Looking for updates? Check the comments below.

Also Join us at our forum!

Bleach:Zanpakutou Senshi is a totally free Total Conversion of Half-Life 1. It´s based on the japanese hit Anime Bleach.

You will be able to play as Kurosaki Ichigo and Grimmjow Jaegerjaquez for the first release.


This game features:

  • Open sword system Hack 'n' Slash
  • Dashing and dodging
  • Blocking and deflection
  • Kido attacks and bindings

This game is under development and we don't know when it will be done yet.

This mod is not related to Studio Pierrot or its production.
© All Bleach related characters, locations and other copyrighted materials remain the property of their respective owners.
Follow us:

Image RSS Feed Latest Screens
Grimmjow combo tree Implementing custom animations (outtake) The Fifth Tower Showcase
Blog RSS Feed Report abuse Latest News: Update #22: Progression with the melee system

3 comments by eliasr on Apr 20th, 2015

Back in February we were able to show Grimmjow doing a single attack, with no delay between the attacks and the animation wasn't even part of what you will see in the combo tree. ( see: (Very) Early melee implementation )
Since then we both worked with the melee implementation and support for Linux servers. We made good progress on both parts, but we are not done with them yet.

Melee update:

For starters, there have been quite some changes since the last news we posted here. Including a very basic concept of the melee implementation. In the last news update we talked about how we wanted to prevent the melee system from becoming too dull, by using some small reiatsu related attacks that either was ranged or had areal effects.
These effects can take quite some time to implement, as both the behaviour needs to be written and the graphics also takes some time to make.

While we do have an idea of how to support effects, ranged attacks and AoE attacks will most likely not be implemented as part of the melee system.
Not only does it require time, but it also adds complexity to the melee system. If we plan to include an ranged or AoE attack, then it has to be working properly before we can fully test the melee system.
However, should we get time to make these attacks, then the attack will first become a part of the kido system. If we then consider it stable for being part of the melee system, then it can be added there.

Charged attacks, clashing and stamina are also postponed for now, but these requires less work, so we will surely keep them in mind as we progress.

Since February we have been able to create the combo nodes, which will contain the information necessary to perform the attack, and use them in the melee system. The results can be seen in this video:

As seen in the video, the synchronization between animation and attack isn't implemented. This will be resolved when each attack has a delay value assigned to it. The video was recorded back in march, but visually the melee system resembles that.

Meanwhile we have been testing whether the client prediction was necessary or not. The result was that it improved the smoothness of the animation when the latency is high. Some of our former developers didn't include that in their weapon development, and as such we are currently updating them to include that.

Last week we held a meeting, where we planned the combo tree for Grimmjow. (Ichigo's will come next)

Grimmjow combo tree

The RRR combo is shown in the video. There will be updates to some of the current animations as well as new ones. Getting this done has made it possible to continue working on the attacks codewise while Angel makes the animations.

Codewise the combo tree now supports a value to define attack delay as well as an effect identifier. The effects themselves has yet to be coded, but since they resemble each others it won't take too long to make.

Another thing we are working on now is to be able to traverse the combo nodes in a tree. It seems redundant to write a system which builds a tree, compared to simply writing the tree manually. This is due to the tree being relatively small and the attacks are all static. You can read more about it here: En.wikipedia.org

So to sum up, our next milestone for the melee system is to synchronize attacks with their animations, traverse the combo tree based on button clicks and add more animations. The effects will come at a later stage.

Linux support:

If it is easy to support, then one should support it.
It isn't the first time we considered making Linux support, but it is the first time we are actually trying.
With aid from this article at the wavelenght: Articles.thewavelength.net , Debian 3.0R6 and a windows batch script running dos2unix. We have been able to convert it all to ANSI c++ and compile a .so file.

That took a bit of time...mostly due to the requirement of gcc 2.91.4, which does not work on newer Linux kernels.
Unfortunately Steam_CMD have been giving us a headache and therefore we haven't been able to test it yet.

The good part however, is if we get it running, then we will maintain Linux support for the mod. We know a lot of web servers running the old Goldsrc engine, runs on Linux. And we know it is important to support that, as Linux servers are so much cheaper than windows servers.
So we will do our best to resolve the issue and get it working.

That is it for now. Joshua might turn up with some images from his maps at some point.

