This is the second game developed by Howard Scott Warshaw. In Raiders, Indy needs to collect several items on his way to collect the lost ark.
Starting the Game
On boot, the game starts its own internal stopwatch. This is responsible for spawning certain items in the treasure room at any given time as well as making the sun shine in the map room.
The following shows treasure room items relative to the position on the timepiece:
N - hourglass
NE - ankh
E - chai
SE - hourglas
S - ankh
SW - hourglass
W - chai
NW - ankh
For every 2050 frames that pass, not only does the headpiece spawn in one of the Marketplace baskets for 126 frames, but the timepiece also changes positions.
The 00x0C address (00x8C in Stella) stores the ark’s location. For every frame that Indy stands on the forklift elevator, the hex value changes. When the player sends an input to one of the controllers, the value stays frozen throughout the game. The ark location system can be thought of like a slot machine with a predetermined pattern.
This TAS was created in pursuit of the fastest possible time starting on first input. Unlike previous runs, we save time waiting for certain items when we're at the opening screen rather than during gameplay. I need to wait for the chai to spawn, so to find the best start time I multiply 2050 by 2 (ticks on the timepiece), then subtract that by the shortest length of time it takes for Indy to walk to the closest headpiece basket (188 frames), resulting in a start time of 3912 frames. At 3912, the ark will be located at the 11th mesa (00x0C hex value 08). For reasons I will touch on shortly, I want for the ark to be at the 18th mesa, instead (hex 0F). According to the RAM address ark location diagram below, we need to wait 7 more frames (frame 3919).
We should be good collecting the headpiece since we’re still well within the 126 frame window before it despawns. In total, we don't move for approximately 1 minute and 5 seconds.
Mid-Game
I collect the whip, headpiece, gold, and chai. While walking to and from the Temple Entrance and Room of the Shining Light, I perform some small tech to save a few frames. Then, I bribe the Black Sheik with the chai to teleport to the Black Market. It’s impossible to avoid death from the Raving Lunatic without bribing him with another set of gold.
The Glitch
Now it’s time for more technical info!!!
The hex values and mesa locations for address 00x0B (00x8B in Stella) remain the same as 00x0C. 00x0B changes when Indy teleports to different mesas using the hourglass or ankh. When leaving the mesa field or other rooms, its value resets to 00 (the 10th mesa). The ark will never be located here during normal gameplay. The only condition for winning is that Indy must dig into the Well of Souls located under the same mesa assigned during the player’s first registered input on the controller (when 00x0B matches 00x0C).
The game only executes inventory management inputs (item selection, dropping, and usage) on every other frame. In the Room of the Shining Light, when Indy uses the whip on one of the dungeon walls and then again after exactly 4 frames (2 framerules), the game immediately sends him falling off the 10th mesa. Though unusual, I didn’t think this bug could be taken advantage of to complete the game faster because it doesn't allow you to start on hex 00. One week later, a breakthrough occurred. I discovered that if this bug is performed on virtually any wall or interactive element, such as parts of the nest in the Spider Room (excluding both dungeons in the Room of the Shining Light), Indy gets sucked into the wall, and his whip transforms into a grappling hook as if standing directly in the mesa field. Not only that, but the value of 00x0B starts changing!
Execution and Completing the Game
After collecting the shovel, I travel back to the Marketplace, purchase a parachute, then walk to the left wall to perform the glitch. The game thinks I’m standing on the 18th mesa (0F). Then I teleport off, deploy the parachute, and finish the game like normal. We bypass the ankh and mesa field completely!
From when Indy first starts moving to when he finds the Ark, it takes 2605 frames or 00:43.473.
Extra Notes
Buying the parachute at the very start of the game, and collecting a third piece of gold, just to skip the death animation triggered by the Raving Lunatic and avoiding the trip back to the Marketplace does sound like a good way to save some more time, but it is 24 frames slower.
eien86: I am delaying the judgement of this movie in wait for the upcoming changes to PG rules and possibly to supporting alternative methods of timing. I kindly request the author to be patient until we can resolve these situations.
eien86: Much has been discussed and analyzed about this submission so I am not going to add much further. This submission beats the current RTA WR by a huge margin thanks to manipulating the initial game RNG by waiting in the title screen after reboot. This wait makes it much slower than [3559] A2600 Raiders of the Lost Ark by adelikat & Alyosha in 01:34.56 in terms of TAS timing, so this movie cannot obsolete it. However, we can nevertheless accept this one in its own category, which makes it clear the target is to minimize RTA timing.
