Official Trial Testing

[weasel]

Hey everyone,

It's about time that I make some "official trials". There are going to be, at first, 15 trials to accomplish. They'll ramp up in difficulty, starting out at tutorial-level and ending with something particularly tough.

I'll post some levels I'm targeting for publishing and would appreciate any feedback you may have!

(note: this isn't a thread for people to suggest levels)

So let's get started.

I think this is going to be around #5ish or so:
http://www.protonaut.net/?level=615

[weasel]

And here's a harder one, maybe around level #10 or so:
http://www.protonaut.net/?level=616

[OfficiallyHaphazard]

5: http://www.protonaut.net/?replay=150
10: is hard. Still working on it.

[weasel]

Here's a much easier one, maybe first 5 range:
http://www.protonaut.net/?level=620

[weasel]

5: http://www.protonaut.net/?replay=150
10: is hard. Still working on it.

Here's a win for ya:
http://www.protonaut.net/?replay=152

edit: that level is way too hard.

[Linkzcap]

First Level: http://www.protonaut.net/?replay=160 3rd try

Second: http://www.protonaut.net/?replay=161 Epic Catapult Fail!

[Azxs12]

Here's a much easier one, maybe first 5 range:
http://www.protonaut.net/?level=620



http://www.protonaut.net/?replay=172 I'll try to beat the other ones.

http://www.protonaut.net/?replay=173 Failure

[Linkzcap]

And here's a harder one, maybe around level #10 or so:
http://www.protonaut.net/?level=616

I think this is my favorite level of all time.

So many fails...

http://www.protonaut.net/?replay=209
http://www.protonaut.net/?replay=210

[weasel]

This one is called SwingSets.

http://www.protonaut.net/?level=706

I had a lot of fun making it. It has potential!

[aeiowu]

This one is called SwingSets.

http://www.protonaut.net/?level=706

I had a lot of fun making it. It has potential!

i love this level. i think it really needs to be toned down toward the end though. there are three things that are just too punishing in my opinion.

1. the counter-weight (Fe) throwing you off the last swingset.
2. jumping as you get off to get oxygen on the last swingset.
3. jumping to get the final Hydrogen balloons to win the level.

All three of these things come at the very end of the level, which is already quite a healthy challenge. So getting that far just to get screwed (or at least that's how it feels) by trickery really sucks.

Don't get me wrong though, i think there's a place for those levels but I feel the trials should be a little more lenient. IMO they should showcase the wonderous things that this system can spawn and presenting them in a way that leaves a bad taste in player's mouth should be avoided with these.

I tried to make my first legit level tonight. it's an attempt at making a more casual level that doesn't penalize you for mistakes (falling). it still has the ability to do that though... http://www.protonaut.net/?level=703

note: there's some kind of invisible hydrogen element in that level so it won't let you win unless you get it. It's about 40px just to the right of the second clump of balloons.

[Linkzcap]

http://www.protonaut.net/?replay=48

Iron Fail in Swing Sets

Pretty hilarious!

[aeiowu]

1392: http://www.protonaut.net/?replay=61

still not perfect yet...

ok... here's a better one: 1114: http://www.protonaut.net/?replay=62

[weasel]

BOOSH http://www.protonaut.net/?replay=64
1009

[Linkzcap]

961

http://www.protonaut.net/?replay=73

Puts BOOSH to shame!

[OfficiallyHaphazard]

BOOSH http://www.protonaut.net/?replay=64
1009

Both of this and linkzcap die for me.

[Linkzcap]

Mac?

I use Vista.

[Azxs12]

This one is called SwingSets.

http://www.protonaut.net/?level=706

I had a lot of fun making it. It has potential!

http://www.protonaut.net/?replay=75 Not the fastest but...

[aeiowu]

Both of this and linkzcap die for me.

same.

it's interesting, it seems that the bit with the wall jump in the beginning might be the culprit as my replay seems to work cross-platform up until the final bit with the wall jumping to get the last hydrogen bits.

[weasel]

I believe the problem with cross-processor-replay-incompatibility lies in complexity. The whole chaos effect in motion. Since it has to do with decimal rounding, there will never be a big difference in replays - just a little change here, a little change there...

In a level with mostly solid surfaces and player movements, there isn't much disparity. Since I handle all the player movement calculations myself, there's no random decimals to calculate; the physics is fairly straightforward (except being flung around by bonds or falling from heights). The only out-of-control elements are things that reside mostly in the physics engine - how elements roll about, how bonds bounce off each other, how hydrogen flies, etc... they are full of decimals and calculations.

Those are mostly mitigated, however, because things don't "activate" until you touch them. Most interactions with dynamic objects are short lived, measured in tens of seconds or less. An oxygen atom might roll an extra 1/10th of a meter on a Mac than on a PC, but the end result is the same since the player usually just plows through them all.

A 3000-tick-long replay for tutorial level one should always yield same-results across all processors, for example. There is nothing really that could happen that isn't calculable. I'll bet the only way you could get tutorial 1 to show a discrepancy, is to purposefully avoid everything - fire projectiles at all of them to dislodge them, and wait a really long time (3-5 minutes minimum), then try to precisely capture each (with one-pixel accuracy; don't just walk through them). That should show some differences.

So simple, static levels are more compatible with replays. Anything that deals with a lot of physics - such as Swingsets, where many [glitchy] objects are in motion for a lengthy time, there are going to be some rounding of decimals that causes discrepancies.

Probably the worst discerpancy-causer is Hydrogen. Since hydrogen atoms are active from frame 1 (they aren't "activated" like other items), they have a longer time to progress to an unstable or out-of-sync state. By the time you get to the hydrogen platform in Swingsets, for example, the hydrogen has been a live for 30-60 seconds, and is all bouncing all over and reacting in various ways. Their position won't change drastically - they are on a fairly short tether - but their exact positions will be different by the time you reach the end of the level.

It's good to note that this desyncing happens because of processor types. Macs with PowerPC processors, AMD chips, and Intel chips should see a difference. Newer generation (Centrino) Intels use a different floating point algorythm than the Pentium line too. This is a problem across all the platforms, and short of writing my own physics engine that is (inherently) imprecise, there's no way around it.

Well, ok, I lie: the way around it is... To encode into the replay "keyframes", where it records the current location and state of every object. This could happen at regular intervals - every 5 seconds or so? So the game would essentially "reset" to put everything back into place. The downside to this method is it's very data-heavy; my replays are only around 1Kb or so in size; a regular keyframe could easily push a large replay to the 50-200Kb size.

Incredibots uses this method; they encode the current location of everything around 4 times a second. Their replay files must be HUGE! :/

I don't want to crowd my database with that much junk, and I don't want players to suffer. I also don't believe replays play a huge part of the game... sure they are big, sure they are fun - but I don't think this is an issue that is on the must solve for release list.

[Azxs12]

And here's a harder one, maybe around level #10 or so:
http://www.protonaut.net/?level=616

http://www.protonaut.net/?replay=100 What an awsome replay 100. :toot: