Here's a new one. Assume I'm building a car-racing game, top-down, and the car is a rotatable sprite. I have built a track, but it's much larger than the screen. I want to be able to track the car's progress as it moves around the track, with the car stuck in the middle of the screen.

Basically, I want to do what we can already do with maps, but I don't want to use a map. It needs to be free-form.

So I end up with two coordinate systems, one for the track and one for the screen. (let's say the track in this case is 3200x2400 to keep it simple, and I'm running at 320x240)

Let's assume I use 'track coordinates' to place objects, like circles and squares to represent some obstacles. So I might put a circle at x1200 y200.

How do I create a 'camera system' to update the screen correctly? I thought about simply dividing by 10, but that effectively just shrinks the universe down to fit on one screen. Do I subtract the x's and the y's, then draw based on that? But how do I tell it how to draw that circle on the screen when it gets close enough to be on-screen?

Thanks.

It's funny... I got it to work using one system, and it works ok, but basically I'm just moving objects around relative to the vectors of the player. I'd like to be able to move the 'viewport' around relative to the track.

Can I draw the entire track (with unviewable portions 'offscreen' and then just move the 'camera' around? In this case, how do I relate the two coordinate systems to eachother?

-l

So...

For every render I am creating a new (screen-sized) with a snapshot of a piece of the larger image, blitting the new image to the screen, and then releasing the new image.

ImageShell is handy. It's exactly what I was looking for.

You guys thought of everything, didn't you?

btw: I sware I read the manual! I didn't see this.

-l

EDIT: WOOOT! Worked perfectly. Learned a quick lesson, though... you have to do some bounds checking. ;)

Thanks, McG.

This is actually much closer to what I am doing.

I'm not using an actual bitmap, I'm building my tracks from pieces, so-to-speak. Like you said, an array of objects which I parse, draw, then take a snapshot, then draw relevent data to screen.

Now I'm trying to figure out how to handle it when a player approaches the 'edge' of the track 'map'...

Right now (within the last 2 minutes) I have it setup to set the vector length to 0 if x and y are within 10 pixels of a wall. I would then add an accident/crash animation, etc. But I'm wondering if there aren't better ways to handle this.

I need to do some bounds checking when the screen image hits an edge.

Man... this would have been much easier if I had used the Map functions. ;)

Now, if it was a free-form driving sim, I could make it 'warp' into a new area. Example: the 'track' now becomes a 'city', and moving from neighborhood (2000x2000) to a new neighborhood could be done with some basic border or hotspot checking.

Wow. This opens up a whole new world of tangents.

I'll never finish a game at this rate.

-l

You know Grue, when you said if someone was crazy enough they could impliment OO through DMA...
Also, if you're having bounds issues locke, an easy fix is use GrabRegion()
rather than ImageShell(), as you don't need to do anything apart from a straight blit anyway, there's no reason not to, and will save you a bit of code as it just gives transparent pixels on out of bounds.

Zip

Quote:Originally posted by Zip

You know Grue, when you said if someone was crazy enough they could impliment OO through DMA...

Actually, Zip, I did a crazy method of semi-OO like this in v2, based on some classes that aen made back on the day. I'd already started implementing the SotS Character Structure in v2 as a quasi-object DMA dealie about this time last year, when v3 was announced and I turned my eye to other activities.

You can even pretend there's types if you close your eyes real tight and do:

#define CHARACTER int

...and the such.

However, v3's structures alleiviated most of my desires for dma objects... instead of passing around an integer pointer to an allocated chunk of memory, instead we pass around indexes to global master-arrays of structures in v3. Different methodolgy, and mostly the same effect. The only thing I lament is the lack of dynamic allocation, but really, put a hard limit that your system will never reach by design, and 'taint no thang, dawg.

However, if Feyd's making an easy-bake oven for dma-structure creation, well heck, maybe I'll go that route myself. Perl's a bit messy for my liking, though... maybe we can wrap that up in a php/javascript object-creation form and have the source-creator all happy-sappy for public use here on the website. After all, not everyone has activeperl installed on their windows box, nor does everyone have shell access on a linux box.

