Blade Lengths of Custom Hilts

[Circa]

So anyone that has played JKA MP and used mods has noticed that when you use a custom saber model, that's on the server and in your base folder, the blade length is shorter than using a vanilla saber. It's slight, but it can make a difference and give you a disadvantage. For awhile I just assumed it was cosmetically shorter but perhaps was hardcoded to all be the same length, no matter what, to avoid such disadvantages. But further testing has shown that's not the case.

 

Is this fixable server-side? The length of all sabers should be 40, as shown in the .sab files. Here is a screenshot showing the difference. Its not the perfect comparison due to not the exact same angle, but in-game you can see it a lot better.

 

vanilla saber: (single_9)

 

 

 

 

 

Custom saber hilt:

 

 

 

 

I could make a video if it's needed, but maybe you guys already are aware of this?

 

Edit: to be clear, this is an issue with JKA, vanilla and OpenJK. Not specific to either. Its something I think should be fixed in OpenJK.

[mrwonko]

What custom saber is this? There's no reason a custom saber should automatically be shorter as they are all treated the same by the game, so my guess is that either the .sab file specifically defines a shorter length or else the distance between the tag that defines where the saber is held and the one that defines where the blade starts is shorter.

[Circa]

What custom saber is this? There's no reason a custom saber should automatically be shorter as they are all treated the same by the game, so my guess is that either the .sab file specifically defines a shorter length or else the distance between the tag that defines where the saber is held and the one that defines where the blade starts is shorter.

 

This happens with any custom saber, no matter the setting in the .sab file. The one pictured above is set to 40.

 

I even tested the vanilla Luke saber, since it's not available in MP by default. I set the "notInMP" to 0 to test it. Still shorter as well.

 

Might be the tags, I can test. It will be a pain to go through all the custom sabers used on my server to adjust the tags to match the vanilla ones, but I'll see if that's the issue. If that's the case, how would this mod I made affect things:

 

https://jkhub.org/files/file/2201-fixed-saber-hilts/

 

I only have these in my base folder, not on the server. They don't seem to affect anything. I made that mod because the tag where the player holds the hilt is at the bottom, which is completely unnatural.

[Xycaleth]

I recall the default saber length is 36 in JKA unless you specify otherwise. All of the single bladed hilts specify a length of 40. You can see the default is shorter too if you try using a non existent saber hilt.

 

I just wonder if the double blade hilts make use of the shorter default. I think that can be worked around though if it got fixed in OJK.

[Circa]

Well I opened all the vanilla sabers in Blender to check out the postitions. Seems like they are pretty inconsistent themselves. The katarn saber is drastically shorter than the rest of them, which explains why custom ones appear shorter, even if you don't have the saber in your base.

 

 

 

 

 

I recall the default saber length is 36 in JKA unless you specify otherwise. All of the single bladed hilts specify a length of 40. You can see the default is shorter too if you try using a non existent saber hilt.

I just wonder if the double blade hilts make use of the shorter default. I think that can be worked around though if it got fixed in OJK.

 

I figured the default was 40 since all the vanilla hilts were 40. Staff lengths are definitely shorter, yeah. In their .sab files it says 32.

 

A possible solution would be to hardcode a length somehow starting from the playermodel's r_hand tag? Or a skeleton bone. That way it would be the same for anyone, no matter the hilt being used. As in, make the damaging portion be the same, and the cosmetic blade be separate. So you could have a length of 10 but still get the reach of 40, or a length of 80 and still have the length of 40. Otherwise the tag/position issue is going to always be a problem.

 

Or change how the playermodel holds the hilt model. Instead of using the origin of the model, have it hold the model a certain amount of units below the blade tag. Not sure if that would be a thing you could do server-side or not.

 

 

 

And before someone comes saying "Just use the vanilla models if you care so much!" - we do exactly that. I switch to a vanilla model when I'm dueling, but its nice to use my own saber model that I made. I'm sure others would agree.

[Wolfeye]

Phantom Academy tested and can confirm there are varying degrees of impact from differing sabers. For purposes of testing, we compared the shortest (Katarn hilt) to one of the longest (Firebrand). We made sure to remove as many variables as we could (zero movement, centerview set, and a keybind for attack so that mouse movement wouldn't impact location of the hit). We also had testers of varying pings run the same test to see if that had a significant impact on it, which as far as we can tell, it did not.

