SSR System Shock Remake Feedback Request - Night Dive Studios

[673171d5034a5ZylonBane]

Unity itself is generalist, but what they're doing might make it difficult to port over/update after they drop support. When i say 'their engine' I refer to the bits that make just dumping assets into unity and hitting run not work.

Well, that is certainly the most convoluted, inside-out way I've ever seen anyone attempt to describe custom game logic.

[673171d503639Hikari]

Mildly insulting but alright why did you feel the need to chip your two bits in?

Moving on. Can we at the very least agree a remake would allow them to not have to deal with legal headachs of somebody else legaly owning the source?

[673171d503749ZylonBane]

No. Any custom code is still subject to legal protection.

[673171d5038b7Gawain]

There will not be editor for SS1 Remastered because of Unity's asset pipeline. Making it to support dynamic asset loading (Like in Cities: Skylines) is a lot of work and not needed. Modding using 3rd party tools is a possibility of course, but can be hard because Unity does change the asset data alot. There are some tools made to edit Unity games.

[673171d503a0cHikari]

So how do the game devs build levels?

Color me curious here because i genuinely don't know and unity seems to have been pretty popular as a medium studio engine for a bit.

[673171d503cedAaron]

Mildly insulting but alright why did you feel the need to chip your two bits in?

Moving on. Can we at the very least agree a remake would allow them to not have to deal with legal headachs of somebody else legaly owning the source?

Because he's very pedantic and feels the need to correct minor spelling and grammar mistakes and any mild inaccuracies he sees on an individual basis. I'd imagine it's what makes him popular in social gathering. Everyone loves it when somebody derails every train of conversation to argue about the minutiae -right?

[673171d503dfbAaron]

Keep in mind you are missing an "l" in legally in that post and an "e" in headaches. Not that anybody around here cares except for *someone*.

[673171d503f85icemann]

Shadowrun Returns was done in the Unity engine. Has a level editor for it. Quite a good one, all in all.

There must be quite a considerable amount of other games done in the engine that have editors also.

[673171d50432fBriareos H]

Ugh, how did I miss this thread.
Here are the basics of what I would have in a System Shock remake:

• Make the user interface feel complex, keep the drag and drop but streamline the controls (SS2-like is a good goal to set -- if possible, customizable UI);
• Implement VR with room scale, be aware that there is no better FPS to showcase virtual reality than System Shock (plus the original already did support it so you're removing nothing, heh);
• Do not remove any of the "useless" locations, logs, minigames, ammo caches, etc. of the original because of time constraints. Unexpected bonus content is reward for exploration which is a core concept of System Shock. Allow interaction with useless objects, allow the player to throw severed heads around ;
• In the same vein, keep the database of small descriptions of everything in the world. Have a small unobtrusive text say "broken piping" when the HUD is over a broken piping tile, just like in the original;
• In general, make it feel as if it is the hacker's neural interface that provides all of the information to the screen. Refine the health and energy meters if you need to, to make them more believable (but keep the cyberpunk feel). In the example of broken piping above, the on-screen text is not representing the thought of the hacker, it's representing an embedded computer analyzing its surroundings and determining it is looking at broken piping. You can experiment with giving it some latency. Some interpretation errors. Make the whole interface bug out realistically when hit by an EMP or other interference source.
• Cyberspace is super important, yet I personally would almost remake it from scratch. It's very difficult to give a good overview of how I would like it to look, though. I'm quite fond of how the Ghost in the Shell series manages that look, and having a hacking interface a bit like it could be a nice nod. Here's an example.
  Note that I'll love forever and will purchase several copies of the first game that manages to give me a feeling of flow while hacking. And no, not the pipe flow of Bioshock.

[673171d5046feRocketMan]

I like your broken pipe example.  What about something like this?

[Hacker looks at broken pipe]
... [1 second passes]
"A pipe"
[0.5 seconds pass]
"A broken pipe"
[3 seconds pass]
"A broken copper pipe"

The good thing about this is it has zero impact on playability since the player doesn't have to stare at a pipe for 5 seconds if he doesn't want to but if he does, the description keeps refining itself as you described.  That would bring the cyber interface notion to the forefront.

[673171d50485bfox]

I like all the suggestions but think that the GitS cyberspace is just as useless and disorienting as that in SS.

