session, The Session Master, The Peer,
The Peer connects to the The Server, The Server finds The Peer near-by Session's and passes The Peer information to connect to The Session Master, if no Session Master's are near, The Peer becomes their own Session Master. The Session Master echo's information from all Peer's in The Session to all Peer's in The Session (roughly 4 Peers to a session, large groups will contain many Session's and Session Master's) for accuracy and security, Session Master's pass information to The Server in the form of before states, a list of actions and after states, The Server evaluates the potential costs of the actions and if all adds up then new information is stored, if not the Session Master is cycled to another Peer in The Session and The Peers with inaccurate information are tagged. Following a few cycles if The Server notices a specific Peer getting inaccurate results, The Peer will be warned that their actions are being logged, an admin or moderator can later evaluate whether the inaccuracies are suspicious or not.
The key concern I have is The Session Master and how seamlessly The Session Master might switch, if The Session Master times out a new Session Master must be found with accurate enough Session information and quickly, alternatively in the event of a Session Master timing out, I suppose all Peers in The Session could act as Session Masters until The Server appoints a new Session Master, this would make poor Peer connections a concern though.
The reason I was considering this is I'm thinking of how to lower the cost of a server, by spreading its workload out, this isn't intended for super accuracy as you might like in something like PvP or high risk games where a single hit may make all the difference, rather games like Warframe (been enjoying this thoroughly as of late) wherein many actions may make a small difference and now and then a larger action will make the noticeable difference. Its a handy kind of game because priorities become clear and cutting corners can work in the right cases.
(07-18-2015, 03:24 PM)TheShyGuy Wrote: [ -> ]I'm haven't done anything 2D specifically, but is the ground collision mask accurate? Is that mask that we see the one that's actually being used for collision? It looks like the sprite's collision mask is what's colliding instead. Does she have gravity enabled to make the player fall onto the ground? Has she changed anything with the way Unity handles collisions (through script).

She didn't change anything in the script, and the ground tile mask is automatic (quad mesh with textures). Also she has gravity enabled (rigibody + physics 2D).

I have no idea how Unity works, but yeah my guess would be that either the ground or character has inaccurate collision hitboxes

Anyway I told her your pointers, she's gonna tinker with them and see if it'll produce any good results

EDIT: She checked the mask, and she found out it's "box collider 2D". Does this ring a bell? The main problem is that both collision boxes don't touch each other (there's a slight gap between them which makes the character sprite look like it's floating).
Is she still having the same problems?
She 'fixed' the problem by lowering the ground hitbox, making it shorter than the actual graphic. Now, Arisa still floats, but the sprite appears to be resting on the ground correctly.
I didn't think I could help at all before but did she try reversing gravity and placing the player under the block to see whether it was the box was too big or just offset? preferably do this with a sprite which takes up the whole hitbox, for accuracies sake, maybe test left and right just in case too.
if the hit box is too big you can now narrow down efforts into which hit box is wrong, possibly via another very basic "nothing can possibly go wrong" box, if both are still too big it may be a quirk of unity's physics engine, if results are inconsistent across multiple tries it could be a problem occurring as result of a collision, perhaps manually handling momentum and such in a collision related event (quite common when trying to get a platformer not to feel slippy), if one is too big you have your target of investigation, perhaps manually setting the bounds would fix. If results inconclusive trying out many collision primitives could help narrow down whether or not its a quirk of unity or something to try and fix.
if the box is offset investigate the object trees (draw objects, meshes, rigid bodies, etc.) for any incorrect transforms.
If all else fails, a thought comes to mind regarding the half-pixel offset on texture sampling which might result in an unforeseen shift in the graphics, past that nothing comes to mind at the moment, but if this is still of use I can update if I get any more thought's on the matter
Been a while since I've been on here. 

Been recently doing a lot of Shader Programming in places like ShaderToy and CodePen:


[Image: CIdVKymWEAALDX9.png:large]
[Image: CMI3oz3VAAEHaJH.png]

This portal one in particular is really cool because there's no geometry, everything you see was raytraced on the gpu and modeled using functions! This was for the shadertoy siggraph competition.
Made a proc-gen maze game over the past month or so: http://kitsusplode.tumblr.com/post/12688...eese-chase




Relatedly, unit testing is new and exciting


also super weird
Cross post from MFGG WIP topic.

DJ Coco Wrote:All credit goes to TheShyGuy for writing the ink shaders. I only programmed the aiming/bullets.
All graphics are placeholders and subject to change.
[Image: z4wQd56.gif]
(The game window is also much bigger than this, I only resized the screen capture to keep the filesize of the gif small)
(08-26-2015, 10:14 PM)TheShyGuy Wrote: [ -> ]Cross post from MFGG WIP topic.

DJ Coco Wrote:All credit goes to TheShyGuy for writing the ink shaders. I only programmed the aiming/bullets.
All graphics are placeholders and subject to change.
[Image: z4wQd56.gif]
(The game window is also much bigger than this, I only resized the screen capture to keep the filesize of the gif small)

...is the placeholder character a block of tofu



....

shyguy u slyyyyyyy doggggggggggggg
Oyyyyyyy~

Been up and away in the world. Decided to come here now to wonder about next steps after actually getting finally down how GML basically works. I might attempt a very small game thingy to put what I learned to execution.

Dunno just what kind of game to try or even some concept. Do an example Tetris-clone or something?
Can I post something made from an engine?

I'm just messing around with isometric pixel art and 3D. This isn't an in-development fangame.
The 3D pixels are so col, man 3D lighting on pixels would be so awesome! I think you got something there. Was the frame rate reduced to kind of emulate a 2D feel?

Here are some problems I'm having with actionscript on video.

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20