RNG in avoidance. Doing it RIGHT.

[zaphod77]

A random phase can keep people from falling asleep. so it's a good idea to include them. however, their difficulty can go all over the place, because, well, they are random.  They tend to range from joke to moderately challenging (for the easy ones) and from hard to unfair for the toughest ones.

There is a better way.

Test the phase yourself with user specified seeds. if the seed is too easy, or too hard for you, try a new one.  Once you have collected 50 or 100 seeds of decent difficulty, randomly select a seed to start the pattern off with.

You now have something that cannot be memorized, yet won't give the player a cheap shot, since cheap shots in random patterns are just not fun. (save the cheap shots for static and streaming patterns.) It plays just like a standard random phase, but doesn't ever screw the player over or give a free ride.

I Wanna Clear the Easy Miku is pretty reasonable with it's randomness (if you pay attention you can deal with it all pretty easily). But for the second one, the randomness puts the difficulty all OVER the place, and it suffers for it. While the random barrages that expand from Miku always seem fair, the volcano and the second negi wave are all over the place.  And the small box section can get really rude at times.

Now if you find 50 seeds without discarding any for being too easy or too hard, then you have designed a good random phase, and probably don't need the workaround.

[Kyir]

So do you have any suggestions regarding the construction of this seeding system? I'm curious how you would go about it.

Aside from that, I think the assumption that RNG is necessary to keep a fight interesting is flawed. If you're having that problem, you might also try looking into positional targeting and aesthetic accents, both of which I've found to improve the boss experience a lot. If you have a bunch of medium-difficulty random apple sprays or whatever it's not going to be that much different from a medium-difficulty attack that's the same every time in terms of enjoyment.

[patrickgh3]

I had this exact same idea a while back. In theory it seems like it would work, but I think the hard part is putting it into practice. Ideally your system would be easy to plug into any project, and it would be straightforward to use with as much automated as possible. It's hard to gauge how hard it would be to make a system like this.

Easy or hard to make, I think the main issue would be getting people to actually use it. People who make fangames often skimp out on QA towards the end of a project (myself included), and this idea, even though it would be useful, is essentially just more work to do for a gain that's not immediately apparent. I mentioned the idea to Klazen a while ago and he said something along the lines of this, and for the most part I agree with him.

Although if you feel motivated to make a system like this, I don't want to discourage you from making something you want to make. And if it turns out to be *super* easy to use, then who knows, it might gain traction.

As for Kyir's comments, I agree it's definitely worth thinking about other ways to achieve a similar effect of less variance in difficulty, including the ones he mentioned. Though you shouldn't worry about making a *perfect* avoidance attack, because there are multiple schools of thought on what makes a good avoidance and you can't please everyone.

[Kyir]

So I gave this some more thought and also realized you could just use random_set_seed. The problem really comes down to the duplication of effort required just to make and test one boss attack. I would estimate about 5 minutes to thoroughly test each seed at different places in the stage, which means at 50 attacks (the minimum you suggested) the amount of time needed for that one attack would be about four hours. It could be significantly more if the pattern is elaborate.

That's not to discourage you from doing it personally, but I think you would be hard pressed to get other people to do it, especially since occasionally getting walled is already the expectation for many people when it comes to RNG-heavy bosses.

[Derf]

I agree with the comments of this thread so far. Programmatically this is not hard to achieve, but logistically I think it would grind against people. It would be very hard to test individual attacks in a vacuum since the issues may occur when they link up meaning every trial would have to be a full run through. Though contradicting that, you might be able cut costs by decreasing the amount of seeds logarithmically between attacks as people will, generally, see each consecutive attack less than the previous one (provided they don't beat the boss in one attempt). They will always see the first attack, and they will see each attack either at least as many times, or more likely, less frequently.

Though I think another idea that might work would be to code prevention techniques for walling. Such as having the boss keep an internal log of its previous attack directions and making sure each new "random" attack is not close to it, etc. This seems like good practice to have generally but I've not seen it in a lot of avoidances.

[zaphod77]

With the help of god mode, it's not too hard to just watch the phase play out, and check if there was anywhere something impossible to deal with even from far away.

So it really only takes about 10-15 minutes to do this test.  You only really need to go to 100 if you are truly paranoid about someone memorizing them all.

