We're there, Fallout's lowest point

No idea, haven't been modding FNV for about a year or two now.
If the paid mods work well enough to be profitable I don't see why Bethesda wouldn't add them onto Fallout 3 and NV, especially if they decide to port them like they did Skyrim.
 
No idea, haven't been modding FNV for about a year or two now.
If the paid mods work well enough to be profitable I don't see why Bethesda wouldn't add them onto Fallout 3 and NV, especially if they decide to port them like they did Skyrim.

Guess at that point I will have to switch from the Steam version to the GOG version as I do not want several GB of extra stuff on my hard disc I am not using and would actually have to pay for to access it.
 
Why not just use the FO:New Vegas engine for a total conversion, using the provided mod tools? They give you permission to use the IP.

Personally the Gamebryo engine that Fallout: New Vegas uses is complete and utter garbage. I honestly feel bad for Obsidian Entertainment for having to learn how to use it. The ancient engine is impossible to optimize not to mention the myriad of other outstanding issues it suffers from. The engine is mostly rubber bands and glue at this point in time, to be honest it's like a sad dark joke. (The leper of the game engine world...)

Personally I would use the Source engine or Source 2 (if it ever came out). I would choose these engines simply because optimizing the game itself is super efficient compared to any other engine out there that I know of. Plus if the developer knows what their doing (Like me) they can do some really cool stuff with the engine. Hell of a lot more than what is possible in the Gamebryo engine. If your curious create a new entity in Gamebryo, and a new one in Source, you'll notice the difference in difficulty immediately.

Ever notice how all of Bethesda Games Studio's titles built on the Gamebryo engine are super duper janky, full of bugs, and suffer hilariously derailing game braking issues that I'm sure we could sit here for weeks pointing out? While I do admit every engine has it's faults and the Source engine is no perfect angel, it does have by far one of the best suite of tools to optimize and debug it. Some may not feel that way, but from my experience the now old Source engine is still capable of some awesome feats. If you need proof just look at some of the games, mods, and even cinematic movies made within it.

Personally my approach would be to leave Fallout first person, but bring back the travel system of Fallout 1/2. In this case I could effectively provide a rather high degree of visual quality, and a great many unique locations for players to explore. For those familiar it would be like taking the approach of Super Mario 64 to save space and preserve performance. Or I could go back to the oblique cavalier projection perspective, and focus on turn based mechanics. Both are easily possible within the engine.

However my point was that someone can provide an obvious higher quality product that could be compared to the establish Fallout brand. In other words "to throw egg in their face and eat their cake".
 
Are there any built-in debugging tools for the Gamebyro engine or is it all reactionary and based on papyrus log analysis from when your game kicks the bucket and CTDs?
 
I have no qualms with the Gambryo engine—as I understand it Bethesda has a very customized version... one that (buggy or not—and it is)... is WELL suited to making RPGs. It's ironic that they don't make RPGs with it.

One notable problem I have read about [IRRC] gave me the impression that the engine does not seem to support the crossfading two animations—or if I don't have that quite right, it means that you seemingly cannot (for instance) play a diagonal running animation. My guess is that if you press the key for forward, you get the animation for forward, and the side-step keys slide the camera sideways; left or right.

Funny though that Larian HAD diagonal running animations in Divinity II.



I don't think the issue is Gamebryo per se... I think it is Bethesda's custom version of it; which for whatever reason, they don't fix or change it.

Regardless though, the other engines don't grant permission to use the Fallout IP in one's mods.
 
To be honest the Gamebryo and Creation engines are awful for open world role playing games.
Here's why I consider the Gamebryo and Creation engine to be a poor engine for games in general:


All entity meta-data is loaded in a similar fashion to the player entity. So states like basic attributes, skills, inventory, etc... are constantly parsing instead of being called upon when needed. Basically the engine wastes a lot of resources on absolutely useless tasks.​

Cell Based loading
; This significantly lowers the amount of entities allowable in a loaded cell aka a hard limit. Anything that surpasses the limit will basically break the engine resulting in game breaking issues or a crash.

