IRF

Further to my previous post, I've implemented my suggested alternative rope motion in the attached file. ** Code changes: this involved firstly changing the command at #937D from a XOR (IX+$00) to a XOR (IX+$01), to pick up the 'Is the rope left- or right-of-centre?' flag, instead of the 'Is the rope swinging left- or right-wards?' flag. But that had exactly the opposite effect of what I intended (the left key made Willy move rightwards on the rope, and vice versa), so I inserted an additional XOR #80 command to toggle the movement back to the correct way round. (I needed to insert two additional bytes for that, so I deleted the redundant two-byte instruction at #93B1, and shuffled all the code in between.) [There may be a simpler way of doing the above without using two consecutive XOR's, but I couldn't think of one.] ** Now that I've dealt with the technical stuff, please have a go with the rope in the Swimming Pool in the attached file. I think this makes most sense of all - pressing left moves Willy leftwards on the rope, regardless of its swing direction, and pressing right moves him rightwards on the rope - simples! In the original game, if you keep a single movement key pressed as Willy swings on the rope, then he switches from climbing up to climbing down the rope (or vice versa) at the moment when the rope reaches the furthest extremity of its swing and turns round to start swinging the other way - that doesn't seem to follow simple logic (hence SymbolShift was prompted to initiate this debate in the first place!) But in the attached, if you keep a single movement key pressed as Willy swings on the rope, then he switches from climbing up to climbing down the rope (or vice versa) at the moment when the rope passes through its vertical position, from one side of the screen to the other. i.e. as the rope changes between having a 'sloping down to the right' versus a 'sloping down to the left' gradient. This seems to me to be a more logical distinction to make, because if you consider the horizontal component of Willy's motion, he is always moving in the direction (left or right) which corresponds with the key that you are pressing. Thoughts? Logical Rope Movement.z80

SymbolShift wrote: "Anyone I know who jumps on a rope for the first time is totally baffled on how to control Willy, as the climbing direction reverses when the swing direction changes. That said, after a while, you do eventually get used to it. I have coded the above method in my game (since it's true to JSW), but I did try a simplified version (left = up, right = down, no matter what swinging direction) and it seemed way more logical." Your suggestion doesn't seem more logical to me, because when the rope is left-of-centre, pushing the left button (to move up, in your system) would cause Willy's horizontal position on the screen to move rightwards. It would only be a more logical method, it seems to me, at times when the rope is to the right of its central position (i.e. to the right of where it would hang if it wasn't swinging). Perhaps the most logical set-up would be to have the up-down functions of the left-right keys toggle to having the opposite effect every time the rope passed from the left-hand side to the right-hand side of the screen, and vice versa (as opposed to toggling every time the direction of swing flipped). That would involve the key functionality being determined by the value of the Rope Animation Frame variable (whether it is a positive or negative number, treating the byte as a signed number so that values > than 128 are negative), instead of using the Rope Direction Flag (the highest bit of the guardian type byte in the guardian definition buffer) to determine which keys send Willy up/down the rope.

Was he a clever de-bugger though? 😄

SymbolShift posed this query in another topic, but I think it belongs here really? : I was super impressed by the WRITETYPER run's from RuffledBricks and DigitalDuck. I wondered if any progress had been made on using AI to find the quickest teleport routes for this? It seems like a task well suited for AI, since there are so many possible combinations for the human brain to calculate. That said, I also understand this would be a complex beast to write, and possibly require a supercomputer to process 😁

I presume the above is referring to the use of AI and WRITETYPER in Jet Set Willy, not Manic Miner? It might be worth replicating the query in the similar topic under the JSW heading, so it's in the right place?

