Submission #8462: eien86's NES Galaga: Demons of Death in 08:17.78

Galaga goes pew pew

Category Rules:

Although the game is unlimited, it is divided into 4-stage loops. The last stage of every loop is a so-called 'Challenging Stage'. There seems to be only 8 types of challenging stages, after which the game provides no new content. The author of this FAQ is the only source I came up that provides this information.

On the other hand, the achievement guide for the Xbox 360 port of this game gives it's last achievement at stage 30, which more or less confirms there is not much more after 8 loops. However, I haven't played this port so I cannot confirm.

I added the perfect score condition because, although it is faster to end the challenging stages with an imperfect result (<40 enemies killed, prompting a faster score tally), I believe this would go against the nature of this game and the audience would not appreciate it.

The previous submission, #5260: LazerLemon's NES Galaga: Demons of Death "High Score" in 00:57.56, was rejected mainly for being suboptimal. However, it was also rejected for the choice of category. In their judgement, Noxxa stated that a movie for this game should at least contain the 8 challenging stages. My submission seeks to address this requirement.

Compared to the previous submission, #5260: LazerLemon's NES Galaga: Demons of Death "High Score" in 00:57.56, this movie improves the first loop by 255 frames:

I ran the bot to find the fastest way to beat each loop, so RNG manipulation is done within each loop independently. For some stages the bot found that the optimal way to beat them is to let one or two enemies through. It seems if you kill enough of them, the ones that pass through will just flee, which is faster than having to kill them.

Reading the FAQs, it seems it is possible to let yourself be captured and then rescued for double firepower. The whole process is just to slow to justify any extra frames saved from being able to double shoot. That is, at least for the 8-challenge category. If we played ad infinitum, then probably it'd be the best decision.

After running the bot for each loop, I did a manual pass to remove shots that the bot added but didn't land. I also added small celebrations after each loop and then reajusted the ship's position and RNG value by pressing L+R a certain number of times.

Please consider using this movie for encoding, as it contains extra inputs for entertainment purposes.

ThunderAxe31: Claiming for judging.

ThunderAxe31: Dropping my claim, as I'm improving this submission. I'll add myself to the authors when my movie is finished.

nymx: Claiming for judging.

nymx: Setting to "Delayed", pending improvements from ThunderAxe31.

nymx: Woah! what a trip this was. To start off, I've been seeing some of this game worked on for years by various people. The one that first caught my eye was a version featured at a GDQ 2017 event. Now that we have seen a judgment for a failed submission, we now have the criteria that completes all unique content and difficulty.

The run was extremely impressive. BOTing this game showed tremendous strength over the failed submission. I watched this run side-by-side, over that rejected TAS, and it was clearly a knock of for eien86. What I didn't expect was to see ThunderAxe31 finding improvements. This just tells you that the very nature of BOTing is extremely difficult to account for all variations that could yield better results. I don't fault eien86 for this, since I share the same frustrations with my own BOTing; however, this BOTing can't be ignored for its strength. During my analysis, I noticed that some aliens were passing through and I was curious as to what was going on. The way I see it is...the BOT didn't need to finish up some small parts at the end of each attack wave, yet it detected the completion with fewer inputs. This just demonstrates that the BOT found detection errors with the game. Even though it was pointed out in the submission thread...I am personally glad to see such a condition in this game being demonstrated.

So...we have a complete game of unique content and no more difficulty to explore. Even though some frames were found by ThunderAxe31, the re-syncing effort proved to be quite the effort and thus shows that it is not easily beaten....which is exactly what we want to see.

Accepting! Great work eien86!

Spikestuff: Processing...