Joined: 1/24/2018
Posts: 367
Location: Stafford, NY
Note that the submitter is also the current RTA record holder for this game using a more conventional gameplay approach. It's hard to tell if the TAS approach could be reproduced by a human runner though given several points require frame perfect timing...
c-square wrote:
Yes, standard runs are needed and very appreciated here too
Dylon Stejakoski wrote:
Me and the boys starting over our games of choice for the infinityieth time in a row because of just-found optimizations
^ Why I don't have any submissions despite being on the forums for years now...
This movie currently has the same (blank) branch label as [3559] A2600 Raiders of the Lost Ark by adelikat & Alyosha in 01:34.56, but their goals look different, and one is unlikely to obsolete another. On the surface it may look like the difference in time measurement is just nominal, but in this case there's more to it.
A bit of historic background.
Traditional movie timing on TASVideos has always been from power on to final input. In some cases it makes sense to instead rely on in-game time (if the game has that). Real-time speedrunners on the other hand (their hobby is traditionally called RTA) usually time their records based on gameplay duration, and also sometimes use explicit in-game time.
Now when it comes to comparing 2 records (most of the time it's done to determine if the new submission is actually better optimized than the existing publication), we don't blindly compare overall movie durations. Because that can be misleading, for example if the new movie uses a more accurate emulator that causes more lag, even though pure gameplay is now shorter. Or maybe the new movie uses a different version of the game that has longer loading screens or menus.
To determine that the new movie is an improvement, we directly compare actual gameplay:
Movie Rules wrote:
Improvements must be in gameplay to warrant obsoletion. You should not be losing gameplay time compared to the published run: If you find a new timesave that's 30 seconds faster, your movie should be at least 30 seconds faster, unless there is unavoidable time loss as a result. Any time saved or lost from version differences and/or emulation accuracy is considered unavoidable, and is discounted. Your improvement movie may even be longer than the published run, but it is still considered an improvement if it improves upon gameplay.
Most of the time it's clear what "gameplay" means, if it's identical in the 2 movies barring the part that is now done faster in one of them.
If a level is now completed faster, but menuing is sloppier, that still means gameplay has been improved.
Even if gameplay itself is now a bit sloppier, but there's a new trick that compensates for that, and saves more time than is lost from sloppy play, that still counts as an improvement.
It's also possible that gameplay is the same, but menuing is now done quicker. We still accept that as an improvement if there's really no more time to save in gameplay.
Then there's also [4032] NES Super Mario Bros. 3 "game end glitch" by Masterjun & ais523 in 00:00.78, which doesn't even enter what is traditionally called gameplay (like the previous movie did) and triggers the ending right from the title screen. It was still considered a valid improvement, because zero gameplay time is shorter than non-zero, and whatever is left can also be improved, and it hasbeen! So again, when regular gameplay can't be improved anymore (there can't be negative amount of it), optimizing title screen inputs still counts.
Measuring gameplay duration using in-game timer has some common traits. That timer often discards loading screens and menuing, which makes it useful to measure gameplay duration indirectly (by simply looking at the value) and to compare 2 movies indirectly (the one with the shorter value is considered more optimal).
I say "indirectly" because most of the time people relying on IGT don't watch the 2 movies side by side in order to see how much frames are saved or lost in each scene (which is also what one is meant to do when TASing a game and attempting to beat an existing record). The IGT value itself is a decent initial indication, and further analysis could be done later if needed.
RTA timing is somewhat similar to just relying on IGT, but what to do if the game doesn't have one, or if it's inaccurate? Then the speedrun community of a given game decides when the timer is meant to be started and stopped. And the portion of the game that is covered by that timer quite naturally means what that community considers to be meaningful gameplay for that game. The exact entity that has to be optimized during play, and is compared when records are being compared.