The two attached snapshots are both from the cavern 'Wacky Amoebatrons' (one from original Manic Miner, the other from Manic Jet Set Willy), and comparing/contrasting them illustrates several subtle differences between the two games: - The Coronavirus-esque vertical guardians animate more quickly in the original game (their 'pulsating' happens more quickly), compared with Manic JSW; - The conveyor also animates more quickly in the original game (first and third pixel-rows are 'sheared' by two pixels per time-frame; those pixel-rows are moved along by a single pixel in Manic JSW - which causes the 'messy bedsheets' in Willy's Master Bedroom!); - Observing the horizontal guardians moving along their platforms, you can see clearly the difference in behaviour between when they change direction at the rightmost extent of their range (some animation-frames are skipped in Manic JSW as the guardians turn around, but that's not the case in the original game), versus when they change direction at the leftmost end of their range (smooth, 'normal' behaviour in both games); - Finally, I've just noticed that the cyan horizontal guardian in Manic Jet Set Willy's 'Wacky Amoebatrons' cavern moves at the faster rate (advancing every time-frame), whereas in the original Manic Miner this cyan horizontal guardian was of the 'slow-moving' type (advancing every alternate time-frame). Manic Amoebatrons.z80 MJSW Amoebatrons.z80

I'd be particularly interested to hear from @Norman Sword whether or not the odd behaviour of the horizontal guardians (as they change from moving rightwards to moving leftwards) was a deliberate design feature, or a bug in the routine which moves the guardians. I suspect the latter, especially since it's asymmetrical and doesn't affect the guardians at the opposite extent of their bounds.

Further investigation reveals that it's not so much the saw guardian turning around more quickly which makes this room tricky (the saw is already advancing leftwards towards Willy when he enters the room). Rather, it's the rolling egg above the nest which bounces back more quickly from the outer edge of the nest, leaving Willy with less time to collect the item safely. I've created two snapshot files with Willy in a 'safe place' inside the nest, one from an original(ish - no changes to Cuckoo's Nest) JSW file, and the other from Manic Jet Set Willy - see attached. In both files, the game was paused when the egg had just reached the rightmost extent of its range (in its 'sloping down-to-the-right' frame of animation). If you load up each file in turn, unpause the game (press the 1 key or something to unpause, not a jump/move key as that would put Willy in danger!), then you can see the different behaviour of a horizontal guardian as it turns around and starts to move left. (I find it helps to keep the A-pause key depressed whilst doing quick jabs at the 1 key, in order to see the animation frame-by-frame.) In original JSW, the egg remains balanced on the spot for a single time-frame (with no change of animation frame), then moves to its 'lying horizontal' position in the next time-frame, and then moves to its 'sloping down-to-the-left' frame of animation in the next time-frame. In Manic JSW, the egg jumps straight from its 'sloping down-to-the-right' to its 'sloping down-to-the-left' frame of animation, as soon as you unpause the game, thereby missing out two animation frames. N.B. In both files, when the egg reaches the leftmost extent of its range, it spends two time-frames in its 'balanced vertically' frame of animation, so the 'turnaround bug' doesn't affect the horizontal guardians when they are turning from leftwards to rightwards motion. The consequence of this bug/feature/change is that in original JSW, it takes 13 time-frames for the rolling egg to get from its furthest position at the outer edge of the nest to a point underneath the item (which I believe is meant to be a tiny bird nesting atop the giant imposter egg?), whereas in Manic JSW, that same part of the egg's traverse only takes 11 time-frames. JSW Cuckoo's Nest.z80 Manic JSW Cuckoo's Nest.z80

Danny, I noticed that in your recording of the Mutant Telephones cavern, you collected the item that is located under the portal before the item in the middle of the cavern (to the right of the slow green horizontal guardian). So it seems that the effect of the 'bug' on the timings of the guardians may have even driven you to change the order in which you collected the items in this case? (So it affected not just the subsequent moves, but the preceding moves as well!)