How do you like our videos? We don't have any one on our team who is a magician with video editing, but we try to keep them relative small and content related. Feel free to give us feedback.

In the past 3 months we have progressively written small updates in the comment section. What do you think of that?

And don't forget to follow us on Facebook. We have updated the profile and made it easier to maintain the page as well.

Thanks for watching and following.
See you next time :)

Media RSS Feed Latest Video
Downloads RSS Feed Latest Downloads
Design sheet

Design sheet

Apr 24, 2012 Other 8 comments

This is the design sheet for BZS mod. Latest version: 26 April 2012.

Post comment Comments  (10 - 20 of 568)
eliasr Creator
eliasr Mar 24 2015, 9:39pm says:

Due to a midterm examination project, coding will be on halt until April.

+1 vote   reply to comment
eliasr Creator
eliasr Mar 24 2015, 9:38pm says:

We were able to conclude that client prediction did make attacks seem more smooth, compared to attacks without. As such we will have to incorporate that with the kido attacks.

Thanks to Joshua Riches for dedicating his time to test this.

Test details:
The test were conducted on a dedicated server, with one client on LAN (1 -5 ms ping) and one client on the Internet (~ 400 ms ping).

Test 1: no prediction
No errors was found in the usage of the weapon. The delay between animation and received damage was so short, that no delay was noted.
The predetermined delay between attacks was also confirmed by the internet client.

As such, omitting client prediction would not indicate any serious issues.

Test 2: Client prediction
As Test 1 there was no errors found during testing. However, the attack animation was considered to be more smooth on the internet client, with client prediction.

Test conclusion:
A large latency can cause a lower quality when client prediction is omitted. While this test covers a very large latency, the test does not include more than 2 players.
To provide the best game experience, and avoid possible unforeseen consequences, client prediction is the best solution.

+1 vote   reply to comment
eliasr Creator
eliasr Mar 18 2015, 10:23pm says:

Was able to resolve some minor setbacks and update the list of several issues.
While there aren't any signs that the client prediction has any effect while on a fast LAN, it might still have an impact on larger distances.

Hence a testing session between Joshua and I will have to be done before continuing with the melee system.
Meanwhile there will be some other bugs I can work on, even though I don't have much time before April.

+1 vote   reply to comment
eliasr Creator
eliasr Mar 17 2015, 3:44pm says:

So far current testing shows:
Really bad to have two different versions (client and server) when playing. That needs to be resolved.

Melee attack provides damage.

Currently investigating:
If client hosted games and hlds hosted games are identical.
client prediction vs server globals.

Since these tests requires multiple players, it will take some extra time.

+1 vote   reply to comment
eliasr Creator
eliasr Mar 11 2015, 11:08pm says:

While steam-cmd dosn't exactly help on linux by appending a relative path to an absolute path, it might be that the software center will install it slightly different.

Right now I got the windows hlds running, and it just reminded me that there is quite some tasks to be completed.

There is an annoying bug, that makes the character flicker when it rotates; I'm trying to deduce whether it is related to prediction or something else. Right now it seems that it isn't about the prediction.

It is however, also important to decide whether there is need for weapon prediction or not. We already know that using the global timer on the server to set the delay between attacks works, but we don't know if there are any consequences of using client prediction vs. not using it.

Keeping it all on the server could result in players with lag or slow computers, has the same experience as those without. It could also result in less ability to cheat. (clearly we haven't tested that)

But client prediction could allow optimization, that provides a more smooth experience, even if there is lag. But this do also open up for possible abuse, such as speed hacks.

However, I will not concern myself about cheating. If the prediction of the weapons has any significant impact, then we will use it.

My semester project is beginning to take form, and i'm yet again way behind on my reading schedule... (wonder why) - It is possible that the upcoming weeks will have less progress.

+2 votes   reply to comment
eliasr Creator
eliasr Mar 2 2015, 9:40am says:

Phew, finally something seems to be working the right way.

I changed linux to the 3.0r6 woody (linux 2.2 kernel) which isn't fully supported by VirtualBox... After the longest installation process in my entire life (including windows 3.1), Lack of x11 drivers, lack of support for smb, recursive sftp, numb scp , an ignorant lftp AND incompatibility with dos2unix converter... I took the dos2unix converting to the windows platform, compressed the result to an .iso so it could be mounted as cdrom (at least there was support for that) and BAM Linux 2.2 and GCC 2.95 does it job without standard library issues.

