TI-99/4A Speed Run (Parsec)

[Opry99er]

Since there don't seem to be many speed running categories for the TI-99/4A, I thought I would challenge myself... 

 

I decided to try to speed run Parsec.  Now many of you are asking (and rightfully so) "How can you speed run a game that has no defined ending?!"  Well, I had to make an arbitrary cutoff point for this speed run.  I chose the Killer Satellites.  
 

I started off innocently enough... running a pretty good first level with minor flaws in my gameplay... I went through and refueled and didn't lose a life on the first level at all.  Second level, same thing... except I did lose two lives on the saucers. More on this in a moment.

 

Beating the Dramites on the first two levels is pretty easy, before laser overheat becomes an issue.  

 

Level 3 was decent, one death on a laser overheat, but no huge time loss, honestly, since it was on a Dramite.

 

I reached the killer satellites at 12:07.  Not bad, but not optimized at all.

 

So next I decided to try to refine my methods a bit.  I found that I could get a 2-cycle kill on the Urbites and Bynites on the first level, and the same for the Urbites on level 2.  The strategy for the Dramites didn't change at all.  When I say "two cycle kill," that refers to the number of times you have to pass the same Horizontal plane as the enemy.  Each time you cross the horizontal plane of the enemy, you have to do damage.  By cutting down the number of times you have to horizontally engage the enemy, you save frames and, subsequently, time.

 

Another optimization I employed on my second attempt was to avoid refueling altogether.  As you all know, refueling takes a ton of time, and any time not engaging the enemy is time wasted off the run.  So, any time I got the refuel

warning, I would intentionally scuttle my ship.  This knocked a big chunk off my time.  I cleaned up my run a little more, not missing a single first pass of any enemy on level 1.  
 

I think it's important to define "first pass" here, because this could be the difference between a fast run and a slow one.  Hitting every normal enemy on the first pass is a great strategy on level 1, mostly because they all enter the field of play in a predictable pattern.... (well, ALMOST... I'll come back to this in a moment.)

 

In level 1, you can just position your ship below where they will be dropping in and spam the laser to Make sure they don't get past you.  This works great on level 1 because you only have to hit each normal enemy once to kill them.  It isn't absolutely crucial to get them ALL on the first pass, but making sure the field is clear before the last enemy in a particular wave enters the play field is important because if you miss the last one on the first pass, you will have to deal with the enemy picking up speed and screen-wrapping.  
 

So, my second attempt yielded a much better time, even though I lost two ships to the saucers (again) due to the random nature of their starting position, and the fact that they approach from the rear.

 

 

Second attempt?  9:01.  A good score, but still unoptimized.  I lost two ships and I had a scenario where one of the normal enemies on level 2 was screen-scrolling out of play at the bottom of the screen.  I had to wait for it to raise its horizontal axis and come in dangerously to pick up the kill just as it broke over the horizon.  
 

3 more attempts, and each time I kept having trouble with the saucers at some point.  
 

Then, a minor breakthrough...  Someone mentioned to me that the pattern on the saucers is dependent upon the number of extra ships you haven queue when the ships start their wave.  If you have more than 3 extra ships, the saucers pattern will always be random.  If you have 3 or fewer, it is a very predictable pattern that is repeated every time the same way.  So I decided to make sure I had only 3 extra ships each time I approached the saucers. If I had more, I would scuttle until I hit the 3 ship number.  
 

This knocked almost a minute off my time because I could accurately predict where the saucers would come into screen during their wave.  And so, even though I died a couple times (once on a laser overheat and once on a stuck input during the level 3 asteroid belt) I was able to get a time of 8:15:88.  This is unoptimized because I still lost lives, I didnt master the first pass strategy, and I wasn't able to get all the two-cycle kills I needed to get on the bosses.  
 

 

 

 

I will be doing some videos of this run as I get better... but I wanted to go ahead and get this run information out there in case anyone else wanted to give this speed run a try.  It's a short run, can be done in less than 10 minutes, and it's a great way to improve your skills.