Based on this post of Norman Sword's: https://jswmm.co.uk/topic/649-manic-jet-set-willy/?do=findComment&comment=14251 ... the general ethos of this project seems to have been to compress the two games into one, not to change the data (other than where necessary to allow fix certain critical bugs, to allow the game to be completed without loss of life). I think most of the room/cavern layouts, guardian behaviour, etc have been faithfully replicated, but the odd unintentional difference may have crept in during the process of transferring the data. However, I do think that the jerky movements of the horizontal guardians, as they turn around from moving rightwards to leftwards movement, is something that arises from a bug in the code which handles the movement of horizontal guardians. I don't think it's likely that it was a deliberately designed feature? Perhaps Norman Sword might wish to comment on this when he next visits the site? ** BTW Danny, I've just noticed that in your post which directly follows after the one I've provided a link for above, you commented on how, like Andy, you had encountered more difficulty than usual in the Cuckoo's Nest (particularly when playing the game at the fastest speed). As with Andy, I think this was probably because the saw turns around at the nest and comes at Willy much faster because in that moment it skips straight past two frames of animation (the 'turnaround bug', as I've started referring to it). You didn't specify which rope room in the above quote, but I believe you are talking about The Cold Store? Again, the Turnaround Bug is probably to blame (or, should I say, to be thanked, since you're reporting an improved performance here!) - the horizontal guardians fall into a different relative alignment with the phases of the rope's swing, and this happens relatively quickly after Willy enters the room because they all (except for the penguin on the floor) move within quite narrow bounds, so it doesn't take long for them to reach the edge of their ledges whereupon the bug makes them skip forward a few steps.

Actually, that red seal seems to have the rightmost extent of its bounds go further right by one character as well, which brings it even closer to the green seal. I thought the dataset for the caverns was identical, but it seems not in this instance? The green seal also goes one more character to the left in MJSW than it does in original MM. The two seals, at the closest point of their approach, get six frames of animation closer to each other in MJSW than they do in original MM (four frames because of the range of the seals being wider by one character, and two frames because of the 'turnaround bug' I've discovered).

Sorry yes, I should have aimed the question more widely. It makes sense what you're saying, in light of what I've discovered. Basically, if you're approaching a horizontal guardian from the left (like the saw in Cuckoo's Nest), whilst it's moving away from you but about to turn around, then when it turns around to come towards you it does suddenly leap forward by two animation steps faster than usual - making it harder to evade. I've just done a couple of screenshots from the Abandoned Uranium Workings, one from original Manic Miner and the other from Manic Jet Set Willy. In both cases, I've captured the last moment when the two seals are moving towards each other - in the next time-frame after the one I've captured, the red seal has turned away from the green seal to avoid colliding with it. You can see that in the case of Manic JSW (bottom image), the two seals get much closer together than they did in the original MM (top image). If the game was set up in such a way that the two seals normally have a near-miss, then with the new game engine they probably would have collided (killing Willy in the process!)

By my calculations, having done screenshots of both recordings (your 'old slow' recording and your 'new slow' one), you lost 560k points during the first jump through the beam in both recordings, and you lost 720k points during the second jump through the beam in both cases. (It is perhaps possible to miscalculate if you freeze the game using the emulator's pause button, and if in doing so you capture the game 'mid-calculation'? I used 'Stop Playback' just after Willy had commenced each jump, and then used the in-game pause button so that the freeze point fell at the same point in each pass through the main loop, so that the update of the score tallies should have been consistent.) This is also exactly the number of points I would have expected you to have lost based on the number of characters of the solar beam that Willy occupies during each time-frame of his jump through it. N.B. The discrepancy of 100 point which you mentioned is, I believe, due to the in-game clock having ticked over to the next minute in one of the cases. (You lose 100 points every time a minute is added to the on-screen clock.) In summary, I don't believe you could have reduced the amount of points lost to the solar beam any further than you already have managed to do. 🙂

Both of those things could be explained by the erratic behaviour of horizontal guardians as they change from moving rightwards to moving leftwards, as I outlined in my previous post in this topic. e.g. If Willy falls into the flashing portal when the penguin is in there, if the timing is right then the penguin may have partially* moved back out again faster than usual by the time Willy has dropped in there? (*Partially, as in sufficiently far for Willy not to clip it with his back foot as he's falling whilst the front of his sprite is pressed up against the wall.)