[673171d504cefOlfred]

So how do the game devs build levels?

Color me curious here because i genuinely don't know and unity seems to have been pretty popular as a medium studio engine for a bit.

You can either build the level with adding blocks and slapping textures on it and so on. You can even sculpt a terrain. But that is not very performance friendly.
So the better way to do it is creating small tidbits in some external model editor and then add them together.
Depending on the game you mostly see a combination of both.

[673171d504fe5ZylonBane]

[Hacker looks at broken pipe]
... [1 second passes]
"A pipe"
[0.5 seconds pass]
"A broken pipe"
[3 seconds pass]
"A broken copper pipe"

I don't see how this is at all desirable. The whole point of useless details like this is that they're fun and enrich the world. Requiring players to have staring contests with walls is the opposite of that. This is something that SS1/SS2 already got right. Object identification is something that will be happening constantly, thousands of times through a typical game, so it should be as plain and unobtrusive as possible.

Now if this were a gameplay function, like being able to get a better analysis of an enemy by scanning them, that would be a different matter altogether.

[673171d50523cJosiahJack]

Unity is a 3D model game engine.  Everything you see is a 3D model, typically made in XSI or Blender or Maya and the like.  This is different from earlier game engines which used BSP generated from "brushes" which were mostly cubes.  Quake and its derivatives such as Half-Life use the BSP approach.  Half-iLife 2 blends the two paradigms.  Most "modern" engines are 3D only woth notable exceptions such as Unreal's csg formed BSP.

Since NightDive is using a 3d model based engine, all game level geometry has to be created in a separate 3D modelling package, then imported to Unity and arranged in the scene to form the game level.

Issues:
Unity does not support per polygon occlusion culling and therefore having the entire game level as one large object is a major performance drawback as the entire level is rendered every frame.
Unity does not support per submesh access properly for 3D models using multiple materials.  This means that each object needs one and only one material in order for the cybernetic interface to retrieve the proper message from in-game raycasts for info such as "molybdenum panelling" and "broken pipes" which meeds access to the material name.

Getting info of the world around, I believe, should be instant and only if the player wants to view it.  Think of all the screenshots that would say "broken pipes" if it was automatic.

[673171d5055e3RocketMan]

I don't see how this is at all desirable. The whole point of useless details like this is that they're fun and enrich the world. Requiring players to have staring contests with walls is the opposite of that. This is something that SS1/SS2 already got right. Object identification is something that will be happening constantly, thousands of times through a typical game, so it should be as plain and unobtrusive as possible.

Now if this were a gameplay function, like being able to get a better analysis of an enemy by scanning them, that would be a different matter altogether.

Yeah but what are we talking about here?  Pretty much useless background objects.  It's just like you said.  It's a useless detail that's fun.  It doesn't require anything of the player.  In the very worst case scenario, they never stop to stare at anything and they get "A pipe" or "A severed head" instead of "A broken copper pipe" and "A severed head of Abe Ghiran".  Sure there's some extra bonus info for the player that they might miss out on but the base information is always there and getting the rest of it is just icing on the cake.  Heck, they don't necessarily need to wait either.  It could just appear after a couple of viewings.  It's just a cute idea that I see as pretty harmless but I don't care if they put it in there either.

If they made it part of the target id implant (forget the actual name of it), that'd be cool too.  I think both are cool because it's extra info for the player that goes along with the cyber interface motif.

[673171d505973Gawain]

When I said editor I didn't just mean level editor but editor in general.
What Olfred said is correct.
Some Unity games do indeed have level editors (Legend of Grimrock is one). And you could say that Shroud of the Avatar and 7 Days to die also has level editor. These work by using predefined pieces or the terrain is calculated runtime from pointdata/pointcloud.

Grimrock has tile based levels, so you basically select what kind of wall is in every direction. Wall models (and rules for adjacent walls) are made so that they mostly fit together. Object position, instance specific data and game logic using predefined operations are also easily red from external file.
Grimrocks level editor was released somewhat after the game. They polished their own internal editor (and most likely changed it to support runtime and made the data to be loadable).
I call all non editor stuff runtime stuff. Since unity basically goes to different state when you run your game inside the editor.

None of these is really the reason why there will not be editor for SS1. I just wanted to give some details before I try to explain the reasons why there will not be.

