User Files from FatRatKnight

BackJumps into Perma-Sprint. If ever you wanted to see just how powerful Perma-Sprint is, here you go.

The start here is to pick up a Mortar weapon early in the stage. Then the BackJumps are used to reach 4+ pixels per frame, so that we can peacefully charge up Sprint mode while outpacing the camera. We go for a swim right after, triggering Perma-Sprint and you can see what that means for our hero. We even found a door we can simply clip through because the terrain there is hollow, and Perma-Sprint is so fast we can pick up the Mortar in there and continue on our merry way all before the camera catches up.

After that, I'm not entirely sure what precision planning is happening for the last two victims. Disorganized notes, I say. It reflects where I stopped in my TASing of this particular stage as well.

This little segment was worked on by me before we just sort of dropped things. I wanted to see how early we can spawn a snail, didn't attempt to manipulate the luck to have it going west quickly.

As far as we can tell, it's best to mostly follow T8's route through this level. A few early items are picked up. The second key we pick up is called the Embedded Programming Key, where we BackSlide into a terrain embed to grab it off the desk rather than jump. The first actual wall we clip through is actually "hollow," with a single tile gap inside it that allows us to bypass it with a speed much lower than 16 pixels/frame. We then work our way north and stopped at the snail manipulation.

The rest of the route is basically planned to be T8, except that we set up Perma-Sprint on a bench while waiting for the imps to kill victims for us advance upon victims that we horrifyingly are unable to save.

This was the script we developed in the DTC. Has hitboxes, position, and speed. And a few random things. This was my version, without the side-radar.

All this one does is look at a few addresses to figure out when stages begin or end, as an assist to figuring out when we completed them.

Shows terrain boxes. Has an incomplete thing I was trying, the list of special terrain IDs are at the top, but that table wasn't used. Has something else I hacked together to show special IDs I'm looking for instead, rather critical in attacking the Sky World boss where I wasn't supposed to.

Colored boxes are used to point out bits of terrain I'm not sure about. Otherwise, the white boxes are close to accurate of what I think they are. Busies up the emulator a bit.

Shows hitboxes and a bunch of information. It's also our RAM Watch, but that was underutilized anyway. It has my guess on Ronald's hitbox rather than anything firmly known.

It also has my early guess on the wall hit detection, although now I realize the magic pixel is just at the feet level and not the vertical orange bars I put down. The green vertical bar was to assist in figuring out the item collection, and seems mostly accurate.

The "average speedometer" and "distance check" on the left needs to be calibrated to the start of each stage.

Skipped a few money pick-ups, but I still have $190 to spare when I get Mega Engine. There doesn't seem to be that many slow choices to skip now.

I have no idea why there are two invisible nitros in track 24. I happened to be viewing the CPU's nitro count when I saw it climb to 1 when I thought I picked up all the nitros. Sure enough, when I drive over that spot, I get a nitro. The location is after the sequential pair of zippers, the first ones you come across after starting the race, the ones right before the first time you fly past the crossover. As you hug the far wall, the same side as the zippers you will pick up two mystery nitros, the second one being near that dirt patch ahead.

If a CPU car drives over rough terrain, and no one is around to see it, does it slow?

Well, manipulation worked. Apparently the RNG was already set up correctly. At least enough for my script to find some possibilities when I left my last upload.

50540 cash after race 15. Enough to ditch a few money bags I picked up and still get Mega. 46140 cash after race 14, the difference between Black and Silver would mean an earlier Mega. 42090 cash after race 13. Skip Red and Skinny's, get Black, and we might have Mega then!

It does feel right to get Red for r2, Silver for r4, Skinny's for r5, and Mega for r16. Black for r3 and Mega for r14 is apparently possible, but then it's Black+Standard vs. Silver+Skinny's for r5 to r13, among other differences.

... Yeesh, the lag. It's the reason I went into that side spot for the extra life in race 13. Not taking it meant lag. It might be that taking every money bag I did avoids lag, so who knows?

This is what race 5 looks like when you manipulate a few letters to be exactly what you need. All those times where that yellow car frustrated you in the first game, watching this yellow car speed ahead is probably making you remember those times.

Four Score used because I needed player 4. Player 4 needed because of luck manipulation. Said luck was looking at player 4's letters collected for two of race 4's packs. By the arrangement I had for race 4, my script claimed there was only 2 possible timers out of 256 that could have worked. It's possible selecting a different player could work, but being player 4 maximizes chance of things working out in race 4.

