I'd consider removing ACE from the movie goals since no arbitrary code execution was used in this run.
I have one suggestion about editing the explanation of the item storage glitch, the way it was explained by the author is very confusing:
Having this in place, and joining the track Dingo Canyon (first in the array of the game), the game goes out of bounds with LEV_ID being negative due to the operation the game itself does to switch from common trophy races to crystal challenges, which usually happen in Battle Arenas (like Skull Rock).
What actually happens is that when you have the battle mode end of race code running, the game makes a copy of an array to the stack and uses it to award the player. The catch is: the index number of race levels are less than the index number of battle levels, so when calculating the reward with this glitch activated, the game will access variables out of bounds in the stack. Luckily for Dingo Canyon, the variable it accesses is one of your camera values, so it ends up being possible to manipulate a reasonable range of the memory, including some variables about the game completion status (e.g items, highscores).
I have a consideration to make about the category description. Quoting the author:
This is CTR: Crash Team Racing Any%. This category uses the Battle Mode Warp Glitch but not to jump directly to the credits, since that method does NOT have oxide beaten in the game ram, but its just a cutscene warp. This instead is the fastest completion of the game by having the final boss beaten in the game ram. All skips are allowed, the goal is to beat """Oxide""".
This is not the fastest way to watch the cutscene while having Oxide beat in RAM. As you can check in the item table, it's possible to manipulate the Oxide star (08FBB0 & 0x4) and then warp to the credits. This brings the discussion of what truly means to beat the game, and what state of the game should be considered to award completion. If you consider the oxide star as beating the game, then one item manipulation in Dingo Canyon is enough. If you consider star + cutscene, then you can obtain the star via item manipulation and warp to the cutscene. If you just consider the final state the ending cutscene, then warping to the credits is enough. I personally believe that obtaining the star via item manipulation (just like the keys are obtained) is a valid way to skip the final boss, since the game gives you all awards from doing so.
I believe the author probably considered the idea of reaching the final state of the game something else (otherwise the strategy used for this run would be suboptimal), thus re-writing the rules of the category may be necessary.
Final considerations:
Hub 1 was very well played, the gameplay is entertaining and optimized. The item storage glitch itself is a boring glitch to watch, but it's a quite interesting exploit which may peak people's interest to read about in the run's description.
EDIT: clear description of the item manipulation glitch and final considerations
I think this run accomplishes what it was supposed to. It's a good base run, the TASer does some cool tricks to keep the movie entertaining through levels that are auto-scrollers, and it beats each level with a good amount of optimization for a first TAS. My vote is a yes.
However, I believe that TASing goes way beyond trial and error with save states and frame advance. This run lacks researching the technicals of the game in order to exploit better some glitches, manipulate the A.I in better ways and even find more exploits if there are any left. I do understand that researching a game is a time sink, and for a complex project like this it could exponentially increase the development time (which was already a lot), so I understand why the TASer opted to go for a less optimization route.
Here are some points that I'd like to make about improvements that could be done for a next run:
1) You can manipulate the ball in ballistic levels to always spawn in a side that you can shoot, and you can manipulate A.Is to get out of the way of the ball, so they wouldn't fight for the ball.
2) Dash lap skips. In Dash levels, you get a lap to count every time your old Y value is greater than the finish line Y value, and your current Y value is less than it. In a frame, the game first calculates A.I collision, then it processes the lap count logic, and then it updates your position using your velocity. This 2 step calculation can cause the game to update your Y value of a frame earlier to a value before the finish line, and if your current Y value is after the finish line, then the lap will count (since it will only correct your old Y value after the finish line evaluation function completes). I have done a script where I inject ASM in order to read what is the mismatched Y value, and see if you can make a lap skip. You can find it here.
While lap skips were exploited in his run, he mainly opted for doing so by using missiles or hitting A.Is with juiced wumpa speed. It's possible to make lap skips happen without it, therefore he could've abused it more in the run.
3) General lack of optimized A.I manipulation, some rounds of the same minigame are a few seconds apart and I suspect they could be better if the A.I movement was researched.
I haven't done further research in this game to look for more tricks that could be used in a TAS, that's all the input I can provide for now.