Modular everything, basically the entire engine is a jigsaw puzzle that is stuck together with rubbers bands, tape, glue, gum, and anything else that could be used as an adhesive. If a module is updated, it broke another, then that get updated... then it continues on cascading through them all. Instead of building a comprehensive system that is efficient and easy to debug, they turned it into cooked spaghetti and dumped it on the ground.​

The engine doesn't support in engine animations, or any real features observed in the games. They are all produced via modules... that must be loaded and remain loaded individually. To give you an idea forward kinematics, inverse kinematics, translation, nor deformations. So stuff like a train car in Half-Life 1 are impossible to program into the engine without using workarounds, hacks, modules, or by exploiting bugs. For example imagine a single entity, except all the variables that determine it's actions are exterior to its programming. In this case separate controllers are applied to the entity to get it to function. Instead of programming in the required information into the entity itself. Remember the Broken Steel tram? It was a NPC character wearing a hat running a path. The NPC, train hat, paths, and trigger initiators all had to be loaded, instead of a path and a single entity...

Height map terrain generation, literally one of the most annoying ways to make terrain. In this case the terrain can only deform on a single axis, which means static assets that use a lot of computational power must be used to make simple terrain like cliffs.​

Literally no optimization tools, look at Fallout 4's Sanctuary. The entire area is just a bunch of static assets layered on top of each other, wasting all your computers resources to render something you're not going to see.

Most of the engine's operations are tied to the frame rate... *sigh* like physics calculations... So if the frame rate drops or increases, so does calculations the engine does. It's like playing an old DOS game on a modern PC without adjusting the CPU speed.​

The lighting within the engine is so out dated goldsrc (The engine the original half-life used) is still considered better. Plus the module for lighting is so inefficient that a good deal of computer calculations get soaked up by dynamic and static light sources. There is also a max limit of light sources within a cell... Heaven have mercy on the soul of the modder or map designer that placed one too many lights in a cell.

The engine constantly trips over itself trying to load separate modules, this often results in conflicts or weird easily preventable bugs that directly impact game play. Kind of like backing a horse up a mountain side. A great example of this is Fallout 4's bug of missing inventory items that still weigh down the player. Which is caused by a desync of modules that are intended to operate in tandem.​

It still uses a manual NavMesh system, Instead of programming an entity with a basic AI collision detection calculation for pathing, it uses a system that requires the entities to rely upon an actual separate mesh, that overlays everything the NPC can traverse. While some engines use navmesh systems, they instead only mark faces and vertexes of existing geometry for parsing without having to generate completely separate faces and vertexes.

Spawns and gibs, every time something is spawned it has to be fully loaded as previously stated, this also includes items and gibs. When something gibs the spawned entities often spawn inside each other and require a collision calculation to "bust apart", instead of using dynamic slices based on the existing mesh and having the engine calculate the slices. Plus the sudden extra calculations cause frame rate drops, and bugs quite often.
I could keep going but :falloutonline:


To be perfectly honest the engine used to be ideal compared to what was available on the market/capable with previous hardware and software. However it's now hopelessly obsolete. The unity engine even out performs the Gamebryo and Creation engine, and I also very much dislike Unity as well because of its haphazard modular nature.

Personally if I was choosing an engine for Bethesda I would have licensed the one of the following engines: Unreal Engine 4, ID Tech 6, Cryengine V, Frostbyte, or Source SDK. While some of them would indeed be quite the learning curve it is possible for a company to build their own developer tools for them or use the existing ones for them. Plus when licensing an engine the company that made the engine can be solicited for developer information to help their team understand how to use it if it isn't readily available on the internet. However depending upon the engine it can get very pricey...

I understand someone is likely to make the comment about the engine being good for open world games, it's not at all. The other engines can also provide an open world experience by cordoned loading of preset locations that preload when the player is approaching it's boundary prior to the player transitioning. This isn't hard, it just requires programming.

The power to make an engine compatible for a role playing game isn't the engine itself, it's the programming applied to it to define the game's rules. The engine only defines the 3D environment in which the player character exists.
 
