So I've just discovered something very silly, but is still a timesave, albeit a small one...on the bomb mission...you actually don't need to pick up any bombs to complete the mission. You just have to get to the end. So that's only a couple of frames saved, but now that potentially opens up the idea of a new route, as long as it reaches the end. I'm looking around right now to see if I can find a potential new route, but in the meantime, it looks like we can make the tas a couple of frames faster, lol :p

Edit: Okay so it turns out this level is actually very linear. The few alternate paths only lead to a dead end, with maybe some ammo or health, but you'd still need to back track to get back on the regular path. Man, as a kid, I remember getting lost way more on this level.

Haha nice! Do you still have to pick up the last one? I thought I tested that. NOT shooting the two guards at the end was one or two frames faster for the record (if you pick up/disarm the bomb) :) I also THINK I've tried ignoring the pickups on Incubators a long time ago but might be worth checking these things again?

By the way on a technical level I have found that while the game runs at 60 fps, many checks seem to be done at 30/second, which might be the reason many tricks are 2-frame. This also shows at the end of each section where a change of a single frame during the section makes the transition to next section (doors etc) either 1 frame earlier or 1 frame later than expected. So 1 frame saved or lost often becomes 2 or 0. On some occasions I have had loading times / transitions between zones change by more than 2 frames but I still don't know why.

Interesting about the frame rules. You can skip EVERY bomb, but I'm pretty sure you'll need to pick up every incubator to unlock the door.....but maybe if we could bypass that door using a clip somehow...maybe that's a discovery waiting to be found.

This just in: it turns out you do, in fact, need to pick up the incubators. It was a long shot, but I was hoping the only checks the game was looking for was to go through the exit of each of the 5 rooms. But alas, you do need to collect the incubators. We just need to find a way to clip past the door and we'd be golden.

Edit: I've found a faster route on Incubators. Only 1 second though. Let me grab a clip, and I'm still looking for other possible routes.

Yep that's how I remember it being. I think those things we pick up are batteries to bring to the incubator room. Not much to work with to get a clip through the door but it would be a massive skip if possible. Don't really know how to do it but in theory I could look at where the game stores the data of which batteries have been picked up and track if that is used for something else in the game and can be manipulated.

So this route may or may not be faster with tas strats in mind, but in a real time run, it's a lot easier to save a second with this route. The grenadier on the original route is tough to get by quickly, so that's why this route is faster. We'd have to time it with tas strats to see if it's faster with tas.

Interesting! I had to do a quick TAS-check and it's actually 23 frames faster :D Plus probably easier as you say, those ladder jumps past the grenadier are tight.

I noticed in your runs that you don't do the vineless T-rex escape (jumping off the car directly to the platform. It's barely faster but it's fun. I found that in a real run so it's not too precise. Trick is to stand far forward on the hood, turn right to force the camera to pan and jump up from below ground-level. Camera pans just in time to let you on the platform. Might take some practice and it has definitely ended some runs but still worth it

Hey uhh, I'm pretty sure I found a faster Helicopter kill that the tas

:D That's sub 33 right there! Does it only work on the cage or did you switch due to ammo? Imagine shooting the helicopter in the first fly-by earlier in the level :P I think it's untargetable then but I'm not trusting any of my assumptions anymore. Not sure if you actually have to kill the cage? I think it was just faster to do it than not, with ordinary weapons that is

I tried killing the helicopter without killing the cage, but I actually don't think you can. Either way, in real time, it's actually quite difficult dodging the cage, and it's slow since you pretty much can't jump at all, or else you'd get hit by the cage. And yes, the bazooka only works on the cage, suprisingly.

On the first fly by, you can hit the helicopter, but I'd wager it's invincible at that point. If we're lucky, best case scenario is that the damage we apply to it there will carry over into the final fight. If you can damage it, then maybe a tas could kill it? If that's the case I have no idea what would happen, but I doubt it'd even happen in the first place, but it's worth a test, at least to see if damage carries over. For reference, (on hard) it takes 8 shotgun hits to kill the helicopter.

Also, hype for the sub-33!

Could changing the order of the standard levels be beneficial to use that weapon in unexpected places? Can't think of much. Maybe the fans on Blockade, but there are no convenient ammo pickups there?

I've thought about it, but I don't think there is any other beneficial place to use the bazooka. You can use it on the triceratops, however it takes 5 seconds to go out of your way to get ammo. One bazooka shot = 4 shogun, tranq rocket shot, but I don't think it's worth it anyways because you can fire off your shots very quickly anyways.

One thing that's been looming in the back of my mind, which I'm sure is possible to pull off...arbitrary code execution. Could it be possible somehow to write our own code and wrong warp to the end credits? I know these SNES games love to have their own versions of ACE, but my pea brain wouldn't have the slightest clue on how to even go about figuring out how to do that.