In the process of investigating Danny's recordings of 'Manic Jet Set Willy', I have just noticed that the animation of the MM guardians transposed into Jet Set Willy does not faithfully replicate their behaviour in their original Manic Miner setting. EDIT: In fact, the motion of the JSW horizontal guardians isn't faithful to the original either (see below). If you study the vertical guardians in the Solar Power Generator, they only animate (i.e. change frame of animation) every other time-frame, whereas in original MM they animate every time-frame. (In JSW you can have slow-animating or fast-animating vertical guardians; it seems that Norman Sword set the MM VG's to have the slow setting). You can see this even more clearly in the Final Barrier, where the winking eye winks more slowly than it does in original MM. There may be some instances of vertical guardians where this difference in animation speed makes manoeuvres around VGs more difficult (or alternatively more easy) - especially if such guardians change significantly in size during the course of their animation cycle (I'm thinking of the first Amoebatrons cavern, for example, where they pulsate inwards and outwards). But that isn't likely to be the case with the Solar verticals, which remain the same size and circular in shape (with just the inner pixel pattern within the circles changing as the discs rotate). Perhaps of even more significance: I noticed that the horizontal guardians (both fast and slow) within the Solar Power Generator miss out two frames of animation as they transition at the rightmost extent of their range from moving rightwards to moving leftwards. Visually, the effect can be seen as if the guardians are 'bouncing' off the right-hand wall of the cavern. Curiously, this doesn't happen at the other end of the horizontal guardians' range (when they switch from moving leftwards to moving rightwards). I have also observed this effect within both the Final Barrier and the Central Cavern (the latter having an eight-frame horizontal guardian, so it doesn't just affect the four-frame MM HGs). EDIT: In fact, it also affects the regular horizontal guardians in the Jet Set Willy rooms. Observe the yellow monk at the bottom-right of First Landing. When he turns from moving left to moving right, his nose is at its shortest extent (as is the case in original JSW). But as he turns from moving right to moving left, his nose is at its longest extent. Or see the rabbit in Ballroom West. When it stops pirouetting at the right-hand end of the table, it does a big, sudden bounce forward as it turns round to go leftwards. (That's a good example as the edge of the table shows you exactly the extent of its range.) This difference in the motion of the horizontal guardians (which may be caused by an error in the part of the code which handles the guardians as they change from moving right to moving left?) is likely to have a potentially significant effect on the gameplay. In fact, it explains why, when I was looking for Danny's reported 'sudden flash' of the solar beam horizontally across the top of the conveyor in an original MM game file (the file only being modified to deplete the initial air supply in the solar cavern by a single unit, to try and replicate what Danny was experiencing in his recording of 'Manic Jet Set Willy'), it didn't happen in quite the same way. (Changing the initial air supply from an 'odd' to an 'even' amount of units did change the way that the Slow HG interacted with the solar beam, but not in quite the same way because the timings are all different due to the missing animation frames of all the HGs. So the 'sudden flash' wasn't as sudden - it didn't last for just a single time-frame as Danny encountered.) ** Danny, in light of the above observations, I wonder if you found when you were playing 'Manic Jet Set Willy', that jumping over/under/past certain guardians (particularly horizontal ones) was either more easy or more difficult than anticipated, based on your 'usual' routines/routes in the regular games? I'm thinking maybe something along the lines of the horizontal guardians in The Forgotten Abbey, or The Endorian Forest, or The Cold Store, not being in sync in the same way as they usually are? **** Incidentally, another aesthetic change is that the conveyors in 'Manic Jet Set Willy' seem to animate by one pixel at a time rather than two pixels - which causes Willy's/Maria's bed to look rather messy! Also, the 'Forgotten Abbey' is now less forgotten, as the 'secret entrance' is more obvious due to the visible animation of the conveyor in The Wine Cellar.

Well done, Danny! I would guess that the remaining discrepancy may possibly be due to Willy being cell-row-aligned versus not cell-row-aligned (i.e. At a certain stage of his jump) at the moment when the beam first hits him during that pass through the beam. (Or conversely during the final instant when he passes through it?) Meaning that in one recording, there is a single time-frame when his air is sapped by 8 units (2*4) but in the other recording his air is sapped by 12 units. This is where my suggestion for Norman's colour-changing solar beam only to have its PAPER colour change for those cells actually occupied by Willy - with it reverting back to yellow once it has finished passing through his sprite - would come in handy.