Ahh, that was just a quick example of the sort of thing that would be possible. I'm going to be working on a little sidescroller first when I get the infrastructure done, and once I've gotten an actual game under my belt I'll see about moving onto something really hard like an RPG. The main reason for doing this (aside from the coolness factor) is that polymorphism will be very useful for the windowing system I'm about to write, and it will save on a lot of tedious and error-prone DMA handling that I would've had to do by hand.

By the way, I've pretty much finished the code generator that allows me to do all that. I've written four examples of increasing complexity using it, and I haven't seen any new problems with it for a while. I'm working on the fifth (and probably final) example now, a simple DMA-based string library that I'll use to illustrate how to encapsulate a DMA structure into another one without causing memory leaks as long as you remember to free the enclosing class, and then I'll toss the Perl script that does the generating to anyone who wants it, along with the examples.

I've done a fair job of bringing OOP to Verge (which may or may not be a good thing ;P), if I do say so myself...certainly better than I expected. The constructors will automatically fill in fields if you specify the field names as parameters, and will also allow you to pass non-field parameters for processing by any custom constructor code you include. The destructor statements for cleaning up after a class are passed along to its subclasses automatically, and they can add on to them to do their own cleanup. Get/Set functions for each field are generated automatically (though you can specify read-only, write-only and private fields if you want). Function inheritance and polymorphism work beautifully. You can specify an expression that will be used to come up with an extra amount of memory to tack onto the end of the class, if you need it (for something like DMA strings or variable-sized list nodes.) Field names are expanded in your code, so that something like

Code:


void PlayerGetAttacked(int Player, int CurrentEnemy) {


  $HP$ -= $Monster.CurrentEnemy.AttackPower$;


}


EndCode

will expand to

void PlayerGetAttacked(int Player, int CurrentEnemy) {


  dma.quad[ThisPlayer + CLASS_Player_FIELD_HP] -= dma.word[CurrentEnemy + CLASS_Monster_FIELD_AttackPower];


}

with all of the #defines generated automatically.

Comments indicating the different fields for each class, their access levels and whether they were defined in the current class or inherited from a superclass are generated automatically. Ditto for the inherited functions - comments indicate the function signature, whether they were first defined in the current class, inherited from a superclass, or overridden in the current class.

*cough* Okay, I'll stop gushing now. I'm quite pleased with the way it came out, and if anyone wants to see it then I expect to submit it after I get out of class tomorrow afternoon.

A quick excerpt from the polymorphism example before I head off to sleep:

Log('Everybody flex! (polymorphic style!)');


Log('Don't hurt yourself, Mage.');


Flex(mage);





Log('Show us that bod, Fighter!');


Flex(fighter);





Log('What's under the Ninja's shirt?');


Flex(ninja);





Log('Just give us a little peek, UberNinja.');


Flex(uberninja);

gives us this v3.log file:

Everybody flex! (polymorphic style!)


Don't hurt yourself, Mage.


He tries to flex, but pulls a muscle.





Show us that bod, Fighter!


He flexes his rippling muscles.





What's under the Ninja's shirt?


This guy is too cool to flex.





Just give us a little peek, UberNinja.


This guy is too cool to flex.

in which Mage inherits his Flex() from the Living base class, Fighter overrides Living's Flex(), Ninja overrides Fighter's Flex() and UberNinja inherits Ninja's Flex().

P.S. You might get your red/black tree in a few days if I don't find any major problems in my generator tomorrow. ;P One of the examples is a simple binary tree with insertion, sorting and in-order traversal with a callback function, like the ForEach() I did for the linked list...wouldn't be a whole lot more work to subclass it and override the insert function to do the red/black balancing.

PHP has, I believe, perl-style regexp.

The javascript would be to manipulate the html forms, for adding new rows of inputs for new data members, etc.

My whole point is that if it has a happy-shiny GUI face on it, more people will use it, even if it's only meant to be used by people who can handle CLIs.

People generally are more likely to use something that panders to them. I know I'd either like a web-interface around such a util, or a quick'n'dirty VB app... and I offered a web-interface because I'd be much quicker at it than at making a windows app.

Actually, I could probably get away with cheating, really.

Just use the webpage to build the makfile format, then send it to the script on the system, then fetch back the output.

I'll have to ponder your arcane format when I have more sleep in me, however ;)