CPU Maze Movement

I was thinking about a movement but me and the SDK don't yet understand each other so I will post it here in case someone wants to make this. (Or tell me its a crap idea)

I would describe it as a CPU controlled 8 direction movenet or a simple maze movement.

- Restricts object to specified directions

- When the object collides with a backdrop set as obstacle it tries the other specified directions in sequence for a clear way to move.

- a method for a choice of directions.
So at maze intersections you can trigger the object to select (randomly) a new direction to try move.

- Perhaps it can also to try to move towards a player action at the maze intersections. So it mindlessly tries to take the first clear direction that appears to head in the players direction.

It does not seem that complex but I don't see me having time to understand the movement SDK here shortly to make it work.

That is a great idea. Would like to see more CPU controlled movements in general.

I think this would work better if the maze junctions where not defined by objects as that gets messy with overlaps, missed collisions etc. Rather something similar the the path movement setup allowing the user to define the maze by laying out junctions points and joining them with lines.

I don't know how easy / hard it is to interact with the frame editor from the movement SDK though. The path movement of course is inbuilt but is something like that possible from the SDK? I am not sure.

The problem in general with CPU movements are making something generic enough that it works in most cases and not just for the one game type the creator had in mind, while keeping it featured enough to cope with more complex demands. Also I think the properties of movements have to be simple to understand for new users, too many options and you may scare people away from using the movement altogether!

So, for example, Pacman ghosts do their thing based on a movement being set alone?

If that is the basic intention, then yes, that would be another way to make a whole bunch of games very, very easy to create.

I would also like to see a player-controlled movement, (see rail movement) as well. That way, instead of using paths and other methods, the "Pacman" would only move in restricted, yet precise ways.

It would be far more intuitive and a whole genre of games would be quick-start. Create maze graphic background(s) w/ specific and consistent width, set Pacman to movement and set it's properties, set ghosts to movement and set it's properties ... tweak.

This would allow focusing on effects, graphics for each level, power-ups/lives/bonuses, etc.

I like this direction for MMF2 and wish we could use CM's to conquer various game types, over time. I especially see this coming to a peak when movements have direct access from the event editor without a special extension.

It may be possible to automate quite a few, rather common game movements and types this way, making less skill and events necessary to create them: Galaga, Centipede, Dig-Dug, Sonic, etc.

I wish someone adapted the easy grid extension to support all the features of the isometric grid extension. I'm talking of all the movement and file features. It's nice to have it for isometric games, but for top down games it is still missing.

If we also finally had a working pathfinding extension again all sorts of grid based movements would be easy and quick to do.


Speaking of easy and quick (this is an offtopic rant :/) - my biggest issue with MMF is that basically you can do anything you want - but the price you pay is "readability". Everything you want to do has to be achieved by using workarounds. And workarounds are hard to read. So if you don't touch an application for a month it will take you hours and hours to get into it again.
Having more comments than actual code doesn't make it better either. At university I was told that if code is written properly (and the language allows to write properly), no comments are nescessary to quickly read it.
So there are a few things that MMF really needs to comply with that demand (variable names displayed for qualifiers, qualifiers renameable, behaviors for qualifiers, comments attached to events, integrated OR/XOR and CASE statements that are not spread over several lines, etc. etc.). But most of this is not doable with extensions. So I guess this is the wrong place for this and I better go to bed now. Good night.

Quote:

---

Originally Posted by Random

Speaking of easy and quick (this is an offtopic rant :/) - my biggest issue with MMF is that basically you can do anything you want - but the price you pay is "readability". Everything you want to do has to be achieved by using workarounds. And workarounds are hard to read. So if you don't touch an application for a month it will take you hours and hours to get into it again.
Having more comments than actual code doesn't make it better either.

---

I would like to see the Commenter Object in MMF2. I miss that object.

Quote:

---

Originally Posted by Random

At university I was told that if code is written properly (and the language allows to write properly), no comments are nescessary to quickly read it.

