Submission #7571: ShesChardcore's A2600 Dragonfire in 00:35.15

(Link to video)
Console Atari 2600 Emulator Bizhawk 2.8
Game Version unknown Frame Count 2106
ROM Filename Dragonfire (USA).a26 Frame Rate 59.9227510135505
Branch Rerecord Count 2804
PowerOn Authors ShesChardcore
Game Dragonfire
Submitted by ShesChardcore on 7/3/2022 5:52:48 PM

Submission Comments
This submission is for the Atari 2600 game Dragonfire. This is a fast paced and challenging game and I feel like it's one of the hidden gems on the console, simplistic though it may be.

Game objectives

  • Emulator used: Bizhawk 2.8
  • Aims for fastest time
  • Finishes all levels, to where difficulty stops increasing.


Dragonfire is comprised of two screens. First, you must cross a bridge to get to the dragon's castle, ducking/jumping fireballs along the way. Any fireball that hits you kills you and resets your position. Once past them, you must grab all the treasures in the dragon's room and exit out the top left of the screen where a door appears. The dragons increase in difficulty with each color, increasing move speed, number of fireballs they shoot, etc.
Once the white dragon is defeated, the game loops white dragons from there, and the difficulty stops increasing. You can use the game select to start on a higher level but with how short this game is I figured it'd be better to just do all the content. Memory address 21 is the "difficulty" which stops at 8 when you reach the white dragon.
Like most games of its time, the layouts are seed/frame based. The seed handles both the bridge pattern and dragon room layout together. There are various "good" bridge patterns. Any time you can get by with only having to duck and jump once is a good pattern. Depending on level and position, your duck will last 3-5 frames, which is totally fine. If you get a bad pattern then you lose a ton of time trying to wait and get past it. For this reason, I threw out all bad bridge seeds immediately when I was seed hunting.
The initial level seed is set on the reset frame. From there, the following level is determined by the frame you hit the door in the previous level. I dug through a ton of seeds to try and find optimal pathing but was unable to find anything faster than what I'm submitting. Good layouts were few and far between, and I intentionally bled some frames in certain levels to maintain a good path. Sometimes a faster level led to a slower level and it was tough to try and find a balance.

slamo: Claiming for judging.
slamo: Optimization looks good, the only delays are just for RNG manipulation. Maybe better RNG is possible, but it won't be easy to figure that out. Maxing out the difficulty and finishing a level is a suitable place to end the movie. Accepting!

despoa: Processing...

Last Edited by despoa on 7/8/2022 2:45 AM
Page History Latest diff