The Remaining Lives bonus in Manic Panic effectively prevented the use of 'kamikaze' as a tactic to boost the final score at the end of the game - it isn't worth sacrificing a life in order to collect a cavern's items twice, as the loss of the bonus points for that life more than outweighed it. (I've not seen any reference to a similar feature being present in this game though.) Conversely, in one of our MM projects (I can't remember which one - Manic Mixup maybe?), I set things up very differently: rather than having a set initial number of lives for the game, and once they're gone it's Game Over (but extra lives can be gained at certain score thresholds); the game I'm referring to assigned a specific number of lives at the start of each cavern, with the number of lives tailored to the difficulty level of the cavern (e.g. for the easiest caverns you were only granted a small number of lives, etc). Now that I think about that system, it actually provides a positive incentive to kamikaze lives in order to boost your final score! e.g. There's no point in leaving the Central Cavern with any lives to spare, since your number of lives in the Cold Room isn't restricted by the fact that you lost lives in the Central Cavern, so you may as well collect all the items/kill Willy before reaching the portal/repeat until you're down to your last life. Sorry, I've rambled way off topic there! 😲

I wonder if 'as manufacturer intended' should ensure that all the skulls (e.g. in the Ballroom screens) animate using all four available frames of animation? (Forgive me if that's something you've already done!)

One other thing I noticed when I was playing around yesterday is that if you lose a life in the solar cavern, then there's a 50:50 chance that the slow HG starts off in the right place after Willy 'respawns'. So I wonder if using the 'cheat' teleport feature* could enable you to achieve the right set-up for the cavern? It wouldn't look as smooth in the final video though. (*Assuming that feature is available in this game for the MM caverns? EDIT: I've just checked through Norman Sword's posts earlier in this thread, and it seems that there is no access to the MM caverns via a cheat teleport.) A related thought: you could improve your final score* by collecting the first two items, and if the solar beam zaps Willy whilst he is above the conveyor, then you could sacrifice that life and try again until the right circumstances come about. (*Unless there is some kind of 'Remaining Lives' bonus in this game, like there was in 'Manic Panic'?) Of course, that tactic could have been employed to even better scoring effect in the Kong caverns (by sacrificing a few lives after toppling Kong from his perch), but I presume it would breach one of jetsetdanny's principles when making a walkthrough recording (i.e. the game - where possible - should be completed without losing any lives)?

Further evidence - the three images attached here are all screenshots of the first frame of the solar cavern, from each of Danny's complete walkthrough recordings (Slow, Medium and Fast). Achieved by watching the recordings back, using the Stop Playback feature in ZXSpin at the point just after Willy had jumped into the 'Amobatron's Revenge' portal, then holding down the A [pause] key until the remaining air had counted down. The only difference (in terms of gameplay) is that the red tractor (slow HG) has its back to the wall in the 'Slow' image, but has already advanced one frame of animation away from the wall in the 'Medium' and 'Fast' images. (There is also a cosmetic difference in terms of the INK colours of the items, which are determined by the value of the game's internal clock, but that doesn't affect the gameplay in any way.) I also found that if I re-watched the Slow recording until just before Willy jumped into the Amoeba's Revenge portal, then used Stop Playback, I could time the jump into the portal myself (and then hit pause whilst the air was counting down) so that the red HG was one step away from the wall in the opening time-frame of the solar cavern. This is further evidence that Danny's suboptimal score in his Slow recording was just unlucky (baked into the scheme of things from the moment that the solar cavern was set up, but there's a 50:50 chance that it could have worked out differently), and that there isn't anything inherent about the execution of the game in its Slow mode which caused the 'quick flash' of the solar beam through Willy when he tried to navigate up the conveyor. It also confirms that it would be a fruitful exercise for Danny to try and improve his final score (should he wish to do so) in Slow mode, by following the steps I set out in my previous post in this topic. 🙂