Another similarity between that and relying IGT is in how the rest of the game is treated. A perfect example used to be Sonic games (I've heard it's not anymore due to pause abuse?). Meaningful gameplay is whatever is meant to be competitively (and/or collaboratively) optimized and compared. The in-game timer indicated that, and then there was the score tally that takes the longer the faster you beat the level. That score tally is not considered meaningful gameplay for Sonic and is completely disregarded.
If it wasn't disregarded, the speedrunning scene for Sonic games would have been dead long ago, because once someone finds the optimal combination of in-level time and score tally duration, it's impossible to consistently keep improving that overall duration by finding new and new tricks. New tricks would just get canceled out by longer score tally.
Now that's yet another similarity between IGT and RTA timing: portion of the game that is not covered by the timer is completely disregarded. This is why RTA runners may stay on the menu screen until they're comfortable to start the actual run.
And finally where I'm getting at with all this story.
This submission's title screen can't be completely disregarded... because it's what makes the meaningful gameplay portion of the game shorter. That screen is a part of the overall RNG manipulation and is explicitly used to get more optimal randomness in the actual game.
Normally, non-gameplay time is disregarded exactly because it doesn't affect gameplay itself.
To solve this, at some point we introduced this clause into Movie Rules:
Movie Rules wrote:
Runs that optimize In-Game Time (IGT) are only allowed in Standard if they contain faster gameplay than a real-time focused run.
It means that if IGT is shortened by using a glitch, then IGT itself is not a relevant metric anymore for optimality. If it's hacked then its value is moot.
Now that's only a limitation for the Standard class. In other classes one can do whatever with IGT and it'd be fine as a showcase in its own regard, as long as the goal is understandable.
This submission doesn't aim for IGT because this game doesn't have it. And it doesn't aim for shortest TAS time either, because it takes longer overall than [3559] A2600 Raiders of the Lost Ark by adelikat & Alyosha in 01:34.56. Instead it aims for RTA timing, but it defines it in such a way that allows for (potentially) unlimited delay on the title screen without counting it as a part of the movie, but doesn't disregard that delay entirely and instead uses it to manipulate in-game events.
So I think it can be said that this movie partially disregards the title screen time: negative implications on movie length are discarded, but positive implications on in-game duration are used.
I haven't seen such a situation before, so I feel it needs some discussion. I do not believe it can go to Standard because the goal is defined in a very tricky way. And it can't obsolete the existing publication because their conditions are not equal (and under equal conditions this one is longer).
So it could go to either Alternative or Playground depending on what everyone thinks. Note that Playground will soon be reformed one final time, which I'll soon make a thread for.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
I feel that this type of run (showcasing the fastest a game can be beaten when timed from the point of actually starting the game at the tile screen to the point of reaching the endgame, regardless of the initial start delay) has publication value along side a run that is timed from power-on.
Because it’s not a traditional timing method from a TAS standpoint (being neither in-game time nor power-on based time), I feel that Alternative class is the best landing place for such runs. Runs like this could be branched something along the lines of “RTA timing” or whatever would be a more appropriate distinction.
If "shortest time starting from first input" (which is effectively what this run is) is accepted as Standard alongside "shortest time starting from power on" then you get a bunch of movies that are identical to those already existing, just without skipping through any opening splash screens/cutscenes/etc.
It makes sense to judge the goal based on how different the gameplay is, and that makes it Alternative for me. I don't think I agree with Playground - it's a well-defined goal, not particularly arbitrary, and does complete the game. If SMB "minimum presses" is Alternative, this should be eligible too.
I think it's definitely worthy of publication in any case - I found it more entertaining than the existing movie, as in the "relevant" portion there's a lot less waiting around, and a lot more of "okay, where am I now?"
Definitely a fun watch in my book.
Especially because even in boot up to last-input timing, this TAS is only 12 seconds slower than the previous movie, even with it delaying a whole minute at the start. (Note: my 12 seconds figure is based on the video, so I think this movie has some blank frames at the end).
This run seems like it would be an ideal case for the proposed alternate timing field.