 

Im thinking a time in the low 7 minute range is definitely possible, so there's tons of room for Improvement.  I am not even a GOOD Parsec player in comparison to many others, but I am doing the best I can to optimize this speed run because I feel like our community should be in the speed running business.  
 

I'll try to post a video tonight of a good run.  I'm completely open on the strategies I learn, so I won't be hiding anything.  If anyone has questions, I'm sure many of the better players on this forum will be able to answer anything I cannot.  Again, I'm no expert at this game, I'm not even a good speed runner.  But this could be a lot of fun for us if we can generate some interest.... maybe bring some new players into the fold for our little computer.

[Opry99er]

For posterity (and for new players coming into the fold) I want to explain why some deaths are okay and why others are not.  This has to do with the way the game handles the enemies.

 

On normal enemies (Swoopers, LTFs, and Saucers) they always start at the beginning of the wave and run the wave to completion.  That means that if you die during a wave of normal enemies, you have to start the whole wave over again, losing a ton of time.  
 

On the flip side of that, if you die on a boss (Urbite, Dramites, Bynite) you start back where you left off.  So if you destroy 3 Urbites and then die on the 4th, you re-spawn at the 4th Urbite.  
 

This is why I don't even attempt the random saucer wave... if you get 15 seconds into the wave and then die due to random pattern, you have to start back over at the beginning of the wave.

 

Intentional Scuttling should never be done during a small fighter wave.  Scuttle only between waves or when the game calls for a refuel. 
 

Much more coming.

[pixelpedant]

Very cool!  I've thought about speedrunning on my TI-99 before, but never about Parsec with respect to it!  It certainly would be nice to see speedrunning of such an iconic TI-99 game. 

 

I've otherwise thought about speedrunning Wing War, as while it doesn't have an "ending" state it does have an explicit (repeatable) core mechanic which culminates in completing a supercrystal and bringing treasure back to your lair, which can pretty reasonably be regarded as the natural goal for a speedrun and generally takes about 10 minutes played very efficiently, which is a good length for a speedrun, usually. 

 

However, an even more natural TI-99 speedrun for me might be Fathom.  Since it's a fairly long action-adventure with an explicit victory state.  Its downside vs Wing War is that an "optimal" run would depend on some rapid dives in which you just drop as quickly as possible and hope you don't hit anything.  There would be less banking on randomness of this sort in a Wing War speedrun, and more careful routing. 

 

A complete run of Fathom's nine levels takes about 25 minutes. 

[Opry99er]

I guess it's important to define some quick parameters for the run.

 

You should start your timer when you "press fire to begin."  You should stop your timer when the screen changes to blue and the first killer satellite appears on screen.

 

In addition, I would like to focus on a glitch-free speed run for now.  There are three glitches which can be used to change the game play dynamics.  The first is the "Level Warp Glitch".  This involves intentionally scuttling your ship before "Pressing Fire To Begin."  This can cause a level warp once you defeat the first wave of swoopers.  It's unclear if this was intentional... or if it can be manipulated to a consistent trick... but it seems right now to be random.  Sometimes it warps you to the end of the first level, sometimes it warps you to another level.

 

Another glitch is the "Clip Glitch."  This involves backing your starship up so far that the flame tail of your engine appears on the right side of the screen.  In this state, you are essentially invincible to the armored bosses.  This is a really difficult one to achieve, but I will be avoiding it as I work to optimize the run.

 

The third glitch is the "Beam Trail Glitch."  This involves spraying laser beams all over the screen by pressing fire on the joystick and the corresponding 'fire' button on the keyboard.  This will leave laser trails onscreen which help against Bynites.  When they turn invisible, the beam artifacts illuminate the invisible enemies.  
 

If everyone is okay with these ground rules, then we can proceed forward with these as the guidelines for our first official speed running category on the TI.  

Edited June 7, 2021 by Opry99er

[Opry99er]

10 minutes ago, pixelpedant said:

Very cool!  I've thought about speedrunning on my TI-99 before, but never about Parsec with respect to it!  It certainly would be nice to see speedrunning of such an iconic TI-99 game. 

