Even though I performed the zipping glitch once and kept it all the way, instead of re-colliding the chipmunks each time, the lower route is more lag-prone and I spent more time in the air by doing small jumps to keep chipmunks collided. This TAS is effectively 15 frames slower and my final position is worse because there are 3 levels of crates I need to cross.
Zone O1. Top floor, upper route at the beginning.
I tried taking the upper route once I reach the end of the second vertical pipe. For this, I didn't sacrifice a chipmunk for the other one to use it for a vault and performed a zipping glitch instead.
Unfortunately, you can't jump over a such a big target as a mechanical dog. The only way out is to jump down where Aglar takes the lead.
Aglar manages to keep the route optimized while keeping the chipmunks on the edge of the screen, although it's not appropriate, because, at some point of zipping, the game may randomly push you in the opposite direction or stop you. I don't know why this happens.
Still, using such position, he manipulates luck by making the mouse above the tall block of crates jump away, hence making less creatures on the screen afterwards.
Zone O2
User movie #638525257538743066
No, taking the lower route to have zipping glitch for more time doesn't save time, because the mouse jumps over the crates and hits chipmunks.
User movie #638525258334574985
Also, there is no way to perform one long fall right after executing zipping glitch. One chipmunk lacks one unit of X position and lands on the very edge of the crate instead. If I delay the drop, I get hit by the mouse on top.
Maybe something will be discovered in the future, but I think Aglar has probably found the potential limits:
1. A jump to an upper platform (64 units higher): Link to video.
The way to do this is, one chipmunk does a minor jump, the other chimpunk performs a jump of a specific length (of holding A button) and you can achieve stability by making the collision push both chimpunks straight up, without giving any X momentum to cause resonance. This way you can reach the shown level of height. After this, the game separates the chipmunks on their way down, but here they already land on the upper platform and the state of collision survives.
Also, Aglar's very first jump with zipping preservation is just hilarious. He manages to handle the collision in the way so Chip lifts up in the air without jumping.
2. A 39-units jump-landing while preserving the state of collision: Link to video.
If you check inputs, you'll notice that both jumps with collision preservation and throws to reach the state of collision (zipping glitch) are patternized for different conditions. One step aside makes chipmunks separated.
To summarize, maybe an improvement there exists, but I haven't found it in 6 days of testing and examinations. I think Aglar's TAS is indeed optimized.
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
Chip 'N Dale: Rescue Rangers (NES)
Is it allowed to kill characters using a special menu? You press Select (it shows up the status screen). There, pressing A+B at the same time shows an special option "Give up? No, Yes". If you select "Yes", you can select the character to kill. After that, it disappears, loses a life and respawns as usual.
Here is how speedrun utilizes this trick:
Link to video
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
That was spectacular to watch! Especially that primitive enemy AI which makes them jump in the pit.
Yes vote, and I agree with your opinion to upload TAS as is
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
In scope of TAS movies provision to test emulation accuracy, I'd like to provide two movies (following previously posted Duke Nukem Advance movie):
- Serious Sam Advance: User movie #638512881522132559
- Doom: User movie #638513118358303271
Speaking of Doom. For some reason, GBAHawk doesn't make an in-game save. I don't know if I should create an issue on GitHub or maybe I missed some warning on this regard, but since battery-backed saves were mentioned in previous verifications, I think that means the emulator must be capable of doing saves.
For comparison, here is my BizHawk 2.6.1 TAS, where game is being saved just fine: User movie #74380673068812668
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
I was interested in watching a TAS of Columns 1, despite the game being "infinite". Thank you for this!
I have strong doubts your run is optimized. The beginning looks great, but after some time into the game you stop building complicated combos and pretty much just keep playing as is, hoping for easy column sequences to come.
The stack of columns is predefined. It doesn't depend on RNG or some timer. You could just place 20 blocks, note them down, rewind the movie and plan how to build a good combo.
What's the point of having Player 2? Just to have your moves mirrored?
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
CoolHandMike wrote:
How can we make sure each of them is accurate?
The topic is definitely related to enhancing the possibility to TAS distinct games. My personal opinion: let's value the presence of TASes higher than possible inaccuracies of source ports, re-ports and disassembles.
If some unofficial port of a game is TASed, then OK, why not? Then, an official version of the game is TASed - great, it can definitely obsolete the previous submission or the former TAS can be moved to alternative publications.
What about Bizhawk. Since it already handles multiple emulators, its architecture, I'm sure, is quite complicated already. It makes sense to take the path of GBAHawk then, and have only special "cores" to launch individual games in the way you provide.
Vanilla Doom source ports are already accepted. There are two main source ports to build actual frame-by-frame TAS demos:
- XDRE - for Doom itself and some other "standartized" source ports (compatibilities), like Boom and MBF.
- DSDA-Doom - for Doom, Heretic and Hexen games, and also Boom, MBF and MBF21.
All this is thanks to Doom engine's native way of replay provision which technically never goes out of sync, unless you edit demo header. Vanilla Doom, Heretic and Hexen demos can be played back on real DOS machines and not rely on specific hardware.
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
coolman wrote:
I don't know how interesting it would be, but I would love to see a TAS of The House of the Dead: Overkill. Maybe a maximum score goal would be interesting?
You may find it out in the respective thread. I dropped the work after beating 2 missions out of 7 back in 2018.
Here's also a full game TAS playlist, but not by a tasvideos member.