Kianotti, acredito estar acontecendo um erro de compreensão aqui. Primeiramente, queria deixar claro que ninguem mostrou o desejo de te banir do site, ou de você parar de fazer TASes. Tambem, o Spike não odeia o Snes9x. O que aconteceu foi:
Como o Spike, o Bobo, e outras pessoas postaram anteriormente em inglês:
Apesar de não ser um emulador recomendável, o Snex9x ainda é aceito. Sua versão v1.43 ja é obsoleta e não mais aceita para TASes novas, a exceção vai para TASes antigos que já estavam em um grau de conclusão elevado. Desta forma, se você for escolher usar o snex9x, utilize a versão v1.51 ou versões mais novas que ela. Adicionalmente, o Spike deixou claro que este site é para TASes. Uma TAS não é somente um movie feito em um emulador com função de re-recording. Na verdade, uma TAS é um gameplay criado nestes emuladores com o objetivo de otimizar o jogo ao máximo, utilizando diversas estratégias que podem ser encontradas no site da TASvideos e nos fóruns de discussões - com o objetivo de fazer um super gameplay próximo da perfeição que um humano sozinho com seu video game não conseguiria fazer. Ou seja, mesmo que uma TAS não seja perfeita (o que é muito difícil de chegar a perfeição completa na maioria dos jogos), espera-se que a TAS pelo menos bata o recorde humano, uma vez que disponibiliza-se de diversas tools para esta tarefa (que um humano não tem). Também, em categorias que existam uma TAS no site, a nova TAS apenas será aceita se bater o recorde da TAS atual. Assim, se você continuar enviando movies para o site que não enquadram-se no que uma TAS realmente é (apesar de ter sido feita em um emulador para TAS), estas serão rejeitadas. Uma forma de aprender a fazer TASes é olhando os tutoriais do site, e aprendendo antes de tentar fazer algo. Várias pessoas aqui já te mandaram vários links com a intenção de te ajudar, seria bom você ir atras deles neste momento ao invés de continuar postando movies que não são relacionados ao objetivo do site.
Eu, Redhotbr, complemento dizendo que não existe problemas se você não sabia destas informações. Errar é humano, apenas não persista no erro. Caso contrário, você estará gastando o tempo de juízes, editores, postadores do TASvideos para nada, e então serão tomadas medidas para que isto não ocorra. Também, o Bobo pediu que você usasse o botão de editar ao invés de criar múltiplos posts sem alguma informação. De modo geral, eu aconselho você a se familiarizar melhor com as ferramentas do site, para que as suas mensagens que tens a transmitir contenham informações úteis, sem qualquer tipo de spam ou flood. Desta forma, você conseguirá usufruir deste site sem nenhum problema. Lembrando também que spam/flood são cabíveis de punição dentro da TASvideos.
Honestly playing a slower variant of any% is very arbitrary, as it's slower than the fastest any%. So, I only see the point of doing a slower any% for entertainment purposes, unlikely this TAS that uses zipzagging which makes most of the run be extremely boring to watch. Your any% got my "YES "vote, unfotunatelly this one gets a "meh".
Talking about optimizations: as Spikestuff has pointed out, this TAS isn't fully optimized, but overall it's pretty good. Your work is truly respectful:all the effort that you've put into this games certainly will help more TASers and speedrunners to lower their times. I'm sure that a non-zipzag version of this would be a lot better to watch.
Olá. Mupen não é mais aceito para fazer TAS. E também, ele é um emulador horrível para sincronizar o movie.
O único emulador que hoje é aceito pela TASvideos para fazer TAS deste console é o BizHawk, então sugiro migrar para ele. Aqui esta o link para download, na própria página deste site: http://tasvideos.org/BizHawk/ReleaseHistory.html
Recently HP1PS1 has got a new major skip that saves around 10 minutes. I myself kept trying to figure out new stuff, and I found this controversial skip: http://pastebin.com/DJt3edBF
So I would like to know the answers for these two questions in the pastebin before anyone start TASing this game. Also, we need to answer these questions in a RTA way, cause this skip should save around 3~4 minutes over the current route.
EDIT: SRcom forum discussion: http://www.speedrun.com/hp1ps1/thread/xhwl0
Recently I started TASing Crash 1 100% in NTSC-U using BizHawk 1.11.6. I completed the first level and now I'm working at Jungle Rollers.
Link to video
I'd really appreciate if anyone make me a good lua script with useful values to run in NTSC-U BizHawk.
I just finished a Hot Air Skyway lap TAS done on BizHawk. I managed to beat Zah666's former record by 0'00''22 and finally achieved the sub 22''50! Video link: https://youtu.be/ael5BueKl28
I have also created a topic on speedrun.com listing the TAS records for this game, so make sure to check out: http://www.speedrun.com/ctr/thread/xpsd7
It seems to be possible pass by the doors in the adventure mod without having it keys, at least the speedrunner Pwulla did it in the hub 3 to 4. I used BizHawk to cheat and do the same having every trophie before Hub 4, and this is what happened:
-> Races are unlocked, this means you can beat every hub 4 race and get 16 trophies.
