Topic: NES-Style Castlevania Belmont Engine for GameMaker! (Version 0.3 UPDATE!)

[TheouAegis]

Here's an update of my engine: It's not ready to be released because of bugs that Game Maker is responsible for! ... Okay, I may have had some part to play in all of this.

Current Bug List Preventing Release

Depending on my code, if Trevor is a certain distance away from the first ledge in the video, he will fall through it when jumping towards it either always or when he attacks prior to reaching it. This bug has been crippling my engine from the outset and after every update I make to my coding I check if the bug has been resolved or worsened. ... Apparently it's worsened, so I had to scrap my last edit. I am pretty certain it is my coding partly and GM itself partly. Yes, GM has various bugs. The one that caused me to stop editing my engine two nights ago involves mirroring sprites with image_xangle -- GM increases the size of the bounding box by 2 pixels if image_xangle=-1.

I had an issue where walking back and forth on the stairs would make Trevor shift left or right a couple pixels away from his norm. I traced this to a line of code I forgot to edit (it had a sister code elsewhere which I edited but forgot to edit both of them at the time). So currently the only bug I am aware of on my stairs seems to involve my coding for attacking. When Trevor attacks while ascending or descending, he offsets the sprite away from its norm. I don't know why this is, since I've overloaded my code with so many conditionals (by the way, I am working on cleaning up my conditionals since some have been resolved with recent bug fixes). Therein may be the root of my problem, if I have a conditional in the stair code hindering movement and not in the attack code where I have been looking.

I don't have Items coded yet.

I don't have a single enemy coded yet, although that's probably not as important to all of you as Trevor's code.


If anyone can help me with the AI coding for crows and owls in Castlevania III, I would appreciate it. Other minor enemies seem to more or less follow a clear-cut boring pattern of their own, but crows and owls base their movement on Trevor's position relative to them. They appear to have two or three specific movements and then change them up depending on where Trevor is. If you can narrow down their coding into a set of timelines and conditionals, I would appreciate that.

[TheouAegis]

Gah! I fixed all my bugs, or so I thought. Then right before I clapped my hands and declared Trevor bug free, I discovered 2 more bugs that either weren't there before or were there but had been covered up by the last one I fixed. I realized when Trevor fell through a ledge and dropped continuously that it was because my collision detection was set to keep moving if he was inside a ledge. Yes, I wanted that but not while he was trying to land. Located it. Fixed it. Adjusted collision detection. Now he defies gravity while in a ledge and I can't find the conditional that's letting him do that! So I turned off my comp and pretended to go back to work until I get some sleep and/or food in my belly.

Don't forget to read the post above. I still await help deciphering crow/owl AI.

[Jorge D. Fuentes]

I will play some CV3 today to try and figure it out.
I needed to do this for my Castlevania Chronicles: Dracula's Curse game as well and, while I got a 'similar' mechanic, I never looked at the details on how the ravens work.

However, here are the owls:
1. The owls have spawn points in the forest area, recognized as 'black spots' in the forest trees.
2. An owl stays hidden for about four seconds (they do three 'blinks'), and all you can see are its eyes.
3. It takes three animations frames for the owl to come into view, and then one for when it opens its wings.  I don't know the timing of these three frames, though  The entire sequence seems to take about a second.
4. After it opens its wings, it will do an arc dive attack at whatever spot Trevor''s upper body sprite is, at that time (it seems to come after your face).  You can 'fool' the Owl by being in the air about to jump, so that when it comes at you you can whip it.  However, if you're on the ground, it will be difficult to hit him before he hits you, if his spawn point is, say, right over you.  It seems to come to the spot you're on the moment after its wings are extended, wherever that position is.
5. Whether it misses you or not, it will continue that arc up into the three upper rows of blocks on the stage to whatever X position that happens to be on (this means he goes up high enough and far enough where only an axe could reach it... maybe...).  It then will do one more dive attack, this time aiming at your lower body (it seems), and do one fast long sweeping arc.  Whether it hits or misses you, it will continue and go off-screen as if the arc were to end at the area of tiles above the screen.  It then will disappear and re-do its 'blinking' animation at the next spawn point.

The ravens are harder to predict.  They seem to change their movements depending on what animation/motion Trevor is doing.

[TheouAegis]

I'm torn between doing more programming or just uploading what I have done. I have the bugs worked out. Just have to add the Items, which should be bug-free because, except for holy water, they don't involve collision detection.