 

I've otherwise thought about speedrunning Wing War, as while it doesn't have an "ending" state it does have an explicit (repeatable) core mechanic which culminates in completing a supercrystal and bringing treasure back to your lair, which can pretty reasonably be regarded as the natural goal for a speedrun and generally takes about 10 minutes played very efficiently, which is a good length for a speedrun, usually. 

 

However, an even more natural TI-99 speedrun for me might be Fathom.  Since it's a fairly long action-adventure with an explicit victory state.  Its downside vs Wing War is that an "optimal" run would depend on some rapid dives in which you just drop as quickly as possible and hope you don't hit anything.  There would be less banking on randomness of this sort in a Wing War speedrun, and more careful routing. 

 

A complete run of Fathom's nine levels takes about 25 minutes. 

Cool man!!  Yeah, I'd like to get a speed run going for Alpiner too.  Fastest time summiting all mountains including Everest.  Single pass... as soon as Everest is summited, the timer stops.  That's what I was going to do first, but I'm not good enough at that game yet to consistently summit Everest.  The only issue with that game is the RNG.  It appears to be extremely random on obstacles and enemies, which makes for inconsistent running mechanics.

[pixelpedant]

Yeah, Alpiner's RNG is rough.  As I've been learning lately.  Having seriously revisited it over the past week, after not really playing it at all for 20 years or so. 

 

It would be one of those runs where you just have to hope for good obstacle placement and restart if you don't get it, particularly on Garmo and Everest. 

 

And that kind of speedrun game does exist, of course. And furthermore, like Parsec, Alpiner has the appeal of being a first-party classic. 

 

But definitely, I prefer the idea of speedrunning something like Wing War where the world is complex and stateful, but always begins in the same state and changes deterministically. 

[Opry99er]

I'd like to hear some thoughts on PAL vs NTSC here too...  Typically PAL times are slower in speed running, and rather than split the category into two records, I was thinking of utilizing a method that has been adopted by the Mario Kart 64 community...  

 

They have a formula that calculates times.  On Mario Kart, the game displays at 30 FPS on NTSC and 25 FPS on PAL.  If a run is submitted on a PAL machine, they calculate a 17% speed difference, so the NTSC equivalent would be 17% faster than the PAL score.  
 

Essentially, when a score is submitted, the user would have to notate whether the run was completed on PAL or NTSC, and each submitted score would have two values: The standard value and the equivalent calculated value. NTSC scores would essentially be slowed by 17% to give the PAL equivalent, and PAL scores would be sped up by 17% to give the NTSC equivalent.  It's a pretty good system, but might not be perfect for us.  Maybe @Tursi could lend an opinion on this.

 

What are your thoughts, folks?

Edited June 7, 2021 by Opry99er

[Opry99er]

Additionally, I would like to just go ahead and get the big requirement out of the way as well.  I would like for all submissions to either be live-streamed or have video evidence.  Speed running has come a long way in the past 15 years, and so has technology and internet speed.  There's literally no reason why we can't require video proof of a run.  

[Opry99er]

6 minutes ago, pixelpedant said:

Yeah, Alpiner's RNG is rough.  As I've been learning lately.  Having seriously revisited it over the past week, after not really playing it at all for 20 years or so. 

 

It would be one of those runs where you just have to hope for good obstacle placement and restart if you don't get it, particularly on Garmo and Everest. 

 

And that kind of speedrun game does exist, of course. And furthermore, like Parsec, Alpiner has the appeal of being a first-party classic. 

 

But definitely, I prefer the idea of speedrunning something like Wing War where the world is complex and stateful, but always begins in the same state and changes deterministically. 

Agreed wholeheartedly.  Some of @Asmusr's games would be great too... particularly the ones that have defined endings.

 

@retroclouds's Pitfall port too.

[Opry99er]

1 hour ago, pixelpedant said:

Yeah, Alpiner's RNG is rough.  As I've been learning lately.  Having seriously revisited it over the past week, after not really playing it at all for 20 years or so.

 

[Opry99er]

Love that Alpiner video @pixelpedant!!!  ❤️ 

[Opry99er]

Here's a 10:17 run from a YouTuber who wasn't going for a speed run... the player has some skill.  You can see the issue with the random saucers in this video.  In addition, you can see how much time is wasted doing refuels.  
 