-> After beating Oxide Station, Pinstripe will talk to you and you can fight he.
-> After defeating Pinstripe, you will trigger the Oxide cutscene and warp to hub 2.
-> Unfortunatelly Oxide door will still be locked, meaning that you still need to get every key to beat the game.
So, the only option to have a huge skip in CTR any% is get through Oxide's door (if it's possible, we might skip every race and just drive to Oxide race) or get through Pinstripe's door for an early key, so we can skip N. Gin Labs, Cortex Castle, Hot Air Skyway and Oxide Station.
If we can do Oxide ==> GG.
else ==> Try Pinstripe.
If we can go through Pinstripe ==> Check Komodo Joe.
If we can go through Komodo Joe ==> Check Papu.
If we can go through Papu ==> Check Ripper Roo.
Special thanks to Karlie for helping me testing it and creating the Joe skip theory.
EDIT: you can't really skip the door opening cutscene by passing it with 3 keys, the cutscene trigger is just too huge, also you can't really go much to the left to get the trick (pretty sure you need to hit the wall way more to the right spot than I did in this gif) B(. Here is the gif proofing it (w/ cheats): http://giphy.com/gifs/xT0BKKGsYH9TK8cCbu
I have done some TAS tests to proove that Tiny Arena shortcut is possible with Tiny/Dingo and Coco/N.Gin. I didn't try to optimize anything, it is just a proof:
Tiny Tiger: https://youtu.be/M-UjI9mYq0I
Coco Bandicoot: https://youtu.be/_1jezJa4p1o
I hope it helps anyone who is TASing this game or wants to TAS it. Those videos can also help as a base for the shortcut; however, it can be optimized much more than this especially by getting a SG earlier and froggying until the SC.
By the way here is a special, so don't waste your time trying it: https://youtu.be/fEDZOWjkDHw
Hey! I really enjoy speedrunning some PS1/PS2 games (mainly Crash Bandicoot ones). It has been a so fun hobbie and I'll definately keep going. You can check my speedrun times in my SRcom profile . For ILs/Time trials, I use ctr4ever and cyberscore. I also upload most of my runs in my youtube channel and stream live at twitch channel. I hold the WR time for Crash Twinsanity any% nowadays and my main plan is take the Crash Bandicoot 1 100% WR that I utterly hope I'll suceed :D
I am a speedrunner of this game and I have made some 'TAS' of time trials back in the PSXjin days. Here are the video links:
- https://www.youtube.com/watch?v=fCKFo-uVLOs - Hot Air Skyway lap in 22''67.
- https://www.youtube.com/watch?v=S43yTMdndpI - Roo's Tubes lap in 21''28 (begin the lap with SG might be faster than what I did in the very beggining of the video).
- https://www.youtube.com/watch?v=kBUWhOXV5Ns - Coco Park lap in 23''30. First one I made and it was really bad, can be improved by a lot (I didn't beat the rta wr in this 'tas' lol). I did it more like a test run to be honest.
I would be glad if it be helpful to anyone. Unfortunatelly I am not good at anything related to TASing; however, I got some interest in TASes that are making me want to go further and learn how to TAS properly.
If it's not going to have significant impact on run gameplay, then PAL won't be accepted. Skipping a few dialogues doesn't sound as nearly enough of a difference, and a shorter ending doesn't seem like a relevant difference.
Warepire wrote:
If PAL is faster, then it definitely qualifies as a reason to use PAL.
Just being faster is not enough reason to use PAL. Even PAL Super Mario Bros. is faster than NTSC due to broader application of the flagpole glitch, and that is not going to replace the NTSC standard either.
The PAL version of the game also skips a few seconds on each aku aku speach, because the language glitch manage to show aku aku soon after you finish the race, while with any other language you have to wait 2~3s. By what is knowing on the speedrun community, NTSC-U is around 20s slower than PAL on any%, that is a huge difference; however, PAL version crashes after beat Oxide on any% with language glitch (hardlock). Just pointing it out to moderators see it and decide whatever they want.
I would test what is the fastest character to go, and use the fast one. I believe Coco is faster than Dingo because she can take some shortcuts that dingo can't (or he waste too much time doing it), such as Polar Pass, [Tiny Arena?], Hot Air Skyway and Oxide Station. About the category: I myself would prefer an any%; however, feel free to TAS whatever you want =]