Last edited:
Don't they effectively own ID-Tech 6?

I don't disagree with any of your points; though I don't see the Navemesh issue as a bad thing, or that much of an issue. I like that I can use a very simple custom mesh to indicate the traversable areas. I think the AI's poor handling ladders is a worse—but I suspect also related—issue.
example.jpg
 
Depending on how F76 goes they might go Battle Royale. That'd probably be the lowest point.

Fallout: Battle Royale.

Microtransactions to get physical mods that they say don't affect gameplay but absolutely do.

Power Armor is $19.99. Per suit. Per game.
 
Don't they effectively own ID-Tech 6?

I don't disagree with any of your points; though I don't see the Navemesh issue as a bad thing, or that much of an issue. I like that I can use a very simple custom mesh to indicate the traversable areas. I think the AI's poor handling ladders is a worse—but I suspect also related—issue.
example.jpg

I can understand your perspective, back in the day the Navmesh style used wasn't "bad" just wasteful. However there are much better ways of going about it. Consider for a moment you build an open world game, that needs a navmesh....

I'll boil down for why such a system is "bad" now.
  • Each Navmesh in the Gamebryo/Creation Engines is an overlay. Which means it's an entirely separate mesh to the actual terrain, static, and dynamic assets. In this case resources for loading the scene are increased due to the inclusion of an invisible static asset that overlaps the existing terrain.
  • The Navmesh even when invisible requires in engine calculations to remain coherent with the scene. Even though it's invisible it still has faces, and verticies that require constant checking to ensure their position in the game world.
  • Such a Navmesh system is, a band aid that far too many engines/developers rely upon because they can't be bothered to optimize their product.
  • Navmesh's are plagued by the same issues static faces are. If there is a seam between faces or there is a deformation that splits the mesh then anything traversing the mesh will bug out, or get stuck. It's kind of like an invisible hole an entity can fall in. A great example of this is enemies getting stuck inside rocks, and under terrain. It's the not the terrain's fault, it's the navmesh.
To boil it down it's like a train, every train has to stick to a rail to move forward, except with a navmesh the train's rail is everywhere.

Here's a better way to go about it:
  • Let the AI cheat and know everything about the map, from there it can make the most optimal choice of how to travel an area to it's destination.
  • Another way is to program to AI to recognize local assets and information to determine their best path.
  • The wandering way point method, in this case a location is defined as the "center" of a location and from there the AI can use the way point anchor to wander about the area. A great example of this is Alien Isolation. This also allows the AI to support additional directives and can be even given preset tasks to complete while in the region.
  • The "WHERE IS IT" calculator approach. In this case the AI determine where an entity like the player is, and then requests information on how to reach that entity. A great example of this is Minecraft.
I could keep going, but the point is the current navmesh system used by Bethesda is very much out dated and wasteful. Just because someone has 16 gb of ram doesn't mean they need to use it all.
 
The Creation Engine is utter shit. When I modded for it, I guarantee it crashed more than the actual game did.
 
The Creation Engine is utter shit. When I modded for it, I guarantee it crashed more than the actual game did.

A Buddy of mine who used to mod Source with me tried to get into modding FO3, and NV a long time ago. However coming from an engine that is actually functional to the lick and hope it sticks nature of the Gamebryo was jarring.

He's a dedicated dude, but after digging into the engine and the development tools he was left scratching his head by the very odd nature in which much of it was designed. While the physics side of the engine was fine, everything else was tacked on seemingly as an after thought instead of being established into the engine itself.

He did later toy around with the Creation Engine, but quickly gave up on that as well as the obstacles he encountered previously from the Gamebryo were still there.

Our running joke is that Valve spoiled us with their programming library and development tools.
 
Yep. Bethesda put together a halfass mod tool that is noticeably broken and they get praised to Oblivion...until the paid mod Fallout corrupts their Elder Scrolls and they are forced to release another Skyrim port. Almost makes you wish for a Daggerfall to the head.

