Posted: 4/30/2023 12:56 PM
(Edited: 4/30/2023 9:11 PM)
Quote
Dimon12321
He/Him
Active player (480)
Joined: 4/5/2014
Posts: 1128
Location: Ukraine
I'd like to tell you about my failures in TASing this game. After all, negative experience is also experience. Settings: - Speed Units = km/h - for better comparison with speedruns - Transmission = Manual. Automatic one switches gears too early. As a result, you car drives slower. - Stability control (SC) = Off. When SC is On, your car is more sticky to the ground which affect both your throttling (speed gaining) and steering a little. As a result, your car moves slower by 1-2%. When SC is Off, the steering a little more slippery, especially when you drive over sidewalks/pavements. I feel like all it does is tweak the gravity for your car. Known speedrun tricks: Double steering (DS) - https://www.youtube.com/watch?v=Q0bKpCz7GOk Double tapping (DT) - https://www.youtube.com/watch?v=lMGFfoUwtaQ Both techniques are used to make the car enter the drifting without major speed looses, which results in covering the turning sections much faster. Why I dropped it so quick? I have been meshing input for 3 hours to perform a trick to cover the first turn of the Tutorial Race (the very 1st race in the game) faster and didn't get any luck. I am shocked how easily speedruns utilize these tricks!! I even downloaded one of the speedrun videos, copy-pasted its input frame-by-frame and still failed to perform DT. No luck with DS too. I have major doubts that even if you manage to perform any of them, it won't be easy to adapt it to your case to complete your section faster. I can assume stable 60 FPS of PC version helps doing that or maybe some plugins I'm not aware of. GC runs at 25-50 FPS depending on the render state, but I'm sure it still should be possible. Global TAS problems - If for most games RTA timings can be relatively easily beaten by slow-motion runs, let alone frame-by-frame building, NFS: Underground is one of the hardest ones to do so. It's gonna be cumbersome to achieve "Beats all known RTA records" goal. - The game is very long and tedious. Any person, except for some special hyper-patient individual, will go nuts spending years to compose an acceptable 4 hours of input to beat the game. - Even if TAS bots become capable of Dolphin input bruteforcing, they will have to an extremely big amount of work, despite the fact that input variations can be narrowed down very much. P.S: yeah, no input file this time. I made first 10 secs of the 1st race, but it shows nothing useful, since I didn't manage to perform a trick at least somehow.
TASing is like making a film: only the best takes are shown in the final movie.
Posted: 4/30/2023 8:00 PM
Quote
andypanther
Any
Former player
Joined: 6/30/2010
Posts: 1094
Location: Zurich, Switzerland
If you ever hope to get a TAS done for a game like this, you just can't aim for perfection or you'll never get it done. Set yourself the goal of beating or at least tying all known records for each race, which is already a challenging goal itself. If you find out that the version of the game that is available for TASing is at a disadvantage against the version used by a specific record, it is fine to not beat that time. Racing games are one of the most difficult types of games for TASing, that's just how it is, so make sure to tackle a racing TAS with the right mindset.
Current project: Gex 3 any% Paused: Gex 64 any% There are no N64 emulators. Just SM64 emulators with hacky support for all the other games.
Posted: 5/5/2023 1:22 AM
Quote
FitterSpace
Active player (416)
Joined: 3/30/2012
Posts: 404
Rookie of 2017 Gamecube/Wii TASer of 2017 Gamecube/Wii TAS of 2017
Andypanther is totally right. You can't be too hard on yourself about trying to TAS a game like this. Racing games are very difficult to optimize, so you can't aim for perfection if you want to get anything done. In my opinion, I think it would be best for you to give it your best shot at a few time trial TASes so you can familiarize yourself with the game. Don't worry too much about beating existing records. Instead, focus on completing some runs so you can see where you're able to improve them. I think it's much easier to learn how to improve upon a TAS that already exists than it is to make one from scratch. In this case, you'd be the one making the TAS you plan to improve later. When you do that, you'll see things you didn't notice during the making of the TAS that will seem obvious in hindsight. If you keep doing that, you'll improve quickly and eventually be able to make better TASes with less effort. In short, don't spend too much time trying to optimize individual corners. I think it's much better to TAS at a reasonable pace so you can finish what you start. Do some time trials, get more familiar with the game's mechanics, and learn from the mistakes that you made in your previous attempts. Also, be sure to talk to the game's community if it's still around. I'm sure the competitive community has a discord server somewhere if the game has any activity. They could also know about some obscure version differences that might help you.
Finished projects: 007: Nightfire 007: Nightfire "All Tokens" Goldeneye: 007 (Wii) Feng Shen Bang (Famicom)