Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
TASeditor wrote:
Spikestuff, these don't fit into the criterea. They were still done with rerecording, an AI doesn't have that.
He/she/it does have but most of the articles I saw about this project is either exaggaration or bullshit. Youtube videos also doesn't showcase any intelligent AI moments (yes they can control a character, wow, much skills)
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
feos wrote:
MESHUGGAH: See? People can miss things.
It took me nearly a day to realize this couldn't be more ironic. (judging you for something that's not even part of your role: fallin in to this mistake myself here by not looking for the informations (Arc's 2010 improvement post) of this game's thread; not taking this submission as seriously as I did with other TASes)
I'm planning to make another stab on this game from the beginning probably soon. This will be a good reason for me to finally stop doing level 1 WIPs and also for finally making a newbie HUD template (which will probably use this game as an example).
edit: like this. Unfortunately still requires some LUA scripting and programming knowledge but with some tweaking you should be able to use it.
Language: lua
--[[ MESHUGGAH's HUD Template
This template features an example to NES Rush'n Attack
v contains RAM addresses of position values (hexadecimal)
p positon values stored here after reading memory from location v (memory.readbyte)
pp previous position values stored here to compare with new positions on next frame
--]]
-- Variable array: P1 X, P1 Y, P2 X, P2 Y, Camera X
local v = {
0x503,
0x505,
0x523,
0x525,
0x31
}
-- data structure for rush'n attack, 32 bytes per object
-- question mark notes boolean
-- 0x500 alive?, state, 02, X, subX, Y, subY, 08, 09, 0A, 0B, 0C, 0D, palette, sprite
-- 0x510 in jump?, 02, 03, 04, 05, 06, 07, 08, 09, 0A, 0B, 0C, 0D, 0E, sprite timer
-- Position array
local p = {} for i = 1, table.getn(v) do p[i] = 0 end
-- Previous position array
local pp = {} for k,v in pairs(p) do pp[k] = v end
-----------------------------------------------------------------------
local function objects()
-- display ID if alive
for i = 0, 15 do
if memory.readbyte(0x500+i*32) ~= 0 then
gui.text(memory.readbyte(v[1]+i*32),memory.readbyte(v[2]+i*32)+40,i,"#FFFFFFFF","#FF000080")
end
end
--gui.text(1,9,
end
local function hud()
-- Display object's ID and a simple box around them (not actual hitbox positions)
objects()
-- Read new positions
for i = 1, table.getn(v) do p[i] = memory.readbyte(v[i]) end
-- Anything else you would like to display?
gui.text(1,1,string.format("CamX %3d (%d)",p[5],pp[5]-p[5]),"#FFFFFFFF","#00000080")
-- Display positions for players
gui.text(p[1],p[2]+48,string.format("X %3d (%d)\nY %d (%d)",p[1],pp[1]-p[1],p[2],pp[2]-p[2]),"#FFFFFFFF","#00000080")
gui.text(p[3],p[4]+48,string.format("X %3d (%d)\nY %d (%d)",p[3],pp[3]-p[3],p[4],pp[4]-p[4]),"#FFFFFFFF","#00000080")
-- Store positions to compare them in next frame
for k,v in pairs(p) do pp[k] = v end
end
emu.registerafter(hud)
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
Even the fact that I can go game over is new to me. I will check it out. I only tried to minimalize the position sacrifices in the levels (as I stated in the submission text).
edit: I know that if someone dies, the camera doesn't rolls forward. So you mean you somehow do game over fast enough to while keep it rolling by the time Red finishes climbing the ladder?
edit2: ahaaa, so getting game over doesn't stops the camera. Yes, that would save 12 + blue sacrifices frames (my raw guess is ~50 frames total).
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
Published TAS F1539, they sacrificed 2px to jump over the green monster and have feasible position aftewards. If you manipulate DW's subpos (0x460 I guess), you can step 1px left before taking damage from the first bullet, changing the X speed pattern to get the arrow and sacrifice 1px right before the green monster (F1588)
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
If you aren't familiar with this game, check out it's bootleg page.
There's a very simple glitch that allows you to deal 100% dmg to enemies. A TAS would do this something 24 times.
Here's a video of 2 fights between 1P and 2P and a fight against CPU.
Link to video
Main intention of creating this post is getting feedback about would you watch a TAS of this (since optimizing the glitch to be the fastest multiple times requires some time). So... ?
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I've semi dived in to this topic and checked a few things.