Maximum score will not be that interesting because the 4th or 5th boss (I don't remember exactly) periodically calls zombies with police shields and this will repeat forever until the boss is dead.
Any% with Extra zombies is the best goal, realistically. Saving civilians is optional. Director's Cut mode would be even better, but it requires a verification movie (beating the game in Story Mode and saving it)
TASing is like making a film: only the best takes are shown in the final movie.
I assume, neither PC nor GBA ports won't have an instruction of how to compile the source?
Also, including .exe files in repositories is not common even for officially freeware games. If you can split the retail files and the source, you'd better keep the retail as an archive.
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
It's disappointing nobody picked this game in 16 years. Well, let's see what I can do.
The whole game looks relatively easy. Just movement optimization and occasional shooting distinct enemies.
Movement nuances
- You switch weapons by holding Select and pressing L and R bar keys. While you do this, you can't strafe. Select, Select + L - that's two render cycles needed to change the weapon.
- You can't jump over stairs railings or over enemies.
- To activate switches or doors, you just have to be close to them. You don't have to face them. Alien switches have kinda weird hitboxes and may not get activated that easily.
- To grab an item, you must be at the same height or above it. You can't bump into items using your momentum, like in Doom.
- When you turn, you increase the camera speed. If you hold Left, then switch it to Right, you turn in the opposite direction at the same speed.
- The framerate doesn't affect the speed.
Speed units
RAM addresses work only in scope of the current level. You have to perform a new search again.
Run forward (RF) = 1568
Run backward (RB) = 874 (55.7%)
Strafe Left (SL) = 868
Strafe Right (SR) = 875
RF-strafing (either left or right) = 1793
RB-SL = 1231
RB-SR = 1237
To have a perfect angle for RF-strafing, you need to turn left/right by 21 degrees away from the direction you want to run towards.
For RB-strafing - by 32 degrees.
Optimization
The framerate is uncapped, which makes it harder to optimize the movement because you have to redo the input every time you try a different way of movement and luck manipulation. You can't just copy or delete unnecessary frames. You gotta follow for how many non-lagged frames you kept each button pressed before doing progress in a different way, and it still doesn't guarantee you will do the same what you used to do.
The enemies often stand dead on their tracks when you see them. I think it has something to do with performance optimization to spend less time on AI logic.
Enemies
The following list is determined by doing RAW Search. GBAHawk, WRAM memory domain, unsigned int. HP cannot be negative (if an enemy is dead, its HP is zero)
Enemy
Health (HP)
Assault General (Trooper)
40
Pig Cop
60
Hybrid Gray Alien
50
Spiderlord (boss 1, 02BF18, 2 bytes)
1100
Octabrain
100
Enforcer
120
Cyber Alien (boss 3, 0194C8, 2 bytes)
2000
Alien Egg
50
Alien Crab
80
Alien Lord (boss 4, 027D78, 2 bytes)
5000
It's not clear how much health does Boss 2 actually have. First, you need to hit it 9 times to get a call from the general. Then, you must make the boss walk under the crusher and it dies instantly. The crusher deals 4 damage. One smash cycle takes 64 hp away.
Weapons
Weapon
Damage
Mighty Foot
40
Desert Eagle (Pistol)
12
Shotgun
50 (5 pallets, each one deals 10 damage)
Lead Cannon
147 (7 pallets, each one deals 21 damage)
MP5
10
RPG
200 (80 = direct hit, 0...120 = splash damage)
Pipe Bombs
0...255 = splash damage
Freezethrower
0
Shrink Ray
0
Boss 1 rocket
0...40 = splash damage
Shrink Ray and Freezethrower cannot do anything to bosses.
Here is a TAS test which passes first 2 levels (1A and 1B). The encode is included there.
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
I recorded a small TAS of Duke Nukem Advance: User movie #638491602391815310
The game features unusual usage of hardware. The sound chip participates in calculation and doesn't play music during the gameplay. Besides, the framerate is uncapped.
So I think it will be a good accuracy test for GBAHawk, if you have the game around.
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
PurpleSun wrote:
Unfortunately not, as it turns out it was done on a bad dump. Someone had asked him on one of his WIP posts (https://youtu.be/hVRbGtyb4rk?feature=shared), but he said that he only found out the game was a bad dump when it crashed on Lanayru Twilight. Trying the inputs on a good dump didn't get past the title screen.
A fallen star, sitting next to Metroid Prime submission =(
TASing is like making a film: only the best takes are shown in the final movie.
Experienced Forum User, Published Author, Active player
(485)
Joined: 4/5/2014
Posts: 1139
Location: Ukraine
Hello, ded! Nice to meet you!
Unfortunately, our community doesn't have much enthusiastic people who would be glad to do this kind of stuff. Also, in your case, it doesn't sound like an improvement to Fly's TAS.
Obviously, one may tell you to learn how to use frame advance. It's an extra button you have to press, while holding the input, to register it and jump to the next frame. It's like playing on 1% speed, but frame advance lets you control the tempo easier.
If you still don't feel like doing that, you can simplify the task, if you provide ROM hash, emulator name, BIOS versions (if it's PSX version of the game), a list of alleged combos and input movie files where Sub-Zero reaches regular bosses. This way, the person, who agrees to help you, will spend the least time on preparation.
TASing is like making a film: only the best takes are shown in the final movie.
