This topic is for the purpose of discussing #6747: dave_dfwm's NES Rambo in 16:54.13

TEMP ENCODE _{not official Link to video}

Wow. Finally, someone has made an improvement! What about that weird glitch shown in AVGN's video when Rambo falls through the bottom of the screen and gets to a completely different location? I'm wondering if it can save even more time.

Context on what Dimon wrote (6:13 if the timestamp doesn't work).

Oh, thank you! That's it.

Context on what Dimon wrote (6:13 if the timestamp doesn't work).

Context on what Dimon wrote (6:13 if the timestamp doesn't work).

That warp brings him back to the very beginning of the game. This isn't useful since the characters will do their beginning of the game dialogue until you rescue the POW. And when you rescue the POW, you're already just half a screen away from the helicopter to take you home anyway. I don't know how or where that could work in other places, or if you could be warped to other places (not just the start of the game). I can mess around a bit, but I don't think anything will be found.

Thanks, jmosx36, for the encode! The game is somewhat repetetive, but the wall climbing glitch helped break it up enough to make interesting. The action also flowed smoothly throughout. Yes vote.

I have an emulator question and hope someone can answer it. While making my run, I went back and rewatched multiple times and I wanted to fast-forward certain sections (such as the time spent as Co). The emulator gives the option to fast forward up to 6400%, but while using this, it never seems to go faster than just 300 or 400%. Additional speeding up didn't change emulator speed. Can anyone tell me why?

Could it be that your computer can't emulate it any faster? BizHawk is somewhat more resource-intensive than FCEUX. Also check if you have any framerate limiters in your system (such as RTSS or the one inside the nVidia driver), that tripped me up some time ago.

I found out this TAS was slow than RTA 2 months ago, but I am not familiar with this game so I didn't pay much attention to it. It's nice to see the previous author improve it, yes vote.

Syllabus Precedent for the legitimacy of the game and branch was established by 2012 adelikat 1163. The previous run was accepted to the Vault, and the justification was that the game lacks variety. This TAS is about 1:44 faster than the currently published TAS. It is similarly faster than the best time on the speedrun.com leaderboard. The movie satisfies the “not slower than” condition. Opinion The game qualifies for the Metroidvania platform-adventure genre. The film on which the game is based—Rambo: First Blood Part II—is a classic of the 1980s action genre. The combination of these two facts should theoretically yield a vastly entertaining game. 92 percent of Metroidvania movies are Moons or Stars, yet the currently published Rambo movie is one of the few in the Vault. The viewers agreed with the tier placement; the previous movie has unanimous low entertainment ratings. But does this new movie change the entertainment value significantly? It is certainly an improvement to some degree. The gameplay is almost identical to Zelda II’s gameplay, which is well-liked enough that more games aside from this one have copied it. The problem in the previous movie was that there was too much continuous running around in an unclear manner, as if the entire game were nothing more than one long Zelda II random encounter screen. The multiple uses of the vertical wall glitch in the new movie provide a pleasant surprise and cut down on some of the endless running. However, the game is still not close to Zelda II’s entertainment level (as a game, not regarding any particular run), since Zelda II involves not just its gameplay mechanics but also the boss/enemy variety, item discovery, puzzle-solving, assortment of glitches, and route exploration. This game also suffers from the fact that the speedrun does not visually appear significantly different from the TAS. Much of the time lost in the speedrun compared to the TAS is from taking enemy damage, as opposed to any TAS-exclusive demonstrations. Talking to the final two characters appears optimized. Rambo could move further left to initiate the talk with Murdock earlier, but the backtracking afterward would add too much time. Rambo moves the minimal distance left needed to initiate conversation. However, I believe that there are at least minor optimization flaws in the movie. For example, early in the game (starting at frame 4700), Rambo changes direction at an exit, goes left for a little while, changes direction again at the next exit, and walks right for a while. In the current movie, the black screen comes up at the end of this sequence on frame 5755. When I tried it, I knocked four frames off and the black screen came up on frame 5751. I believe this part is a genuine timesave rather than an intentional delay to change RNG. But the author is correct to note that enemy placement can be bothersome and cause delays. Other feedback noted the AVGN glitch, in which Rambo took damage and fell through the floor. It appears to be an extension of the vertical wall glitch, except it is horizontal. When Rambo reaches the edge of the screen, the game has to eject him somewhere. For example, around frame 8100 in this movie, the author creates a right-side exit that shouldn’t exist, and so the game figures out somewhere to place him. In the AVGN glitch, it is oversimplified to say that it is a mere warp back to the beginning. AVGN first lands on a screen in the same industrial area, and then it appears that he triggers an exit in the ceiling rather than the floor. The game may be unsure where to eject him at this point and returns him to the beginning by default. It may operate similar to the fairy ceiling glitch in Zelda II. The Battle of Olympus, which is also very similar to Zelda II, has glitchy ceilings too. Although the glitch ultimately may not yield any shortcuts, it seems that it should at least be investigated further. It is worth noting that the author’s real life return to TASing this game after 14 years away (the previous movie was submitted in 2006) is a perfect parallel to the plot of First Blood Part II (1985), in which Rambo returns to Vietnam to search for POWs at the camp from which he escaped in 1971. The author deserves a standing ovation for this extraordinary effort to immerse himself in the psychological state of the character that he controls in the game. For the reasons stated above, this movie deserves to be Accepted into the Vault.