For those curious, for detecting if Trevor is standing on ground, I used collision_rectangle(bbox_left,bbox_bottom,bbox_right,bbox_bottom+1,ground,0,1) rather than place_meeting. I did use place_meeting for horizontal collision detection because there seems to be a "bug" with collision_rectangle allowing a sprite to keep moving. I didn't see anything in my code that would allow that to happen, so I'm assuming it's a bug. I couldn't use instance_position or instance_nearest because my system for handling tile collision involves scaling the boundary objects (doesn't get rid of empty objects but at least cuts down on the number) which renders various forms of object and collision detection useless.

EDIT: I wanna cry. The damn bounding box glitch in GM has left me with one minor yet annoyingly game breaking collision bug. One more day of debugging and i should finally be bug-free.

EDIT #2: OMG All that time trying to make Trevor walk through a ledge if he lands inside of one and supposedly he can't do that! There are limitations to it apparently. Now I need to find any places where such limitations exist. Like, let's say a "tilespace" is 16 pixels. Each "block" takes up 1 tilespace. So let's say you have a block. Then 1 tilespace to the right and 2 tilespaces down you have another block. If you jump onto that second block such that you land with your head inside the first block, you can move right (towards the second block) but not left (towards the first block to drop down). *bangs head on desk*

Anyone know of places in NES Castlevania games that look like:

       _  _  _
      |_||_||_|

          _  _
         |_||_|

 _  _  _
|_||_||_|

[X]

I've just gone over the maps for CV1 and CV3. There are some block placements that come close but not to what you're inquiring about. CV2 has no-such block placements due to the nature of it's program (not allowing you to jump through blocks from underneath).

-X

[Jorge D. Fuentes]

*****
Anyone know of places in NES Castlevania games that look like:

       _  _  _
      |_||_||_|

          _  _
         |_||_|

 _  _  _
|_||_||_|
**********

If the distance between the block tiles is two block tiles, then there couldn't be a place like this.
Trevor cannot jump at all if the space between his feet and his head is two block tiles.  It's the same for CV1 Simon.

I'm not quite understanding Trevor's position in those blocks.  Further explanation is required.
Let me break it down to see if I get it:

"So let's say you have a block"

Ok

"Then 1 tilespace to the right and 2 tilespaces down you have another block"

Ok, so it's below the original block by two.  This means that Trevor is going to walk off to the right and he would 'fall' two blocks on to it.

"If you jump on to that second block such..."

Stop right there for a sec.
Jump on to that block.... from where?  From below, or from another set of blocks you haven't mentioned?

Assuming that you're jumping on to that block from some place below, I think I can find a place like that... let me see...
Oh, here you go!  Check out the bottom-most room of the Haunted Ship!

-----------------------------------This area here, above this text, ↑↑↑↑↑↑

Trevor comes from the upper left staircases, and fights some skeletons and some ghosts, and then has to jump to the right, then jump up and to the left.  Then he has to do another jump and then climb the stairs on the right.

[X]

There are several sections throughout the game that also contain those same block placements. I saw one in the lost shrine of poltergeists.

-X

[Jorge D. Fuentes]

X is right.
Here's the spot, in that stage:

↑↑↑↑↑↑ Right there, on the upper floor, Trevor/Alucard/Grant have to jump through the two brick sections prior to walking the upper section to reach the Bone Dragon King.

[TheouAegis]

Now extend all those upper-level blocks over 1 more tile so they completely cover the ledge below them. You figured out what I meant, but I was looking for the extreme case. Because yeah, in each of those situations you found (which were the right situations by the way, sorry for being so vague), if you jump from the lowest ledge onto the middle ledge such that you land with only the front foot on the middle ledge and Trevor's head inside the highest ledge, Trevor can keep walking forward along the middle ledge but not backwards (which would lead him through the highest ledge but in turn also allow him to just fall off the middle ledge back down to the lowest ledge).

Well, I made use of collision_line() which seems to cause some slowdown in parts of the game, so I rearranged my code to hopefully reduce the slowdown (so only one collision_line() command would be executed at a time, I hope). The only problem with collision_line() and collision_rectangle() is they detect collisions differently than place_meeting() or instance_meeting(). They allow my code to be more accurate and so far seem to simulate the scenario I mentioned at above at the start of this post, and they make it easy to have Trevor stop outside a ledge while walking but fall inside a ledge while falling (which Trevor can do by 2 pixels either side of a ledge). But right now I'm having an issue which is no doubt caused by collision_line() because I've never had it any other time in my programming: if Trevor jumps and connects with a ledge such that his feet or legs touch the ledge but his upper body is still in open air, he will effectively land IN the ledge and be able to walk through it (because my horizontal collision_line() doesn't need to deal with his feet being inside of a ledge), but if his face or chest bump a ledge he falls down like he's supposed to. I might just need a second conditional to check against, but I'm open to suggestions.