I don't see any benefits of tasing through a web browser. Answering OP's questions:
1. Yes, you can compile them and run it in asm.js but not with a single click, some things requires modification (or usage of emscripten plugins).
2. Totally depends on user's perspective. A program won't work faster/better/easier just because you can open it in your browser.
3. There are many ways to save data but AFAIK the most convenient one is using an online (cloud) storage. The program won't have access to the file system, so you either need to use HTML5 local storage, chrome indexeddb or the classical "save to notepad".
Also note that for android devices the performance is... bad. So far.
edit: "I remember we alway had some sort of issue when trying to TAS through a virtual machine." - those problems are about emulating specific virtual machines (kvm for j2me). asm.js/emscripten has no impact on this.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
3 "difficulties" (more like game modes):
Normal: clock power up to get more time (this TAS)
Expert: same without the clock power up
Land of Waldos: the final level of this TAS, you can play that stage as long as you still have time.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I've finished the first one, result is "Älykkyysosamääräsi on noin 130 , joka on korkeampi kuin 97.7 prosentilla väestöstä".
My Mensa HungarIQa test result (offline/IRL) is 130~135, 98% better than others. These online tests used to avoid very complex logic patterns that I saw at offline tests.
My answers:
1 3
2 6
3 1
4 3
5 1
6 1
7 5
8 2
9 6
10 4
11 2
12 3
13 1
14 5
15 6
16 5
17 4
18 5
19 5
20 1
21 4
22 1 ???
23 2
24 3
25 5 ???
26 1
27 4
28 5
29 3
30 6 ???
31 1
32 3 ???
33 6 ???
34 3
35 out of time lol
edit: regarding IQ... well I used to say it's about understanding patterns by using a set of rule(s) to apply that logic. I also know that official IQ test questions (the very hard ones) have multiple acceptable answer (the exact number wasn't shared with me).
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I did some comparison with and without Scumtron's half pixel save and looks like it's preservable with additional saves if I would be lucky.
* 4-1 bitch (slot 7 knife thrower) now throws a more feasible knife to pass below it and above the bitch. (maximum save is 7px)
* slot 5 dog needs to hit the red block
* slot 7 gunner must shoot 3 to the right side otherwise a bullet hits us later
* slot 6 guy subpos to hit the platform after above mentioned dog manipulated
* slot 6 final guy (mostly because interrupt place) is at the worst possible place, unable to jump over the platform while despawning him with juggling.
If everything gets in sync, probably 2~3 frames will be saved.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I could write a lengthy post but I will stick to short sentences that I think you should think about (not refering to anything in particular since I don't know the full story)
I hope you will be a judge again in the future.
1. Always take your time to make sure you do what you are supposed to
2. Avoid encumbering yourself if it impacts your overall performance
3. People are retarded, you should try to outshine them whenever you can
4. Resolving problems requires discussion
5. Only pussies giving up, universal guys never.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
A fair note, the TAS aims for shortest input time and doesn't tries to reduce the lag frames at the end of the final level (the sprite spam for the destroyed rocket), so the credits only starts 13 frames faster.
Those lag frames appear ~10 seconds after last input.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
All the informations above are known already and correct.
edit: You guys should really learn how to use search and where to find those informations you want to know instead of trying to reach me with a forum post...
Here's the page you probably missing: Wiki: GameResources/NES/AdventuresOfRockyAndBullwinkleAndFriends
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I want to TAS a few FDS (famicom) games but I couldn't find how to remove left-overs / save files from the rom itself.
The movie rules says I should make a verification movie to show it syncs from a clear save, which would mean I need to start the movie from power on, delete the save files, create new saves (depending on player count) and then start it.
For whatever arbitrary reason, I got vastly different results. For example in Dirty Pair, I'm forced to wait an extra 500~1500 frames that I could easily skip when using the left-over save file (which has absolutely no affect on starting the game / playing the game apart from score points / no different cutscene).
Question is, how do I magically hex edit it out of the FDS rom?