Race 6 might be next to impossible to manipulate. There are six letter packs, two of them linked to each other on the same manipulation. Since these two are the same, they must be I, as that's the only one I can collect two of. The other four must be R O A M, as none of the remaining letters can pick player 4, and the other cars all have cheated their letter P by now.

Next to impossible doesn't actually mean impossible, there's some theoretical code path that will make it work. It's just a matter of finding a practical code path.

Mostly just trying optimal turns and such.

Apparently you don't lose speed for turning sharply? But only when at max speed? And not with the default engine?

I messed up in the Tug-o-Truck challenge. Well, sure. I win it pretty fast. I just haven't accounted for the speed dropping down a notch when you try to speed up beyond some threshold.

No real planning was done when TASing this. I just sort of dived in and got this so far.

A script I used, including the TAS Editor helper in there, hence the filename. (Actually, I sort of didn't want to bother renaming the file)

There's not a lot to the script. Mostly keeping track of distance, speed, sub-speed, and the music. This script is also how I measure my distance loss, to more easily compare segments where speed dips below 255 MPH.

Really the music was being tracked because I wanted to dance to it more easily. With a car.

C1 - 0'52"4 1'41"1 2'17"1 - Loss:11605

An improvement over my previous movie. Both in entertainment and in time.

However, it only really completes one course. I don't have any particular plans on finishing it.

One thing I did do in this one the prior one did not do is manipulate the sub-speed by that turbo boost thing at 255 MPH, so that I recover lost speed slightly more quickly. Also, general optimizations and all that. You know the drill.

A part of course 2 was also redone. Since it was lossless to begin with, the only possible theoretical improvements would be entertainment.

C1 - 0'52"4 1'41"2 2'17"7 - Loss:13333
C2 - 0'38"5 1'21"4 2'04"2 - Loss: _ _0
C3 - 0'38"7 1'29"5 2'07"0 - Loss:29399
C4 - 0'39"8 1'27"3 2'20"8 - Loss: 4486
C5 - 0'47"1 1'27"8 2'19"2 - Loss: _ _0
C6 - 0'39"8 1'25"0 2'20"0 - Loss:39412
C7 - 0'43"0 1'36"6 2'22"7 - Loss:18549
C8 - 0'35"7 1'14"3 2'20"5 - Loss:17802

Beats the following submission: #1188: TheSwordUser's NES Rad Racer 2 in 19:50.12

The loss numbers should be explained. After the most optimal start I know how to do, if I remain at 255 MPH the entire time, then the loss is zero. For every frame I spend at less than 255 MPH, this increments my loss indicator by the difference. Theoretically, each 255 loss is equivalent to a frame of gameplay. I think it's essentially 30 Hz, so 7650 is one second.

A trick the old submission hasn't done is to move a bit before the power start. Until the "race start" sound actually finishes, you can't power up, so I take the time to creep forward slightly. It's worth something like 765 distance units (based purely on unreliable memory), or close to three frames of 255 MPH.

The script has a few features.

On the left, taking advantage of the 8 pixels of dead zone, I have a speed visual that basically deconstructs what effect your current velocity has. Colors are used to denote target speed, and the left vertical bar shows how much of an 8 MPH threshold is filled. Tick marks are used for 1 MPH increments, but meeting these tick marks have no impact on the player's position. Filling the rectangles do, representing these 8 MPH thresholds.

As well, the console output is also used to spit out frame counts automatically (using register exec functions) on how long it took to complete each race. There is a disabled feature in this script that would also use this output for checkpoints, letting the TASer know when the player has hit the next segment of the track. Uncomment a line with LogCheckpoints() (under function BasicHUD) in it to activate.

Finally, the numbers display. Shame my memory of the script is basically nil on the details here, but so it goes. I recommend experimenting a bit, and if that isn't helping, open the script and look for my comments in BasicHUD. I do have a habit of using hexadecimal if the numbers line up better that way.

Considering my entertainment attempt was incomplete to begin with, I figure I should at least finish it properly, after all the set up was done. Got lag-free pausing, too, which probably involved firing up the bubble barrier early enough to get the sound to finish.

Hopefully I dance good. Somehow, this movie doesn't sync to the end, walking into what seems like extra lag in s15.