It's true that trying to prevent impossible stuff inside the pattern itself is a good idea. and sometimes it's doable on purpose, and sometimes by accident.  But sometimes it's not so easy to plan it out.  Going to DotKid mode with small hitboxes is a copout.  Making the pattern impossible proof from the start can also make it predictable, so doing at least one pattern this way is a good way to mess with the player's head, creating a pattern that looks like it could screw you over, but never actually does.

Even if this technique is not used, you should test any random pattern at least that much ANYWAY, to make sure it's not cheap.

I agree that well done static patterns can still be tricky and interesting, but once you figure out what to do, then it always becomes boring.  The purpose of the random one is to make sure that someone can't just clear the game after watching someone else do it.

[Kyir]

Getting walled is almost exclusively based on positioning, so unless you're some kind of rabbit-human hybrid and your eyes can look at different things at once, you're still going to have to go through an attack multiple times if you're really committed to making sure everything is dodgable from everywhere.

I absolutely guarantee someone will complain about getting walled anyone even if you have made sure that everything is possible everywhere, so it all comes back to how much effort you're willing to put into things for a community that is never happy with anything.

[zaphod77]

My main idea here was to avoid situations lie I wanna be the music extreme 10 minute mode, where you can and will get walled in through no fault of your own.  Trial and error is a big part of fangames and is to be expected, but only having a chance to beat the pattern 1/10th of the time because it will trap you no matter what the rest of the time is not my idea of fun.

Another thing that bugs me is the A.I. Roulette from hell, where any attack can follow after any other, and some combinations are much harder to survive than others.  One example is that boss with blobs. if it decided not to do the summon a chaser attack you've got it in the bag, but if i opens with that, sme of the other attacks are really hard to deal with with the chaser around.   I finally ended up beating it the first time that attack was unused after i had learned the others.

[Kyir]

I understand your intent behind the thread, I'm just telling you that few game makers are going to invest the time required to properly execute your idea.

[Denferok]

Walls are nowhere near as common as you think, they're too rare to even be considered a problem. Often times you can watch back on an attempt that you thought was impossible and see a way through. These situations are important for you to improve.

Your idea would suggest that the difficulty would always be the same with no variance, which goes back to the original problem where it becomes dull. Sure, with the current system you can just luckbox your way out of it but I personally don't have a problem with that. Getting excited over good RNG only to screw it up later and "waste" it is important for teaching the brain to be less nervous and be able to deal with these factors.

[zaphod77]

Nowhere is it written that you have to remove ALL variance.

If walling really is that rare, then see how many times in a row you can clear the 10 minute stage of i wanna be the music on extreme.

I also get the becoming complacent when the pattern is easy thing, but there's a difference between that and when the pattern being easy is the only thing enabling the clear in the first place.   And in that case, complacency tends to not be an issue.

[just_another_Guy]

If walling really is that rare, then see how many times in a row you can clear the 10 minute stage of i wanna be the music on extreme.

He meant in most decent hard avoidances. No one even cares about that fight anymore, it's considered impossible iirc.

[OhNoezEinPandy]

Take the wall like a man and press "R". If you can't accept that some attempts are straight impossible then don't play it. There are enough people who will keep playing because rng is fine.

[zaphod77]

I got walled in more times then I can count on Genki's true final boss.

Usually i would be attacking the top head, and it would herd me right into the top laser beam out of the triplw. Or an arrow would show up during the big red beam and clip me just as it ended.   or the double green flame would force me to drop down into an arrow or turn the stream pattern that JUT started into an actual wall.  SOmetimes i would be blocked by arrows and fail to dodge the giant beam.

[Sudnep]

I got walled in more times then I can count on Genki's true final boss.

This boss is hectic but extremely easy and there is only one way to truly be walled by it and that's either by doing it wrong which is player error or having extremely (and I mean extremely) bad RNG with the giant laser and arrow/mushroom attack. This is hardly a good enough example nor does it happen enough to be even warranted a change since the chance of it happening is zilch to none.

Even so in the case where the player is 'walled' because it's his first time playing the boss I think it presents more of an epic challenge rather than anything and even if there are those few forced deaths it's hardly challenging enough to bar a player from completing it.

If you actually truly think you're getting walled too often on this boss you're definitely not as skilled as you think you are and probably should suck it up and get better before complaining about 'walls'. Consider a new strategy rather than blaming RNG.