One step closer to linux supported servers.

+1 vote   reply to comment
eliasr Creator
eliasr Feb 27 2015, 11:03pm says:

Compiling for linux is ... lets just start by saying that i'm happy that the debian archive had gcc 2.95. But there are still a lot of manual labour

+1 vote   reply to comment
eliasr Creator
eliasr Feb 26 2015, 1:35pm says:

Today a huge update was made. I wont go into all the details, as that could take up to an hour to write, but overall it looks something like this:

Several bugfixes has been done, removing 2 crashes and resolving a sound bug in one of the kido attacks.

The melee system now get the animation from the combo tree. Correct progression though the combo tree still needs to be implemented.

Added 3 animations to demonstrate WIP progress in the melee system.

A few issues were also found and it will take some time to deal with them. HL uses a event system, which i'm still unfamiliar with. This system might handle some prediction for the client, and in such case it might also handle delay between attacks.
The issues lies in, that the code which originate from the SDK adds a delay to a time, which either is the server time or 0. The definitions (also standard in the sdk) chooses this to be 0 and this in result causes no delay, as the server time is far ahead.

The current workaround is to enforce the use of adding the delay to the server time. This works for the host, but we need to make sure that it doesn't provide problems during actual multiplayer. If it does, then we will have to handle this, before we can look at synchronization between attacks and animations.

Joshua have also uploaded his latest version of "the fifth tower" map.
We will add some graphics soon.

+1 vote   reply to comment
eliasr Creator
eliasr Feb 19 2015, 7:48am says:

To follow up on the last question. It is most likely possible that we are reducing the amount of features, which we want to implement, as some of them aren't trivial. In that case, it is reasonable to consider us at least 60% done.

Right now our focus lies in implementation of the melee attack, which needs to get the info from a player's combo tree. At the same time, we need to sync between attacks and animation, so the attack happens around the middle of the animation.

This is currently a part which we will have to work out with the animator.

Recently I recorded a very very early implementation of the melee "system", if you can call a forced animation and close ranged attack that...
It will probably be uploaded soon, but for the sake of consistency of future videos, you will have to wait till we agreed on it in the team.

+1 vote   reply to comment
lightningterror Jan 31 2015, 1:17am says:

On a scale from 1 to 100% how much did you accomplish till now ? Just curious cuz I'm dying to play this mod xD Well keep up the awesome work guys.

p.s. I think it would be a great idea to show us a video let's say 1vs1 to see how the mod works and stuff.

+1 vote     reply to comment
eliasr Creator
eliasr Feb 2 2015, 2:08am replied:

It will be easier to tell after our general meeting.Perhaps 40% or 60%.

We might be reducing some features, as the effective work being done, doesn't cover the needs to fulfil everything with a satisfied quality.

Right now we can only cover 2 kido attacks (perhaps 3) with many unimplemented animations. We sorted out how to handle the animations, which would be rather fast to implement by now, but it would not provide the full picture of our progress.

Joshua and I once talked about making a teaser, which briefly would show off the ichigo and grimmjow model, and a kido attack. But we are aware that this is not really what are being requested of us.

Earlier we had some trouble with animations being WIP and the developer requesting to retain these from public view. While Angel have made the grimmjow model and done a great work with the animations. We need to check up on the animations on the ichigo character.

While I'm not a fan of the request to retain work, I still find it important to acknowledge the view of the developers and ensure their credits are due.

I might be able to bring some more details the 16.th February. I will follow up on this question; if you don't see an answer by that time, then feel free to bump the comment section.

+1 vote   reply to comment
Post a Comment
click to sign in

You are not logged in, your comment will be anonymous unless you join the community today (totally free - or sign in with your social account on the right) which we encourage all contributors to do.

2000 characters limit; HTML formatting and smileys are not supported - text only

Half-Life Icon
Send Message
Official Page
Release Date
Mod Watch
Track this mod
Third Person Shooter

Latest tweets from

It can take up to a few hours for tweets to begin appearing.

Embed Buttons

Promote Bleach:Zanpakutou Senshi on your homepage or blog by selecting a button and using the embed code provided (more).

Bleach:Zanpakutou Senshi
Bleach:Zanpakutou Senshi
Last Update
2 weeks ago
406 members