Morrowind. Left it out.
 
I think they can't just switch over to idTech 6/7 for Fallout 5, their modder base would have to get used to new tools and then it would take too long for them to fix Bethesda's crap, and that again would make Bethesda look bad.
 
The main issue likely will be if Bethesda have the capability of learning to use a new engine, i remember they haven't use anything else other than Gamebryo ever since Morrowind?
 
Indeed. Bethesda only has experience with one engine.
As for ID Tech 6 Bethesda Game Studios would actually need to license it from Id Software even though they are under the same umbrella of Zenimax Media. I would think there would be some heavy discounting though.

Either way if they did they would likely require ID Software to stop production of their products for a period to train up BSG to use the engine. This is actually the cheapest and safest option for switching engines for them.

The real problem with Bethesda's developer tools is that their not really a SDK, but more a long the lines as a hacked in editor one would see modders make themselves for really old games. Look at the old map editors fans made for ancient RTS games.

If they did switch engines and wanted to make a development kit then they would have to put in time and money to develop them.

Honestly it would be better for BSG to stop development, switch engines, hire the programmers to develop their tools, learn the engine, fix the buggy tools, and then begin production again. Until they do that their stuck with the Creation Engine. Provided of course the license agreement for the ID Tech 6 engine allowed modding tools to be generated for public use.

It really just comes down to time and money, and considering BSG only knows one engine. Getting their employees to learn a new engine would likely encourage lay offs, new hires, some employee shuffling between studios, and a considerable amount of down time to get their house in order. If Fallout 76 is somehow a success they would likely have the funds to support such a venture, if not then there wouldn't be enough buffer capital.

The development of such a tool set, and all that work would be like funding an entirely new game. Except in this case they need to do a much better job to ensure their tools are on par so that they can develop in line with a future game. In this case the development tools would "change" through various iteration as they developed a new product for the first time. However if planned and developed well it could see a massive rise in modability.

To be honest however BSG is never going to retool their entire studio to more functional means (Money Talks, Ideas Walk). That being said I would fully expect BSG to suffer significantly from lagging so far behind other developers due to their exceptionally stingy nature with upgrading and training. Even Fallout 4's release was met with less than luke warm reception from modders and other developers a like due to it's "new" old capabilities.
 
I was under the impression that Gamebryo itself was only a renderer technology, and that the rest of the Creation engine (and other in-house engines based on Gamebryo, were necessarily comprised of modular features that are purpose built; included [tacked on] to suit the specific game or application.)

@Einhanderc7 As to the Navemesh, I took it to be (very) loosely similar to how convex hull calculations work for game physics systems... Allowing for cheaply approximated collisions; inaccurate (but close enough).

Do these other traversal methods allow for the same minutely detailed navigation? (Like for instance, that the designer could allow NPCs whose capsule is small enough, to slip through the legs of a statue, or for a child to slip through an alley too thin for an adult. [etc]

___

I can certainly understand the reasons for it, but I have never been a fan of omnipotent AI, because (as you'd expect) it makes obvious choices based on impossible perception, and perfect foreknowledge. One's own NPCs would know the layout of a dungeon that the party just discovered.

IIRC though... Warcraft (1or2) did this, but they had the Shaman cast the Evil Eye spell (visible to the player) to make their behavior seem plausible. ;)
 
Last edited:
@Gizmojunk
If an entity requires intelligent movement in an area the best approach would to only allow the entity's AI to perceive the environment like the player does (Through programmed response to information). In this case the entity would determine if movement is possible by comparing it's determinate state such as a collision box with the area around it. If the box fits, they fit.

As for programming an entity to react to stimuli, they have to cheat, and then be told they can't cheat. I know that sounds weird, but bear with me here. Alien Isolation is a great example of this:


Another great example of AI programming is Doom 2016's protocols that dictate enemy behavior:


Realistically it depends upon what the entity needs to do, and how it understands the environment around it. This basically means the entities need to be programmed well, to function well. Which is a significant hurdle in the Creation Engine that relies upon a modular system.
 
Back
Top