Danny, it has occurred to me that the above observation could be the key to obtaining an improved score for your recording of your walkthrough of the game in its 'Slow' setting. Should you wish to do so at some point - and assuming your .RZX recording was left 'open' to further amendments (via the Rollback feature) - it would be a relatively easy task to do: - Resume recording of your Slow walkthrough; - Roll back to the last saved point before Willy entered the portal in 'Amoebatron's Revenge'; - [If that point was some way back in the Amoebatron cavern, then save another Rollback just before entering the portal, for convenience later]; - Jump into the portal, and as soon as the Remaining Air countdown starts, save another Rollback point; - Then whilst the remaining air is still counting down, press down the A [pause] key and keep it pressed down until the next cavern is drawn; - The Solar Power Generator cavern should then be displayed, frozen in its first time-frame. Observe the slow (red) horizontal guardian; does it have its back up against the right-hand cavern wall? Two scenarios here: Scenario (A): There is a small gap between the frozen-in-time red horizontal guardian and the wall. In this case you are in the correct 'reality' to complete the cavern without being zapped by the air beam (apart from on the two unavoidable occasions when going there-and-back to collect the item at the top-right), so you can simply: -- Roll back the recording once, to the point where the previous cavern's remaining air supply is counting down [this is only necessary to avoid a short pause in the recording which would otherwise be noticeable to future viewers of the video]; -- Enter the Solar cavern without pausing the game this time, and complete the final two caverns ('Solar Power Generator' & 'The Final Barrier') as efficiently as possible (something which you are well practised in doing by now!) Scenario (B): There is no gap between the frozen-in-time red horizontal guardian and the wall. In this case you are in the 'alternative reality' where the air beam would hit Willy later on as he is proceeding up the conveyor. So in this case, you should: -- Roll back the recording twice, to the point just before Willy entered the 'Amoebatron's Revenge' portal; -- Jump into the portal after an infinitesimally short delay, and repeat the above process (i.e. save a Rollback point during the air countdown, then press A and keep it depressed, observe the starting position of the red HG in Solar, etc - and repeat until you achieve Scenario A above). ** Since you had already accrued a better score prior to entering the Solar cavern in your Slow recording of the game than was the case at the equivalent point in your Fast recording of the game, doing the above should enable you to achieve an even better total final score than you have achieved to date in any Speed Mode. 😀

One final thought: the mechanics of how the solar beam interacts with the guardians can get even more complicated in a modified game where the number of air saps per contact with a cell containing Willy's white INK has been edited. e.g. in the Hard levels of Manic Mixup, the code has been edited to cause five saps per cell of Willy, instead of four saps per cell contact. It means that at certain times when Willy is jumping or falling through the solar beam, when he is spanning three cell-rows, you can end up with 3x5=15 saps of air in that time-frame (instead of 3x4=12 in the regular MM). The effect of that, as far as the consideration of when to move slow horizontal guardians is concerned, can be to 'cancel out' the regular decrementation of the air supply. Tricking the program into thinking that a slow HG moved in the previous time-frame when it didn't (or, conversely, didn't move last time when it actually did). The net result is that a slow HG sometimes either moves at double speed (i.e. a normal HG's speed: being moving along in two consecutive time-frames), or else it doesn't move at all for two consecutive time-frames (appearing to momentarily stop in its tracks). This creates a jerky, stop-start feature of the guardian's motion at times when Willy is jumping/falling through the beam! One could even conceive of a hypothetical scenario where two guardians have paths which cross over, but whose speeds/initial positions have been carefully designed so that they never collide - however, with the above effect taken into account, a slow HG could end up out of its 'usual sync', causing a fatal collision with another guardian! (That's something to think about for game designers if they are designing a solar cavern with slow guardians and a solar beam which causes an odd number of air saps.)

P.P.S. Adjusting the initial air supply in the solar cavern by one unit also changes the initial position at which the slow horizontal guardian is drawn, as the 'Move the Guardians' routine is executed once before the first time that the guardians are drawn. So in one case, it has its back to the wall as soon as you enter the cavern (as per its defined starting position in the cavern data - the 'Move the Guardians' routine having decided to skip moving the slow guardian this time round), whereas in the other scenario, it has already moved one animation-frame away from the wall before you first see it.

