Sorry for bad english.
- running while attacking will faster an normal.
- Jumping while attacking will cause more damage than normal
- Final Boss can be defeated without destroy 4 core
- Holy Sword "SP" attack can take 1 boss life bar
- higher difficulty enemy HP will increase
+ Easy enemy 0.5x hp
+ Normal 1x hp
+ Hard 1.5x hp
+ Extreme 2x hp
- best sword form worst to best. My opinion
Normal Sword < Fire Sword < Ice Sword < Thunder Sword < Holy Sword
- Color enemy "life bar" from RED to Gray:
RED > Green > Dark Blue > Pink > Gray
Samsara: Claiming this one.
Samsara: As pointed out in the thread, this run is noticeably suboptimal. Note that the frame count/timer on Spike's run suddenly jumps forward at 2:07 in the video (it's due to a desync: the footage at that point had to be spliced in), so the final time on that side isn't accurate, but it is a little under 15 seconds faster over that 4 minute section. An optimized TAS of this game could potentially be under 20 minutes if that pace is kept throughout the rest of the game.
A TAS is more than just a regular speedrun with save/load states. They're meant to be as perfect as possible from the very start. You want to do everything you can do ensure that your run is as fast as it possibly can be before you submit it. Some basic tips: Instead of spending a day on the whole TAS, spend a day on a single level. When you get through with it, try again and see if you can do it faster. Keep doing this until you absolutely can't save any more time on that section, then move on. Your input implies that you're playing in real time: Don't do this. Play in frame advance, or use TAStudio if that's more your thing. The slower your game speed, the more precise you'll be, and you want to be as precise as possible.
Don't be afraid to use re-records! They're there for a reason, so you can remove any and all mistakes. A little over a thousand for a 20 minute run of a fairly complex-looking game doesn't inspire confidence, and the input shows that. For example: Here is your input for entering the Option screen at the very beginning of the run. It's mashed through, which implies that it is not optimized. You want to find the first possible frame to get through, which I have done here. This is more or less what all your menuing input should look like, instead of mashing and hoping for the best. Don't autofire either: Autofiring often loses frames over time since it's likely that the first possible frame to advance will land on a blank input frame.
In total, I managed to save 19 frames on the opening menuing alone. The game often accepts input far earlier than it shows, so it could be possible to save even more time than I did. Just keep trying it earlier and earlier. You may have to hold a button for a couple frames for the game to accept it. This isn't always necessary, but make sure you try it anyway.
Spike's movie shows where and how you can save time in the earlier stages. The biggest time saves seem to be through attacking sooner (there are noticeable waits when you should be able to attack), better movement (dropping down that cliff looks to be a huge time loss), and figuring out the best way to clear out enemies in each section. The spawns seem to be set, so you should be able to figure out an optimized order to kill them. The aim should always be to kill the last enemy as soon as possible, so you'll always want to position yourself in a way where you can do that. Since enemies come from both sides of an area, try waiting on the side with the final enemy spawn and letting enemies from the other side approach you from behind, like at 0:27 in the comparison video. This may not be faster in all cases, but it should be a good start for planning. Spike does lose time in a couple places compared to you, so there is definitely some improvement to be had over his run.
I think this game has a lot of potential, so I hope you do decide to keep working on it, because I'd really like to see an optimized run. If you do that, I highly suggest posting a game thread for it here and keeping us up to date on your progress. To help you start off: The first GO arrow should appear on the screen at frame 4858 at the latest, 23 frames faster than this submission (19 from the opening menu and 4 from clearing the enemies a little faster). You should be able to use the comparison video and Spike's input to optimize further. Don't aim to match the times, aim to beat them. It's definitely possible, and with a little extra effort and time spent, I know you can do it. A time under 20 minutes would be awesome, and it looks like it can be done, so aim for that.
Be sure and listen to the feedback in your submission threads as well. This isn't the first time you've submitted something with noticeable mistakes, and submitting too many suboptimal movies too quickly could mean you temporarily lose submission privileges, which nobody wants to see happen. Don't be afraid to ask the community for help or more advice if you need it. We judge TASes fairly harshly, but we absolutely don't do the same for TASers themselves. We're always here to help.
For now, though, I have to reject this for being noticeably and provably suboptimal. Keep working hard and come back with something amazing!