First, we looked at the visual elements, which do support this:
https://jkhub.org/images/71cf77f78eb05d557a6a0b6e0c017643.jpg
https://jkhub.org/images/cebcbbec45c4d3168a2b0aed12ddc2d7.jpg
https://jkhub.org/images/12cf820f9da0c8a3fa151a47ebc80acc.jpg

All show a significant length difference in Katarn v. Firebrand.

Then, we took to our formal range testing. At the edge of range for the Firebrand, it was scoring about 6/15 swings. Each test differed in the number of hits though, it wasn't really a strictly consistent thing. (I did an unrecorded test with Defender which is approximately the same, and got 10/25).

For the same distance, Katarn was much less consistent. I believe there was a 3/15 one, but then we had at least one instance of 0/20. In most tests, Katarn proved to be almost incapable of landing a hit.

Demos provided here for reference:
https://www.dropbox.com/s/twvt48jzk93vnbe/Saber_Legnth_Test_%28Wolf%29.dm_26?dl=0  (My perspective)
https://www.dropbox.com/s/60o1zix0pai7p7h/Saber_Length_Test_%28Art%29.dm_26?dl=0 (Art's perspective)

We would conclude that this finding is accurate, though the consistency of the usefulness of the extended range is still up for debate. It is without a doubt that Firebrand/Defender scored better overall than Katarn did, but the hits still were not 100% every time. The only known variables that could impact this would be networking things (snapshots/packets/rate and such).

(Credit to everyone involved in testing where it is due, especially to Frost for being so patient in being our test dummy )

[Raz0r]

Stop comparing visuals. Client visuals do not match 1:1 what is simulated on the server. This is especially true across mods.

I'd be curious as to what happens on JA++, I did normalise a bunch of things like this when allowing client visual overrides.

Can't check into this too much atm, at work. Tags on the saber model and rhand on the player model are both important.

 

Inconsistent results generally means an unreliable test. Recording a demo of an unreliable test does not make it reliable.

[Circa]

Stop comparing visuals. Client visuals do not match 1:1 what is simulated on the server. This is especially true across mods.

I'd be curious as to what happens on JA++, I did normalise a bunch of things like this when allowing client visual overrides.

Can't check into this too much atm, at work. Tags on the saber model and rhand on the player model are both important.

 

Inconsistent results generally means an unreliable test. Recording a demo of an unreliable test does not make it reliable.

 

o.0

 

How else would you compare them, without comparing visuals?

 

How would you propose it be tested, then?

[Wolfeye]

The results were consistent in terms that the Katarn hilt was vastly lower than the Firebrand hilt as far as landing hits in every trial. I consider any variation from the very few hits that Katarn actually got to likely have been the result of unaccounted for variables, but there was a very strong lean in the evidence.

In this case visuals matter. If the saber tag is lower on one, of course the saber will actually be shorter. That said, the visual was just for the sake of checking it, we're vastly more concerned about the impact it actually has in combat. When one saber is scoring sufficiently higher than another, that has a strong implication. 

So we have:
1. Visual representation support
2. Saber tag support
3. Effective impact (higher hit ratio of longer vs shorter) support.

By no means was the slightly inconclusive test meant to serve as absolute fact, but I'd say the theory is very well supported right now. The goal was to provide some degree of physical evidence for the claim Circa made. The demos were meant to serve as information to show the results we were getting and how the test was conducted, so that people may run their own experiments or may suggest reasons why the results we got were not consistent.
 

Inconsistent results generally means an unreliable test. 

 
You'd be right to some degree. What it has to mean is that there is a variable unaccounted for somewhere. Which is why, since every important in-game variable was accounted for (all with centered view, same facing position, unmoving, same swing, same dummy position..) I would posit that the variable missing lies in JKA's netcode--that you and every other coder for the game would admit is horrible--which could cause oddities in how often the saber generates a successful hit. Most likely do to whether or not it lines up with a snapshot from the server, I'd assume.

Because the hits differ in quantity from saber to saber, I'd be tempted to argue that our dummy wasn't perfectly positioned completely out of range of the shorter saber (probably because the additional length is very minimal) and that both were technically hitting it, but the fact that one saber was hitting with slightly more of the blade increased the chance of it registering properly. It's likely that both were hitting, and the hits that weren't counted remained unregistered because of how little of the blade hit and for how short of a duration it was. The hits registering from the longer saber more often would make sense, because it held a slightly more significant impact.

That said, my alternative to fixing that variable would be to increase the sv_fps of the server to something much higher to try and force every hit, regardless of how small, to be registered. Then run a second test. 

This is all just a hypothesis, mind you. I'm not really a coder, so some of that could be wildly off, but to the extent of my understanding of the game, this would be my expectations.