Spin Missile, a Buggy-Type move.
One would think it could hit up to 5 times, but we all know the boss invincibility, and it could only hit twice.

First, in order for the animation to sync up and allow for damage based on jump frames, I had to delay spawning the door by two frames. This loses the ammo drop for our flame weapon.

Second, no matter when I pause, I get a lag frame when finishing up the ammo tank.

Solutions sought. Oh, and I did a minor change at s9 ammo tank usage, doing so as late as possible.

So, boss stuff. Working with frame rules, pixel perfect positions, and all that mess, I saved a bit of time at the door. The fight itself is synced.

This movie delays 8 frames at stage select to the end stages for item cycle sync. Not sure what help or harm this is going to do for items when the delay is removed. Good luck with those items, too...

ThunderAxe31 been working things. Synced the boss. Fairly easy since adjusting boss actions to happen exactly on a particular frame is the point, done by changing pause duration.

We are exactly 4 frames ahead of the old item drop cycle.

I was working on this. Then I got a nice theoretical improvement to check.

The theoretical is now reality.

I couldn't get this working without lagging the boss intro. Sad.

Down from 56229, we have saved 85 frames.

I was highly annoyed by the fact the boss disappears for a moment just before becoming vulnerable once more. Man, if he just didn't talk so much, his timer wouldn't advance that far, and we can beat him up with another shot before that teleport.

... Wait, he operates on his own timer, doesn't he...?
Pausing advances the global timer, but freezes the object timers.
Dialog advances based on global timer...
Pause during dialog. We delay him frames without delaying dialog.

This line of thinking got us another second and some change. We have another improvement! Entertainment was not really attempted in this version, though.

All this really does is clean up input. By which I mean, I get rid of all those A holds that don't have any effect, input removal at times where the control is locked anyway, and a few Select presses are put in for timing checks.

Nothing exciting here, just making it a little easier for some random person in the distant future to check over our stuff. Just in case.

The best way to achieve the impossible is to refuse to believe the impossible.

I am about 4 frames away from a skip. And the only weaponry used here are basic shots and muzzle flash. Presumably, fancier stuff can get the skip. Or simply timing differences could make it happen.

Tell me it is impossible now. Tell me with a straight face after watching this closely. Is the goal that far away, where this movie is the dead-end that only shows we're denied the skip, or does it reveal a path yet untaken and calls in ever greater hope?

Only thing left to do is optimize this miracle. The s13 boss skip is not only possible, but it is done in this movie.

The way I ended it before was rather plain. Not even vanilla flavored. I added chocolate and mint to how it ended.

Well, the glitch-fest ended when the barrier did its job, so I thought I'd fire a shot to keep the glitch going. So you see a spread shot of glitch, followed by a bouncing glitch-ball. A little extra, really.

Copied in ThunderAxe31's latest stuff on the end of this movie. I didn't check if it contains lapogne36's stuff, though in my brief look I didn't see anything new.

So, now that I have used a Fragile Barrier against a boss, and successfully done so, I can feel satisfied in defying the words of some random person on the internet. Despite, you know, these words most likely talking about dealing with the endgame stages on decidedly finite ammo, and not in an environment where we have far less need for it, but let's be honest. We like twisting meanings around and making them somehow fit a different perception and then claim it was that all along.

Aw, now I went all meta and can't spark some kind of debate on what I actually did. You guys are going to smile too much at this.

Used spikeball shot as finisher instead of star, simply because the damage output was only just barely keeping up with the sub-HP. Star would not have done enough damage. Also, I technically avoided damage, in the interpretation of never getting an ouch animation, by using barrier and taking hits with that to disable my hitbox detection. This was enough to reduce the lag. Additionally, the Fragile Barrier was used against the boss as a single projectile that doesn't lag much.

Well, I removed a specific action on the s8 boss, restoring what the prior movie did. Sync after s7 returned. I changed the s5 boss exit, to show ammo on our other weapons then.

Currently trying to puzzle out how "lag-free" and "no damage" on s2 boss is even possible. I did go back earlier in the stage to find spots to remove "memory leak" from prior shots, and removed a little bit of red zone. I remember holding still doing nothing at the s2 boss was still causing lag during one attack, which was removed when we take damage. Well, this time, the hold still and do nothing isn't causing lag, so there's some slight hope of lag-free no damage run, but we're not left with much to work with.