I started looking at this game after the previous submission and completed a full run which is 1,590 frames faster than yours (when your input is truncated to final input). One major thing of concern is how your input is still coming off as RTA like which can be seen in the Gum Monster fight. You can defeat the boss with Pac-Man only crossing the Gum Monster once, and you don't have to react to the monster's arm, instead letting Pac-Man tank the hit. And the RTA like continues in that input when you look at it and see how you're shooting your pellets holding longer than necessary. Please, take your time and use the tools that BizHawk provides such as TAStudio, cause you'll be able to clean up your input, and also discover that the B button helps save a frame in the Main Menu.

and also discover that the B button helps save a frame in the Main Menu.

What do you mean?

Please, take your time and use the tools that BizHawk provides such as TAStudio, cause you'll be able to clean up your input

Echoing this. Looking at the input directly, it seems like these runs are being performed in real time, just using rerecords. While that can be considered a TAS in and of itself, a true TAS will work frame by frame to ensure that every movement is precise and optimal. TAStudio is a godsend for this, allowing easy input editing and the ability to jump to any frame you want at any time. I would highly recommend downloading one of our published runs and studying how the author lays out their input. You'll almost certainly see things like single frame presses of buttons as early as possible in order to do things on the first possible frame, which is something that cannot be done consistently in real time over the course of an entire run. I've said this many times before to many different people over the years: The act of creating a TAS gives you infinite time and infinite retries to create a single run, so use both liberally! Nothing should be out of reach. Every possible idea you have can be tried, and SHOULD be tried, even the ones you know for a fact aren't going to work. You can try any idea at any time, you can tweak your approach in any way you want, and you can test several things far quicker and far more efficiently than even the most experienced RTA runners can. Something else I've said before is that all you need to create a great TAS is patience and Google: The patience comes from learning the tools, learning your workflow, testing and re-testing strategies, optimizing sections frame by frame over and over until you're confident they're done as fast as possible. Google is for everything you might have missed! Also, other search engines are available. You can approach a TAS from an RTA background, of course, and you can sort of treat the act of TASing the same way, but the only things that should be carried over are mental. An RTA runner will replay a game over and over and over and over again to try and improve their time: This should be kept in TASing, however the replay aspect should be micromanaged down as far as possible. Unlike an RTA runner, who can only "rerecord" to power-on and only knows if they've improved their time once they've finished the entire game, you can rerecord to literally any frame you want and figure out exactly where you're gaining and losing time at any point you want. Each screen becomes its own game to optimize in that way. You can clear a screen in 1800 frames, keep working at it, and find a way of clearing it in 1750 frames, then keep working at it further and possibly complete it in only 1700 frames, and you can move on to the next "game" whenever you're satisfied. This sounds like a lot of work, and it is, but it'seasy work. Zenlike, in a way. As you optimize your workflow, the process gets even easier, too, and you'll find yourself taking much less time overall while still putting in the same amount of effort, if not even more effort. I've knocked out entire optimized runs over the course of a single day before, and while that's an extreme case, it's also not an unrealistic one. I'll stop myself before I continue speaking in circles, but I'll leave with a parting thought on top of everything else: If this is a hobby you truly want to continue pursuing, and I damn well hope that's the case (as I do for every new submitter), it's worth it to stick around the community and learn things from people, being led by their examples and working with them to create better results than what one single person can do. The way we operate when it comes to new submitters and suboptimal runs isn't meant to be discouraging at all, it's simply meant to try and produce the best end results possible. We want people to improve and succeed more than anything, so I hope our words aren't coming off as cold and/or pushing you away. We're not coming from a place of mockery or insult when we talk about what we see in submissions, we're coming from a place of providing information, clearing up misunderstandings, and hopefully allowing self-improvement. In other words, don't give up. We're here to help.

om, nom, nom... minty!