---

I done programming for a long time and reviewed many languages syntaxes etc and all need comments to make them more readable. Its ok for basic code not to comment but complex code needs a roadmap and an explaination. A few of the extensions don't display everthing that is entered so that can make it really hard to read when you can't see the variables.

I would also like to see a text based script for MMF where instead of the graphic being displayed the object is replaced by its extension name found in the dialog insert object. I suggest using different brackets for object name and extension name.

Hmmm my thread seems to have turned into something else :)

There are many walls in MMF that authors constantly (are forced to) run into. I think it is good to discuss what extensions can do to get people over those walls.

Do you remember how fast loops revolutionized MMF's capabilities? In my opinion a lot more work has to be put into that department. And even some extensions should be integrated into MMF itself because they are so vital (Select Object Extension or Direction Calculator).

Sorry for off-topicifying your thread Jeff :/

In my oppinion one of the most important areas is how objects are dealt with at the most basic levels: How they can be referenced, using qualifiers in global events, dealing with multiple instances of an object, etc.

Back on topic: Simple AI movements would be cool.

As far as AI movements go they would be interesting, however in a way I kind of dislike them.

My reasoning behind this is that many people like to have complete control over their games. If there is a bug in the movement then they can't fix it or they have to wait months to get it fixed. If they need more power or something to work a different way then they have to start from scratch as there is no way to just tweak the movements.

While more movements would get newbies up and running quicker, they could result in a slew of poorly made games which will make MMF 2 look weak compared to the competition.

While movements are nice, I believe that more work needs to be put into empowering the users to create such movements for themselves through extensions that make the "harder" parts simpler, and with more examples on how such things are created.

So I would rather see objects to help with graphs and state agents then precanned movements, especially since AI is an area where MMF 2 could greatly improve.

I think its all about balance. I would like to see some extensions that do both.

I like the idea of having some paramters to tweak a CPU movement, but let MMF work out what it does based on that. Something to make it fun to play with and get better results than messing around for hours myself*

I'd also like to see some more complex extensions that allow you to do a little more. There are some of these in 'extension' form rather than movement form. And of coures there is a lovely physic extension in development which I hope someone finishes pretty soon! ;)

So a bit of both please, but I think we are still lacking in CPU controlled movements that have a little more intelligence than the bouncing ball, so I would love to see more of these types first.



* note: I'm maybe one of the few that really like the built in movements, including the Platform movement...oh yes. :)

On the subject of the platform movement, why isn't there a CPU controlled version that has some actions and properties to allow us to put characters in our games more easily? Come on Clickteam :)

for example, allow the CPU platform movement to walk the background as the player does, but when it comes to a drop that exceeds a certain threshold (which the player could define) make it turn around and go the other way insted of falling off the edge. Some actions to control the movment (turn it around, make it jump, climb a ladder, get off a ladder, crouch etc)...

I bet some there are load of good ideas to make this easier.


Another idea- bouncing platform movement (CPU or player - user decides). Same as platform movement, but the object bounces along, a bit like the pinball/bouncing ball mixed with the platform movment!

Here is version 1 of my evented maze movement.

I think the only thing I need to add to this is a check for intersections when the block is moving.
Right now it only picks a random open direction when it hits a wall straight ahead.

I have an idea about the intersections but I have not yet figured out some events.

Something like:
While the block is moving have a collision check the the side corners. If 2 corners are not over an obstacle the block can decide to take a new path.

Then another improvement might be figuring out how to fastloop this so my block can move faster.

I am sure Joshtek will come thru here and do it in 4 events or so LOL! I like to make it long and complex.

Yves -- The collision mask is very cool ... Now I understand :)

A little more OT to mess-up Jeff's thread ;) What fun, for a change!

Speed and ease-of-use are still my number one criteria for MMF. I agree, we have so much farther to go that we may have clearly but a big scratch in the surface, but it is only superficial enough to give us and idea about just how deep we can dig.