I've never done anything like that (this being my first TAS), but the tools I use have ways to monitor the RAM during play. I imagine the way to attempt ACE or wrong warps is to check where the game stores for instance the data for the next area to load and track it to see if that same space is used for anything other than targetArea, that we can change somehow, or make other data use more bits than it's supposed to be able to and hope the game writes it anyway and overwrites other data. Something like that.

Apart from monitoring RAM, a brute force way could be to try to overload the game somehow. Get a lot of sprites on screen at once. Try to get all your raptor friends together at the same time and hope something weird happens (: Other candidates could be red birdos (Dimorphodon) or flamethrowers. Any enemy that follow you around. I'll play around with it when I have more time to spare!

Edit: Some BazookaBoyz sometimes follow you a bit too I think

I looked into the Helicopter fight on Gallimimus and it's not really the weapon used but the timing of destroying the cage that matters. Normal mode is the same number of shots required as on hard. The helicopter always leaves screen to the left when the cage is destroyed. In the old TAS I destroyed the cage too early (with shotgun) which made the helicopter leave screen too quickly. In the latest version I allow the cage some time on the return cycle to keep the helicopter on screen long enough for a 1,5-cycle kill. I ended up using one shot of the bazooka and the rest shotgun, both to avoid damage from the explosions and to skip ammo pick ups (up to 5 frames. initial 2 frames for not holding forward + 3 lost frames that I think is some minor acceleration/deceleration). Sometimes you've got to lose frames to make seconds, as they say. I'm sure someone says that.

I'm still working on optimizations and handling cascade (the unpredictable effects and desync later in the run after a change) but Galli should end at least 279 frames earlier but I may have lost count somewhere because I ended up cutting exactly 420 frames from the screen between Galli and Scout.

Scout desyncs more than many other levels even though there are no clear RNG-elements. Both second and third zones often need frames added or removed after changes earlier in the run. Third zone has extra weird stuff going on, like the flying compsognathus.

Congrats on the official sub 40!

And for the TAS, we need to find some more frames to save for sub 33 :D Current time (at constant 60fps) from player select to first frame of the phone check animation (and start of the fadeout) at the end is 33,010833 minutes. To complete blackness of the fade out is 33,0144, so almost a second still needed

Just curious, how are you going about adding these changes to the tas? Do you just play normally up until a specific level/specific load zone, and then just redo that section and splice it in to the current run? If so, is that what's causing desync? And does the desync happen on a completely different level that didn't have it's own route altered? Very strange if that's the case.

I had a feeling the heli kill might not save as much time for the tas, because I realized afterwards that in the real run, we have the heli fly back to the right side to give it as much screentime as possible. I take it nothing came of trying to damage the helicopter on it's first appearance in the middle of the level?

I'm getting ssome crafty ideas for the final boss kill. I think a 4th cycle kill is doable within a run (currently it's at a 6th cycle). On the first flyby, if I target the homing missiles, I could then destroy it on the 3rd cycle (only the homing missiles can be hit on the 3rd cycle). I could still destroy the nukes and the engine, or whatever is right next to the nukes, on the 2nd cycle. Then the 4th cycle, kill the thing right next to the homing missiles, then the dudes, then the turret, although the turret would be a tight kill, not to mention it's hitbox is pretty small. If I miss the 4th cycle kill, a good lineup for the fast flyby on the 5th cycle should be enough to kill the turret.

I wish I wasn't at work so I could play around with it. After tonight though, I'll be free again for a few days before going back to work.

The TAS is essentially a long spreadsheet of inputs frame per frame that can be replayed. I can go back to any specific frame and make changes, but then it has to recalculate how the game will play from that point. The inputs for every specific frame needs to correspond to the game state I made them for to keep the game running as intended. Mostly I just need to add or remove a few frames just after a change to sync back up, but there are some parts of the game that are affected in unexpected ways. So yes the game can be in sync between a change and a later point that still desyncs requires tweaking. RNG levels for obvious reasons and Scout for weird reasons are the main culprits.

Gallimimus heli(Gallicopter?) still saved several seconds making it a one cycle fight! (or 1.5 since cage has to start the return but at least the helicopter never leaves the screen) I could not hit the chopper at the mid level fly-by, and while you can hit it at the end without destroying the cage it doesn't seem to really take any damage. Cage has to be destroyed, which is still faster anyway.

At the final helicopter (hard mode end), is sticking closer to the targets working for higher fire rate like predicted?

I'm currently working on faster T-rex fight. Unfortunately it seems to have invulnerability during some parts because the amount of hits required for a kill varies a lot. It's very unclear when it's vulnerable and not. All I know is it always blows up during the roaring animation, but I think your sub-40 run landed the last hit during the 'backing up'-animation. I've been doing weapon damage calculation on the cage (since all weapons hit it) with some unexpected finds. Not sure if damage is different for different targets though but I used the scientifically sounding Converted Cage Damage (CCD :D) to find out how many hits are minimum to blow up the T-rex, and then figure out which hits are ignored. Still unclear so far.