Aha, mystery solved! Below are four screenshots taken from the videos embedded in this post of Danny's: https://jswmm.co.uk/topic/649-manic-jet-set-willy/?do=findComment&comment=14272 The first two snapshots show two consecutive time-frames from the 'slow' recording. The last two snapshots are the equivalent two consecutive time-frames from the 'fast' recording. You can see that the guardians (other than the slow HG) are all in precisely the same positions in images 1 and 3, and also in image 2 compared with image 4. Willy is also in almost the same position - in truth, it looks like Danny was on course to complete the cavern slightly quicker in the slow run-through, although the penalty for an additional solar strike more than offset that. For the purpose of this analysis though, Willy is occupying the same character row (which the solar beam either does or doesn't pass through) in all four images. The difference in solar behaviour is actually due to all of the factors I discussed earlier, including the fact that a moving entity can interact to a greater or lesser degree with the solar beam, depending on whether it spans two or three character-rows. (I mentioned that point in relation to Willy in respect of the beam being more or less 'deadly' at times, but it's also true of vertical guardians in so far as they can rotate the beam a greater number of times if the beam passes through them whilst they aren't row-aligned.) In Image 1, the red vertical guardian is row-aligned, so it deflects the beam three times. Since the beam approached the red VG horizontally, that means that it ends up going downwards to the cavern floor. In Image 2, the red VG spans across four character-rows, so it deflects the beam four times. This time it ends up traversing horizontally above the conveyor (and as it happens, through Willy). (Note that the red horizontal guardian occupies the precise same position in both Images 1 and 2. It is just protruding past the edge of its platform, and being a Slow HG it only moves every other time-frame.) Image 3 represents the exact same scenario as Image 1. But in Image 4, the red VG doesn't get the opportunity to send the beam along the top of the conveyor, because by that point, the red horizontal guardian has moved one frame of animation to the right - just enough to pull it away from the path of the solar beam coming down from above, so that the beam continues on downwards to be absorbed by the brick tile in the cavern floor. And as previously discussed, the different position of the red HG (being 'one time-frame behind' in the first two images compared with the last two) can be explained by a different starting value of the air supply for the cavern. A fix would be to round up (or down) the current time counter upon entry to this cavern to the nearest multiple of 8 [that's assuming the air/game clock counter works in the same way as it did in original Manic Miner, incrementing by 4 each time], perhaps by applying an AND #FB (11111011) command to the value of the variable via a Portal Set-up patch? P.S. In the next time-frame after the one shown in image 2, the red horizontal guardian will have moved into the same position shown in image 4, so the solar beam will be heading straight down to the brick tile in the floor - hence the beam only flashes along above the conveyor for a single time-frame during that moment in the first scenario.

Hmmm, I've decided I need to satisfy my curiosity and resolve this issue of the solar beam sequence. It occurs to me that I may have misremembered exactly what changes between an 'odd' versus an 'even' initial air supply in the Solar Power Generator. Rather than involving the higher-up horizontal guardians, it may be the case that the exact position of the vertical red guardian in the middle of the cavern, at a certain point in time, is what prevents the solar beam from flashing past the top of the conveyor in one scenario and not in another. i.e. the solar beam could be deflected leftwards by the horizontal red in both scenarios for a full eight time-frames whilst it's at the leftwards extent of its range, but in one scenario, the precise timing of the red HG is such that the red vertical satellite dish re-deflects the beam down to the cavern floor for all of those eight time-frames, whereas in the other scenario, the red VG is only in the right place for seven of those time-frames, so the beam sneaks past and above the conveyor for just one of those time-frames. What I do recall it is that it was the complex interaction between the slow HG (the only variable when considering the difference between an 'odd' and an 'even' cavern), the positioning of one or more of the other guardians, and the solar beam - at a specific point in time (with unfortunate timing if Willy happens to be 'at the wrong place at the wrong time'!) Further investigation is required - watch this space...