Let me sum up how this Ludum Dare went for us this time. We wanted to work together, so we did jam this time — Makdaam did music, sounds and level design, and I made the simple engine and graphics. The things that I’m mostly happy about and think that we got right:

• I used love2d, having only about two weeks of learning it (and of learning lua). This seems like a sure way to fail, but it was actually fine.
• The AdvTiledLoader library saved me a lot of time, and Tiled let us design levels relatively painlessly.
• Using music as a part of the level design seems like a viable strategy.
• The workflow for my graphics is the same that I used in previous Ludum Dares, and it seems to work well, except for tiles, about that later.
• Making the distributions for all the platforms was very easy with love2d.
• I had an emergency plan of adding shooting to the game and turning it into a shooter if anything went wrong, but that didn’t happen.

What went wrong:

• If you are setting out to make an adventure game, make sure you at least have some outline of the plot. We left the plot for last, and it was a huge mistake — we only did very basic and rudimentary things so it can barely be called a game (you can’t even win). Another person taking care of the characters, plot and levels would be fantastic.
• The freehand style doesn’t work too well with tiles. It was hard to make them tile and still retain their rough looks. I also didn’t make all the necessary tiles, so you can see some gaps and inconsistencies if you look carefully.
• Collisions are hard, but fortunately getting them perfectly right is not necessary for the game to be playable.
• Sound support of different libraries varies greatly between different platforms, even with a multi-platform framework like love2d. The game crashed on Windows just because we pre-loaded the sound files into memory.
• An overlay layer is not enough to make top-down levels look right in respect to obscuring sprites.
• Defining the dialogues and quests directly in the map files is a mistake — makes it too hard to see everything at a glance and make improvements. I think that I should have only defined named locations, and have all the plot and text in a separate file that refers to them.
• We didn’t have the free time and any energy left to work on it for the third day of the jam. We could have come up with a real plot and polished it all much more, but we were so completely tired, that we haven’t.

You can play it here.

Some input from Makdaam:

• There’s not enough time
• Learning milky tracker proved to be more difficult than expected
• Sounds and music are not implemented right anywhere
• 1 out of 50 melody ideas can be rendered almost instantly in a tracker without any training
• 1 out of 10 melodies are usable
• sfxer FTW
• Audacity is enough to make your own synth samples with distortions
• none of those samples will be used in the final LD product, so don’t waste your time in Audacity
• prepare for your neighbor getting a new hifi during LD

Tags: postmortem