Unity is designed for multiplatform development and everything is changed to internal formats (these formats can of course be some standard formats like DXT textures) when you import them to unity.  Original files are kept, but are always reprocessed when you change them. These internal formats go through the build process and lots of stuff is optimized away or changed to optimize rendering, load time and memory usage. Internal formats are also dependent on the selected platform.
No custom models, animations, sounds, textures etc. can be added easily after the project has went trough the grinder.

Cities: Skylines has added multiple open source model, sound and texture libraries to load assets at runtime. Part of their business is the user created content.

Since SS1 is tile based, level editor could be feasible to pull off, but I see no business reason to do so.
Unity is basically itself the editor they will use to create the levels, so they do not need to make one. They however will make (or buy) some tools to extend Unity editor to make level building easier. But these are lost when the game is built and to give these to outsiders would mean they need to give alot more (Basically the whole project. Code could be compiled to binary). This is all based on a doubt that they are making the levels using tile based data.
Reason for this is that unlike Grimrock, there is no need for fast iterations of level layouts, puzzles or interactions with the world. They already have complete design document regarding the levels.

Nowadays levels are mostly built from different sized pieces that are individually modelled. Usually smaller pieces are used to create larger pieces (like parts of a corridor) and so on. No custom level formats are used.

[673171d505d53Gawain]

Unity does not support per submesh access properly for 3D models using multiple materials.  This means that each object needs one and only one material in order for the cybernetic interface to retrieve the proper message from in-game raycasts for info such as "molybdenum panelling" and "broken pipes" which meeds access to the material name.

This is a problem I have encountered also years ago and Unity still doesn't have any good solutions to it.
I did make code that finds correct submesh index and hence the connect material index. But this works by going trough the model data until found (model data needs to be available in memory, so no GPU only memory optimization). You can propably find similar code, it wasn't very complicated, just couple of loops. Problems will arise if the model is complicated. Worked fine for my needs.

This isn't the only problem with submeshes.

[673171d505efcGodlike]

@Nyarlathotep

Dismemberment in the game will be a HUGE one of pros, seriously. Think of physics, gravity in original Dead Space where nice feature.
IMPORTANT: Look at these systemshock.org/ttlg.com community fan made content for the Dark Engine games (mods, levels, tweaks). Please remember about level editor !

[673171d5062a2ZylonBane]

Unity does not support per submesh access properly for 3D models using multiple materials.  This means that each object needs one and only one material in order for the cybernetic interface to retrieve the proper message from in-game raycasts for info such as "molybdenum panelling" and "broken pipes" which meeds access to the material name.

There must be some known solution for this, because it's kind of necessary for surface-specific impact sounds.

[673171d5065c0Gawain]

There must be some known solution for this, because it's kind of necessary for surface-specific impact sounds.

Yes there is, but it is a really bad one. Firstly you need to have MeshCollider with the same mesh as the rendered one.
Using this method needs mesh data to be in CPU RAM and also in GPU RAM. So you will lose some memory optimizations there.
You need to check every submesh and its triangles if it contains the triangle the ray hit.
No other solution currently exists. In most cases you dont want to use same mesh for rendering and collision detection, so you need to do some trickery and raycast against disabled colliders...

Edit: One solution is of course to test every triangle in every submesh (until found) using ray-triangle intersection and not use the physics engine at all. This could be optimized using ComputeShader with data already in GPU RAM.. Maybe I should write code for this and sell it in asset store.

[673171d506809Kolya]

There may be use for it.

[673171d506a3aGawain]

I started to really think about this. If you just need to cast from first person, you could do it just like the original System Shock did the on-line help. Render everything to small buffer using different color per object/texture indicating the ID or type. This probably could be done using Camera.RenderWithShader. Then read the pixel color from cursor coordinates. Anti-aliasing needs to be disabled.

[673171d506b77Kolya]

Didn't the walls in SS1 have many different colours? How is the colour an identifier?

[673171d506e45Gawain]

Didn't the walls in SS1 have many different colours? How is the colour an identifier?

The color is the identifier for that wall/texture. Since System Shock uses maximum 64 textures per level only 6 bits from 32bit color buffer is needed of every wall texture. Rest of the bits are enough for every object type in the game...

Edit: just to clarify. Each wall and object type is rendered with different color. One is red, other is green.