If you watch carefully, the player switches up his strategy on the Dramites at the end of the Dramite wave in Level 1 and he uses the same strat on levels 2 and 3.  The strat he switches into is the one I use exclusively on Dramites.  It is the most effective strat for Dramites on the first 3 levels.  
 

 

Edited June 7, 2021 by Opry99er

[Opry99er]

Tbh, @mizapf is the one I believe will own this speed run....  IIRC, he can play until he passes out without losing a ship.  ? 

[+mizapf]

I was tempted to play ...

[+mizapf]

From my experience, the different speed (50Hz vs. 60Hz) has some noticeable impact on my game play. For instance, I barely use lift speed 2 in 50 Hz, while it seems really useful in 60 Hz in the asteroid belt passage. Also, the enemies dive bit deeper in 60 Hz.

Edited June 7, 2021 by mizapf

[Opry99er]

1 hour ago, mizapf said:

I was tempted to play ...

Do it...

[Opry99er]

Okay, I came up with a checkpoint split list for this run.  I did a couple runs but didn't make it through to the satellites.  I died on the Dramites on level 3.

 

For an example, here are the times I got during my last run.  At this pace, had I finished the run, I would have hit the Killer Satellites at roughly 7:45.  
 

 

Swooper(1) :00

Urbite(1) :15

LTF(1) :35

Dramite(1) :48

Saucer(1) 1:05

Bynite(1) 1:30

Asteroid Belt(1) 1:45

 

Swooper(2) 2:06

Urbite(2) 2:41

LTF(2) 3:02

Dramite(2) 3:25

Saucer(2) 3:50

Bynite(2) 4:16

Asteroid Belt(2) 4:43

 

Swooper(3) 5:06

Urbite(3) 5:23

LTF(3) 5:46

Dramite(3) 6:14

Saucer(3) xx

Bynite(3) xx

Asteroid Belt(3) xx

Killer Satellite(End) xx

[+mizapf]

This is really hard to beat; speed running requires me to play in a more risky way, so I do have some trouble reaching the 4th level at all. ?

[+mizapf]

50 Hz: 9 min 17 sec; may certainly be improved.

 

[unhuman]

I accelerate out of the left side when the UFOs enter....  Gives me a better chance of avoiding when they're random. 

[Opry99er]

Today I got on a great run... I was 20 seconds ahead of my 8:14, on pace for sub 8... 3rd level, I got killed by 2 Dramites, and then I got busted up halfway through the saucer wave costing me 10 more seconds.  Had a clean Bynite wave and made it through the asteroids without losing a life.  Unfortunately due to my difficulties on the saucers and Dramites, I ended up with an 8:14.  Here's one bit of good news... if you take all my best segment times from each of my runs and add them up, you can see that a low 7 is possible.

[+mizapf]

I guess another well-suited game for speed running is TI Invaders.

[Retrospect]

Here's something I found, when playing Parsec ..... The Swoopers and LTF's can be disposed of faster, if you are in Lift-2 rather than Lift-3 .... and this is because in Lift-2 we can get better pixel-accuracy for our shots, rather than what sometimes happens is skipping over the top of them or below them. 

[Opry99er]

Interesting!  I use LIFT2 on Asteroid belts and Level 3 Dramites, but I've not tried it on Swoopers of LTF.  Those are usually no problem for me.  The big issues I end up having are the saucers and dramites....

[Opry99er]

11 minutes ago, mizapf said:

I guess another well-suited game for speed running is TI Invaders.

There would need to be a cutoff point... with no "end" it is impossible to speed run a game.  That's why I picked the Killer Satellites as the cutoff for Parsec.  Usually speed run games are platformers, RPGs, racing games, etc... as far as I know, TI Invaders doesn't have an ending.  Pitfall would be a fantastic game to speed run.  So would "Markus of Marinus" (cough cough... plug).  ? 

 

Alpiner too... to get to the summit of Everest, however the RNG is so wacky and unpredictable, it would be extremely difficult to make sure everyone was on the same playing field.... since the field itself changes every time you turn on the console.