Mr. Arc, that was a magnificent speech! You should apply for a judge post. =)

I enjoyed this movie and voted yes. I agree that it's pretty much only Vault-level content, though that's more the fault of the game itself, and not of the author.

So first off, I thought the wall clips looked a tad slow. I found that it seemed like you can't drastically improve all of them, however I found this improvement of two frames. Simply by removing right input at frame 08108, you can move all later input two frames sooner. This syncs to the end. Additional improvements are likely possible. Secondly, some people were proposing accepting this submission to Vault. If the tier is Vault, this submission becomes substantially sub-optimal as it intentionally avoids damage, even where it could save time. A movie that intentionally avoids damage as an artistic choice is acceptable for moons, but not for Vault. EDIT: Some basic RAM Addresses

some people were proposing accepting this submission to Vault. If the tier is Vault, this submission becomes substantially sub-optimal as it intentionally avoids damage, even where it could save time. A movie that intentionally avoids damage as an artistic choice is acceptable for moons, but not for Vault.

It is true that Vault movies should not have goals that sacrifice time, but it is also true that judicial precedent was established when the previous movie with the same no-damage goal was accepted for the Vault. Doesn't stare decisis have the higher priority?

These look suspicious indeed: http://tasvideos.org/Movies-C3030N-C3035N-Vault.html Precedents are explicitly worded if they happen, and even then we still review if they work in a new situation too. Either the case may be slightly different, or some policy may have changed, so relying on past decisions that don't line up with normal rules requires caution. If it's not explicit, it's likely just an error.

Most of those are very early Vault publications (back when the Vault rules were a lot less clearly defined), or pre-tier publications that automatically were moved into the Vault based on ratings. The previous Rambo publication falls in the former category. In both cases, submissions/publications may not have been adequately scrutinized to conform to Vault standards as much as they were supposed to. Generally speaking, while past judgments can indeed be used as a reference at times, they are by no means infallible and should not be blindly trusted.

I'd like to note that if avoiding damage doesn't cost too much time, it could still be accepted to vault in spite of being able to be obsoleted by a run that does take damage. However, if it is significant, or if there are other issues with the run, it would not be acceptable. So far this TAS looks to me like it could have used a little more time being worked on.

When rambo was put into the vault, I don't think we necessarily knew that it sacrificed time to avoid damage. However, issues like that were known. The precedent, if any, is that while a run that avoids damage can be published to the vault. It can and will be obsoleted by one that does not. Also by extension, if we have a vault run that avoids damage and a better movie comes along that will avoids damage, it makes sense for the new movie to obsolete the old.

Re: the discussion about taking damage. In almost every case, I'm actually not sure taking damage can save time. On most enemies, I can avoid or kill them with no delay. On some enemies this is unavoidable, so there may be a 2 - 8 frame slowdown to time a jump slash or dodge the enemy. However, taking damage results in a knockback which would put you further behind than 10 frames of distance. The direction of the knockback is determined by what way Rambo is facing, and it takes 8 frames (from a run) to turn him around. So you'd have to turn him around to get the knockback the right way, and then use more frames to turn him back around to get going the right direction again. I think the only times taking damage would save time would be with the bouncing ball enemies that pause occasionally to be vulnerable and killed. But as an artistic choice, I think it would look weird to take no damage the whole movie except for two or three places near the end. The previous movie (which had a stated goal of no damage) was accepted, so I assumed that was an OK precedent to go on. However, I completely understand if this current run would be obsoleted someday by a "takes damage to save time" run if it is shown that taking damage can result in more than a 3-4 seconds of time reduced.

Would it be hard to modify the run to just be as fast as possible?

I might have overestimated the amount of time lost from taking damage but to me the real problem is the potential for easy improvements. I didn't check that many spots but I feel there's a lot more potential. EDIT:

However, I believe that there are at least minor optimization flaws in the movie. For example, early in the game (starting at frame 4700), Rambo changes direction at an exit, goes left for a little while, changes direction again at the next exit, and walks right for a while. In the current movie, the black screen comes up at the end of this sequence on frame 5755. When I tried it, I knocked four frames off and the black screen came up on frame 5751. I believe this part is a genuine timesave rather than an intentional delay to change RNG. But the author is correct to note that enemy placement can be bothersome and cause delays.

I confirmed that this improvement doesn't actually change RNG at all and is another relatively free 4 frames saved. EDIT 2: You can cancel the damage knockback animation by attacking, and cancel your attack by jumping, meaning that taking damage would in fact likely save quite a bit of time in a number of places. EDIT 3: Example usage, saving 10 frames (but changes RNG so desync ahoy): User movie #63473280776366799

User movie #63479847765337933: 243 frames ahead of the submitted file so far, with taking damage. I think at this point I'm invested in improving this. Also I got this really dumb lucky time save that saves like 2 seconds: Link to video

om, nom, nom... want more!