My opinion is that some people had a natural proclivity to do "amazing things" with this product line and ended-up looking superlative and then acted condescending and superior to those who could not grasp or use the more complex methods. It was not until a new breed of users came along and, not without effort and disdain, decided that sharing every detail of how to do things was the best route for everyone.

There is no reason to dumb-down MMF. In fact, I am all for more complex and powerful tools for those who desire to use and exploit them to the max.

Can we not strive for an inverse proportion ration here? Encourage and develop powerful, complex ways to users to really test their own limits and amaze others, WHILE adding ways to get down to the nitty-gritty and create something in an amazingly fast and simple way?

Clicking started out with making things easy to do and access. It has demonstrated that power-users found it a captivating challenge that payed off both time-wise and in showing off their capabilities.

We should always continue, no matter what it takes and how much we put into realizing it, this product shine as the ultimate way to do the most powerful and amazing things, even as growing newbie. That translates into as much instant gratification and great results as we can muster up for newbies and casual users who want to inflate their egos or materialize and idea, a dream, or a personal statement that shines with professionalism.

Can't we update our mission and paradigm to expand this cool tool in both directions? Personally, I have gone from the geeky, "look what I can do/know!" mode to more of a, "how much fun and demonstrable productivity can I have in this playground in an evening, a week, or so?"

We CAN do this. We are already well on our way. Defining and evolving just what it is we are accomplishing is starting to become an important factor as we move ahead, it seems.

I just want to say I am currently making a pac-man tutorial for MMF2/TGF2 that include both player controlled pac-man movement and random ghost movement.
The code is allready done. It uses smart detectors (thanks to Random) and that makes the whole AI movement in just a few events. I tried to do it with collision masks first, but that seemed to be too complicated for the new user.

It would be nice to see a 4-direction cpu movement extension that does all the hard work, but for now I think tutorials to help beginners see and understand the opportunities of MMF2 will be their best tools :)

Quote:

---

Originally Posted by AndyH

for example, allow the CPU platform movement to walk the background as the player does, but when it comes to a drop that exceeds a certain threshold (which the player could define) make it turn around and go the other way insted of falling off the edge. Some actions to control the movment (turn it around, make it jump, climb a ladder, get off a ladder, crouch etc

---

I love that idea. I think many users really struggle making half-way decent platform enemies. MMF really lacks in the foundation there. It took me like 9 or 10 years to figure out how to get it right and it still is not easy.

There should be events when the AI touches a drop in the floor, when another object comes in sight / or is in a line (no obstacles) of the same height (+- threshold value), when it touches a ladder, actions like climb up / climb down and so on. If it was easy to make such movements, you'd not need a fastlooped custom movement everytime you want something like that.

Yes, there's lot's of potential there for a platform object. Something that is probably harder for an extension developer to do as this player version of this movement is built in to MMF, so it might only be something we can petition Francois to add in a future build? (hint)

But maybe we'd better start a separate thread about our other movement ideas so Jeff can keep this thread on track (sorry jeff! ;))

Really back on track... I've played with Jeff's example and added a bit of code to cope with moving south or west into the crossroads (you can't get to it any other way).

Adding the T junctions in a similar fashion would require lots more events.

Wow Sarah that is easier then I had planned to try it :)

Just need to figure out how to C++ it :)

I've made an alternative version. It only changes direction on collision at the moment, but it does support multiple instances.

sweet 12 events LOL!

I wonder what you would need to do to make the blocks able to move faster? Use some of the ideas DavidN uses in his Fastloop platform tutorial?

Didn't Popcorn say he already had a solution and is writing a tutorial on it?

Yes but this is a different issue :D

This is so I can explain what I want my maze movement extension to do and to try to create it in C++ and the movement SDK.

If it can be evented I am assuming it can be done in the SDK.