I might make a backup and then try recoding for strict tile collision checking, but my current engine is just soooooo close to being perfect!

[Jorge D. Fuentes]

Indeed, I think you will need another conditional.

CV3 seems to have covered its bases, as it currently has is thusly:

When Trevor Jumps at a platform
AND he can land on it
//Trevor cannot land on a platform with no floor.  A 'Wall' platform.
//Trevor also cannot land on a platform if the jump trajectory does not allow it.
AND there's a platform above
//Seems the game knows where the edges, floors, and ceilings of platforms are.  It needs to, because it needs that information for Grant's Wall-climbing abilities.
AND Trevor is at the edge of said platform's side wall when he lands
//So the game checks specifically for these goofy instances.
THEN he can walk in the direction away from that platform's side wall
//So he cannot get stuck.
and only in that direction.
//but cannot go back to the lower platform.

The game does not have instances where there's another block tile.  However, there may be some instances regarding Alucard's Bat Transformation, that might create a 'stuck' situation, though in those cases, I'm pretty sure the heroes can just 'jump through' that plaform.

[TheouAegis]

Yeah the game allows for jumping up through a platform. Don't know if I"ll go so far as to emulate that aspect specifically. So much mechanics involved in the games. Every time I play it I find some mechanic I missed before, then I feel compelled to go back and try to fit that mechanic into my engine. Blah. Getting sick of it already.

[Jorge D. Fuentes]

At least you're not programming WallClimbing with Grant!

[X]

Well, I made use of collision_line() which seems to cause some slowdown in parts of the game, so I rearranged my code to hopefully reduce the slowdown (so only one collision_line() command would be executed at a time, I hope). The only problem with collision_line() and collision_rectangle() is they detect collisions differently than place_meeting() or instance_meeting(). They allow my code to be more accurate and so far seem to simulate the scenario I mentioned at above at the start of this post, and they make it easy to have Trevor stop outside a ledge while walking but fall inside a ledge while falling (which Trevor can do by 2 pixels either side of a ledge). But right now I'm having an issue which is no doubt caused by collision_line() because I've never had it any other time in my programming: if Trevor jumps and connects with a ledge such that his feet or legs touch the ledge but his upper body is still in open air, he will effectively land IN the ledge and be able to walk through it (because my horizontal collision_line() doesn't need to deal with his feet being inside of a ledge), but if his face or chest bump a ledge he falls down like he's supposed to. I might just need a second conditional to check against, but I'm open to suggestions.

Dude! I've seen this happen in CV2 Simon's quest! It almost always happens in the first mansion I visit. I jump to a ledge that's two blcks away from me and one block higher. And if I land it right, Simon is walking through at least four pixels of the block but is still able to move through it.

-X

[TheouAegis]

Yeah, I had wall-walking enabled prior to fixing my ledge-landing glitch. The glitch I was having when I was about to cry bloody murder yesterday I solved (for now) by putting a vertical collision check in a horizontal collision check event. I was trying to avoid such sloppy programming, but it seems it couldn't be avoided. Then I was having some glitch with my stairs, which I realized was because I messed up on some if-then hierarchy (I cleaned up my code to hopefully reduce memory usage but in the process made it harder to read because of all the spanned-out brackets). So now I have just one bug left to work out as far as I can tell: letting Trevor walk out of a ledge if he finds his head stuck in one, but of course only if there's a blank spot ahead of him. The solution (I hope) came to me on the bus ride home from work this morning, so I'll see if my idea works in a few minutes. The solution was so simple that it eluded me last night... Ok, it didn't elude me, I just felt it was too simple to be right so I made it more complicated to the point that it didn't work and then gave up on it. 

In short, I should have Trevor ready to upload by tonight (cuz I sleep during the day). Cross your fingers.

[Jorge D. Fuentes]

And I guess you're uncomfortable just 'snapping' him out of the edge of the platform and into the center of the block?
That's what I would do.  It's not the best-looking of solutions but it's forgivable.