Jump to content

  • Log In with Google      Sign In   
  • Create Account

We're offering banner ads on our site from just $5!

1. Details HERE. 2. GDNet+ Subscriptions HERE. 3. Ad upload HERE.


makuto

Member Since 01 Nov 2011
Offline Last Active Yesterday, 12:56 AM

Topics I've Started

Designing a good feel ARPG melee combat system

05 October 2014 - 12:06 PM

Before you read this, you should try out what I have currently. Download the full game (made with Blender Game Engine) here, then open levels/level2.blend in Blender and press P over the 3D view (actual release will be a lot simpler).

 

I've been working on the melee combat system. It's a real time ARPG-style system like Skyrim (obviously with drastically less polish). Most of the code is done, but I cannot seem to get the feel right. By game feel, I'm talking things like Game Feel and this talk (fantastic, by the way). There is very little material available that talks about good feel for combat systems like this. If only a Skyrim dev would let some of their feel secrets be known!

 

Currently, the combat feels very floaty. The knockback was something I've thought about a lot, but knockback in combat systems varies quite a bit. It's currently a little too exaggerated.

 

It would help me out a lot if you would try out my game and tell me what you think. I've considered adding screen shake, a slight 10ms pause when you hit the enemy or get hit, improving enemy animations by having drawbacks (don't know what to call those - what I mean is things like raising the sword before slashing/hints that a hit is going to come soon), and sound (actually adds a lot to feel).

 

If I can get the combat system feeling solid, I think the game will be drastically more engaging and fun. Like the Halo devs say, get ~30 seconds of fun and repeat that over and over again in slightly different ways.


Discovering What Games I Want to Make

24 July 2013 - 10:56 PM

I'm participating in One Game a Month and have been learning a lot about game development because of it. Ever since October of last year I've finished seven games, five of which I see in a positive light and one that won first in a game design competition.

 

The problem that I'm facing (and have faced on all of my games) is that I'm never very motivated by the design. I'm designing games that I think will please the crowd and will fit in my ruthless schedule, not games that I will personally enjoy.

 

How do I begin designing & developing games for me, not for others (or do I always have to keep others' opinions in mind)? How do I find the game I want to develop?


Implementing an event or message system for decoupling

14 July 2013 - 07:23 PM

I've recently finished implementing a component-entity system (or COM) which is working well. My goal with this new system was primarily to encourage code/component reuse and speed up development of my games.

 

The problem I'm tackling now is how to maintain the decoupling of components (both sibling components and cousin/other-object components). C++ has been my primary method of development for four years now, so when I tried the Blender Game Engine for the last two games I've made, I noticed some things I wanted to implement into my own engine. The "message" actuator was particularly appealing. The ability to send simple text messages or variables to any object (or to all objects/groups) made interaction between objects simple, yet not extremely limited. I'd like to implement something similar in my own engine, at the component->object<-component, and component->group of objects<-all components in group.

 

How should I go about doing this in the best way (judging in order of importance: versatility, simplicity of use, performance)? I've thought about using a static array and storing global flags/variables (such as position) at an object level so that all components can get or set the variables safely. This solution doesn't allow easy/readable indexing, events (only stores data), and is difficult to get right with the dynamic nature of CES/COM objects. 

 

Game Coding Complete suggested using a polymorphism-based "event manager" consisting of events (which contain any data and must inherit from a base class) and event listeners (which are called when an event is heard and must inherit from a base class; these also must cast events to their true types). I want to avoid that much complexity in order to make coding for it easier and less time-consuming, but the functionality is there. 

 

How would you recommend implementing such a system? 


Rediscover CPU: a very niche educational game

25 March 2013 - 07:55 AM

I've been thinking a lot lately about computer processors and design, and was thinking about actually implementing my own simple virtual CPU as a learning exercise. These days, young programmers grow up on very high level languages without much knowledge of how the language (and the processor) actually work. I am one of those young programmers, so I thought I would implement a tool that would allow me to learn assembly and "Rediscover" things like time sharing OS's, the internet, etc. in a convenient and simplified way.

 

I've been designing the CPU to result in the simplest instructions possible to allow a smooth learning curve (at least for programmers). It's been quite a challenge seeing that it's difficult to find books on assembly language design, not just learning assembly language, but things have been going quite smooth. The CPU will likely be implemented in C++ and take a simple binary input file that will fill the main memory, allowing the player to edit their code using a hex editor. It probably won't be very fast, but it's about learning, so that's OK.

 

I've thought of a list of challenges the player is asked to complete (they self check), such as:

  1. Set certain memory cells to two others plus each other (set/mov, add)
  2. Add certain memory cells X amount of times (jmp, set/mov, add)
  3. Make a checker board pattern on the screen
  4. Use interrupts to stop a process with a key press
  5. Use interrupts to stop a process after X milliseconds
  6. Perform all basic mathematical operations
  7. Add a new I/O device hard disk (really just another binary file on their hard drive) and save something to the disk
  8. Make a Rediscover CPU assembly compiler in any way you want (on CPU or in high level language off board CPU)
  9. Build a time sharing OS that allows you to start, stop, and execute multiple processes
  10. Extend the OS to allow loading programs from an I/O hard disk (ideally make the programs position independent)
  11. Add keyboard input & basic text editing to your OS
  12. Allow programming the OS on the OS (at this point the CPU is independent of the real host computer)
  13. Network a friend and your CPU's together & send each other simple messages or files (networking possible by having a layer over the actual internet that simulates a simple wire)
  14. Extend your networking code to allow multiple CPU's and addresses
  15. Create a higher-level language compiler for your OS
  16. Continue building your OS, network, programs, and I/O devices
  17. Connect to another Rediscover CPU network
  18. Assist in development of new versions of Rediscover CPU

As you can see by that list, this game would be a very nerdy, very niche game (but very educational).

 

Do you guys think this is a viable game? Would any of you play it, or is it just me?


Designer's block with four days until deadline

24 January 2013 - 06:34 PM

So, I've been procrastinating and ignoring a deadline for a competition I want to enter. The deadline is January 28. I have four days to make a game.

 

This has been a stressful time for me because I'm still learning how I personally develop games, as well as what to look for in a new game
idea. My main constraints are these:

  • Shouldn't need any collision resolution, especially with a tile map, due to my difficulty with collision resolution (note that collision detection works already, that's easy)
  • Cannot be terribly violent (in the rules it states an "E-rated game")
  • Must run on Windows via CD or DVD
  • I need to be able to create it in a very short amount of time

Note that right now I'm not really facing very many implementation constraints, besides the collision resolution issues. I have a solid
codebase that has been successfully used on my first completed game, as well as tools such as the Blender Game Engine and Stencyl. I would prefer using my own code base, but I foresee I will have porting issues when I need to go from Linux to Windows, especially because I'm using a SFML 2.0 release candidate (which is difficult for me to get going on Windows).

 

 

I've been debating on whether or not I should just "game jam" it (generate a random topic and sticking with it), come up with a new design mechanically (I was thinking maybe using reward systems as a base), turn in a game I've completed that wasn't intended for this challenge, or if I should quit entirely. All the ideas I've come up with so far are either too time consuming to implement, contradict with my constraints (especially collision resolution), or fit too small of a niche. I have make a game that appeals to a judge, not just to myself.

 

So from a design perspective, what would you recommend me do?


PARTNERS