The other question is, does the verification movie is allowed to have way much longer frame delays before starting the game? Because now it doesn't looks like the way I eject and insert disks is "verified" this way.
edit: forgot to say that I can't even save the rom with clear save files, it loads back the left-over data (using FCEUX 2.2.2)
edit2: with help of adelikat, here's my Matou no Houkai "cleaning" attempt (not approved method?):
1. start the ROM
2. use the game's own menu to delete it
3. launch the game to save the SRAM part (can't verify if that's the only thing it does)
4. start a fm2 movie with bit 2 on the very first frame to use "power-on" otherwise it uses the left-over one.
edit3: for those who don't follow IRC, rules dictates you are supposed to have a clean FDS rom with no save file made by someone else in order to submit a "valid" movie file. "poopy" situation.
edit4: wrote down some stuffs about saves here: Wiki: GameResources/NES/DirtyPairProjectEden
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
The only way to TAS this game was frame-by-frame trial and error
Well first of all, congratulations on proving sonic is slow as phuck, but I don't understand the quoted sentence (you already know the distance you can travel each frame) nor the problem behind the "impossible turns" (example is an L turn at a corner?). So what did you do when you can't turn "instantly"?
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I watched your TAS input to see how the game works and what you do and then made a new movie and tried various things (in TAS Editor, I draw routes with mouse) while keeping speed on maximum.
The music disabling trick is done by pausing at the end of the level. The jumping doesn't delays the ending despite the position loss (which seems to be because of different memory address used for in-air calculations).
Also I never use entertainment tricks if it costs 1 or more frames. This costed 0 frames and can be done without jumping and right after the first stage. But I'm not that evil.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I thought it worth to mention that this TAS was already done probably in 2014 but forgot to submit it (rewatched it by frames to make sure it's done).
And I would also warn the encoder about the visual anomalies on the City part (3:47). I'm not really sure but if it would be possible to decrease the high contrast in the background, it would be not that much hurtful to the viewers.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
edit: I'm going to update this post each time I modified something.
List of improvements added to V7 http://pastebin.com/Zn2xWp87
Timesheet http://docs.google.com/spreadsheets/d/1BkA607KZFXguP2Ps1rvPsqbfyxNQXY_3QJyvJTS8G88/edit?usp=sharing
Version 7 User movie #20597948781701404 (65)
Version 6 User movie #20554752230643976 (59 @ mission6 start)
Version 5 User movie #20551259081411075 (56 @ mission6 start)
Version 4 User movie #20539447403495593 (53, desyncs because of a bug at mission7)
Version 3 User movie #20515682161543102 (54, desyncs at F28357 M4R4, would be 49 if it would synced)
Version 2 User movie #20513288744457209 (22)
Version 1 User movie #20511612708416611 (11)
The input parts of the movie (V7):
- MMM route of Mission 1+2 (Mission 2 contain's AAA's speed cycle, doesn't changes framecount)
- MMM+AAA route of Mission 3 with all AAA improvements
- Team 2/4/6/7/10/11's jump @ Mission 3 Room 4 ~F21300 (list of teams who did this 11 frame improvement over AAA and MMM)
- AAA Mission 4 route with MMM improvements (many frames)
- SAC Mission 5 Room 1-2 with improvements (2 frames, they had different weapon so isn't comparable directly)
- TTT Mission 5 Room 3-4 with better speed cycles (5 frames)
- AAA Mission 6 with MMM's Boss (4 frames)
- AAA Mission 7
The Mission 4 Room 4 "5 frame loss" has been reduced to 2~3 frames.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
Just don't tell me all you do is read the submission text and watch the video encode. I'm not pointing on anyone, but what other submission should have the same case when things were already knew but nobody tried to make a single step forward to discover a game.
I didn't watched TASEditor's tas, but so far my wip looks much better than other "probably moon" TASes. Maybe just because of my "entertainment" taste (those nasty bomb glitch boosts are impossible)
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I expect judgers to take submission(s in general) and verify if they meets all the rules we have, containing it's technical value.
So I expect the judger to research the game (a simple google search) or refer to other sources (seriously, my page is hosted no tasvideos so you probably will get it very highly ranked if you would have searched for it, it's not my fault that it's still on that particular page as I don't have right to move it).
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Experienced Forum User, Published Author, Skilled player
(1892)
Joined: 11/14/2009
Posts: 1349
Location: 𝔐𝔞𝔤𝑦𝔞𝔯
I had no laptop for a week and this happens?
feos, did you checked the tas uses all the already researched and documented tricks?
answer: no, TASEditor probably didn't read my newer version of my gameresource page that I already asked to move to its proper place several times on IRC with no answer. Here you go: http://tasvideos.org/NES/AdventuresOfRockyAndBullwinkleAndFriends.html
I will submit my movie when I finish it, which is 16 frames faster after 1st screen, 40 frames faster after 2nd screen and more frames on next screen because of better route and optimal bomb glitching.
What happened with investigating TASes regarding it's technical value...
edit: google result for the game title brings my game resource page as the 4th entry.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
