I am developing a game called Mad Meddlers.
The story is that 8 characters have a vision of Jesus Christ telling them to not meddle,
they disobey and get into fights over it.

Developing this game has been a very interesting journey with a lot of pivots.
Originally, it was a party game; however, during initial playtesting of the prototype,
people had trouble picking it up and understanding what it was.

So then, I decided to change it into a 3D platforming based arena fighter.
A fighter has a clear objective, established gameplay framework, and it is something that players can understand quickly.
Also, I am passionate about fighters.

Using GGPO rollback networking code for this. I am so glad that I am developing the game to work with GGPO from the ground up, it is so much work to change gameplay to be deterministic and work with rollback.

Here is an old sync test for the game. I hope to make more videos to show current progress.
https://youtu.be/LqNM1OxiaUE

Unique features of the game.
Zombie mode - defeat a player and take over his body.
Tag mode - Play tag.
In depth movement mechanics, long jump, fast fall, and crouch.
Planned roguelike mode called Ultimate Trial.

Game is heavily inspired by Super Smash Bros. Melee, Super Mario 64, A Hat in Time.

Current status. Trying to fix desync bug with reaction to guarding against attack in prototype. For some odd reason, attack box of player that is doing attack on guard is desyncing.

After that, I will playtest heavily and get feedback.

7 days later

Made progress on fixing desync bugs caused by not saving/loading rotation in game state.

Made acceleration and max velocity completely dependent on input so that it does not get desync'd on rollback.

Using linux tc to simulate connection with packet loss and delay.

Changed grab mechanic so that player can escape from grab by mashing buttons quickly.

14 days later

I had a mostly working online game the other day. It worked for so many attacks and frames at 35 ms ping except for 1 desync that happened after a throw.

I received some recent feedback on my game as well. It was very good feedback that showed flaws in my game.
The characters do not feel different from each other, presentation is not clear, stages too big.
Presentation meaning how graphics and sound communicate game concepts and what is happening.

I agree with these points.

I agree that the characters do not feel too different from each other even though in code they have different stats such as speed, jump, attack, and size.
Often in life, the perception of something matters more than what is actually happening.
However, looking at the characters , it is now obvious to see that they do have the same base and only deviate a little bit from each other in stats. I need to exaggerate these difference in how they move and attack so players feel their unique strength. After this, players can decide which character to pick based on which strength they prefer.

Masahiro Sakurai, the designer of Super Smash Bros., mentions character strengths and weaknesses in one of his design videos on Youtube. I will add a link if I find it.

I am glad to get feedback on my game. I find that I am often too deep into programming the game that I am blind to flaws in the game.

6 days later
10 days later

I added projectiles to my game and focused on following character archetypes.
My game has 1 shoto/base, 1 rushdown, 2 zoners, 2 grapplers, 2 spacers.

I sped up the workflow for setting hitbox frame data by putting time into fixing and improving the hitbox animator tool. The tool works well now I can easily align the hitbox to character weapon and save location information to a .json file.

I am almost done with version 11.

I was inspired by this video by Electric Underground to add some shmup elements to the roguelite mode that I was planning.
https://www.youtube.com/watch?v=HSS_3iTIgA4
I changed my plans for roguelite mode. I am going to do the opposite of the open-world gameplay trend and just make a small room interesting and dense. This works much better with the time sync requirement for rollback netcode gameplay.

I am adding input frame buffer of 4 frames to my game. The game does not have sequenced attacks like a traditional fighter, so it works more closely to the platformer input frame buffer.

The game feels so much more responsive than the previous per frame input I was using.
The buffer allows me to feel like attack is activated much earlier than expected unlike per frame input because I can see that the player tried activating attack before it could be activated with input history in the buffer.
I am using a circular buffer for input frame buffer to use less resources.

Input frame buffer also allows for the player to activate an attack in several different ways. For example, to activate a jump attack the different combinations can be done such as attack and jump at the same time, jump 1 frame before and attack afterwards, or even attack 1 frame before jump.

Input frame buffer is much more forgiving and is literally much more responsive because players do not have to wait for the character to enter into a specific frame to activate an attack and it allows for the same response to different input conditions.

    4 days later

    Yes, in the near future. I am looking for a local music composer in the USA.

    Thank you for the reply. Well, I am not from USA. I am from India. let me know if you are interested in remote composer 🙂