Posts for garland


Experienced Forum User, Published Author, Player (19)
Joined: 7/29/2018
Posts: 5
I have been trying to figure out how the RNG works in Golden Axe 1 (on Sega Genesis). Specifically, the patterns the thieves run around the screen. I think the RAM address CB20 is used in some way. If I watch it as a 2 byte unsigned integer it goes up 4 every frame the game is running (not including loading frames). It sometimes randomly goes up 8 on a frame but I could not find a pattern. By freezing this address I was able to reproduce specific patterns the thief moves around in arcade mode at the camp after stage 1. I poked the address with numbers from 0-80 going up 4 each time and found 7 patterns, but I could not see any pattern to it so I'm guessing the game generates a seed based on this address value on a given frame to determine which pattern to use. I tried going up 1 number at a time, the thief doesn't ever load if I use an odd number, and even numbers that are not a division of 4 still work. I took a screen grab of where the thief moved with each address https://imgur.com/a/ZgJ0fHW I also noticed by freezing and unfreezing the address that the direction the thief will move next is generated some time before it moves, and the address is re-used before it moves each time. I was hoping to find a way to manipulate the rng in a way I could use for real time runs but it looks like this won't be possible due to it being bound to time at such a small denominator. For a TAS though, you can pause the game to alter the rng as the counter keeps going even when paused. I would be interested to know other people's thoughts on whether my conclusions are accurate and if there is any more testing I can do? Below are the ram addresses of note: CB20: Game time counter (used for rng) C17A: Possibly another game time counter but it doesn't seem to be linked to seconds like the previous address, I don't think it's used for rng
Experienced Forum User, Published Author, Player (19)
Joined: 7/29/2018
Posts: 5
It is certainly true that Tyris' main benefit in arcade is her stronger magic. However, there are also differences in special attacks/hitboxes. Tyris has a spinning backwards attack for her special. This lends itself better to combos than Gilius' rolling backwards attack. Her hitbox also feels smaller (although I cannot prove it). Ax has the same spinning backwards attack as Tyris. His running attack (shoulder barge) has a hitbox that makes it less useful, although is has the unique ability of being able to hit enemies attacking from behind (if you charge away from an enemy that is about to run+charge you, the enemy will be hit and fall in the direction of your charge). Ax's hitbox feels like it's the biggest (again I cannot confirm). Gilius has a rolling backwards attack. It has different benefits to the spinning backwards attack. It cannot be used as easily in combos due to it's range, but this means it can be used to hit enemies who are off screen at a greater range than any other character. Overall the easier combos with Tyris' spinning attack make her the best choice for me. Most of the time it's possible to keep the enemies on screen which makes Gilius' rolling attack less useful, whereas the frames saved not having to re-position to chain a spinning attack saves more time.
Experienced Forum User, Published Author, Player (19)
Joined: 7/29/2018
Posts: 5
I'm not sure how to update the submission file, perhaps I don't have permissions? I was able to run Spike's lightly spiked version without de-sync http://tasvideos.org/userfiles/download/48797251874350810
Experienced Forum User, Published Author, Player (19)
Joined: 7/29/2018
Posts: 5
I re-downloaded the latest heavily spiked bk2 and I also get de-sync when running in emuhawk. I will try and fix it and see if I can re-upload. It would be great to see this game optimised more, let me know if I can help at all.
Experienced Forum User, Published Author, Player (19)
Joined: 7/29/2018
Posts: 5
Thanks for trimming the ending. I'm a complete beginner, what is a light spike of frames? Also, I posted the you tube video with timing using the same start/end we use for non assisted runs. Which is why the time is shorter than the total duration of the TAS (timer starts when selecting a character, and ends when the last enemy starts their death animation). I didn't think it would become part of this submission but I can change the video if it's important?