A big problem I foresee with KSP 2: pc resources

[Lewie]

On 7/13/2021 at 7:37 PM, linuxgurugamer said:

That mod is very system dependent.  Some systems it helps, and some it doesn't 

Wait....what’s this mod called? Sounds pretty interesting 

[Lisias]

18 hours ago, K^2 said:

For a personal or indy project, absolutely. Go Godot. That would honestly be my first recommendation for anything a team of up to about 10 people or so can make. (...)

I wonder exactly where and when Unity made that wrong turn. This is what Unity should be pursuing, it's my understanding that this is how Unity started. 

 

18 hours ago, K^2 said:

Purely in context of KSP2, I don't see any engine on the market that would work better than Unity. (...)

So the Game Industry is already walking fast into the verge of a collapse, more or less as it was happening on 1983/4.

One of the signals of such market disaster is what I will call (mainly by lacking the proper language skills) the "squashing" of the production chain. The really big and few ones and the really numerous small ones are squashing the middle ones,  that are the ones where most of the really good and innovative titles comes from.

The small ones never manage to really score a big hit,  because they lack the resources for such.

The big ones stagnate on their own mediocrity, protected by their ability to exercise their muscles to supress/harass/buy the competition, instead of exercise competition itself.

The net result is that sooner or later you flood the market with bad products. As it happened o  1983/4.

The present distribution mechanics should prevent things to go south in the exact same way it happened in the past, no doubt. The Stores may force a shift in this balance, but I think this will more likely postpone the unavoidable. Without the middle game publishers, the big ones will just mate with the biggest stores and exercise their muscles from there.

I'm not exactly thrilled with a potential career on this industry by now.

 

18 hours ago, K^2 said:

You can complain about Unity, and I'll pitch in my laundry list of grievances about engine, documentation, and company, but there isn't actually a good alternative for this size project with this size team.

Our loss, IMHO. The problems caused by Unity are a clear and present danger to gamers. If we have no viable alternatives to it, things will keep going as they are now.

And they are far from being good. Really, really far.

[Alexoff]

I am worried about the performance of KSP 2, since recently we were shown a video with a new planet, in which there was a low FPS. The fact that the developers do not hesitate to show us videos with low FPS leads to disturbing thoughts.

[Incarnation of Chaos]

21 minutes ago, Alexoff said:

I am worried about the performance of KSP 2, since recently we were shown a video with a new planet, in which there was a low FPS. The fact that the developers do not hesitate to show us videos with low FPS leads to disturbing thoughts.

Literally could be the difference between the test build and production builds. 

I really want people if nothing else to caution themselves about taking any definitive conclusions about KSP2 until it actually releases. 

If it performs bad, it performs bad. You still have plenty of time to save up for a upgrade in the meantime.

[K^2]

5 hours ago, Lisias said:

I wonder exactly where and when Unity made that wrong turn. This is what Unity should be pursuing, it's my understanding that this is how Unity started. 

I don't have the full story, and probably nobody but their top execs do, but there have been several key turning points, I think. First, Unity chose to monetize via asset store. Which is questionable choice if you're trying to make a good generic game engine for small size professional teams. Even Roblox has a better strategy, combining asset store with in-game micro transactions. Second was their boom in mobile, simply because there was very little competition. Finally, they branched out into non-gaming applications. Do you know that there are amusement park rides that run on Unity? Yes, I'm scared too. Fortunately for Unity and unfortunately for the rest of us, they have taken in enough investments to the point where they didn't have to compete on profitability for a very long time, and maybe still don't. Which means they don't have to have a good business model. They just have to keep growing the customer base.

6 hours ago, Lisias said:

So the Game Industry is already walking fast into the verge of a collapse, more or less as it was happening on 1983/4.

Yes, but this has been the case since mid 2000s at least. The only reason we haven't had a gaming crash yet is that we are in new territory of market growing so fast that the threshold for disaster has been moving away at equal or faster rate.

But yeah, we need new alternatives and new companies making better choices, because the current situation isn't infinitely sustainable. Eventually, the games market will not be able to grow further.

[Lisias]

1 hour ago, Alexoff said:

The fact that the developers do not hesitate to show us videos with low FPS leads to disturbing thoughts.

The SpaceX's Dragon capsule once exploded on a test, then the video was published and a lot of people had disturbing thoughts.

But let me tell you, that explosion was a bless. They found a inherent flaw on the fuel system, and by knowing that a lot of resources and even lives were saved, as that thing was being considered on some other crafts too (not only by SpaceX).

You will have terrible FPS on code compiled for debugging, as lot of checkings is being made befir3 executing every single task to detect any mishap that could screw up something later.

Fail early, fail often. This is good engineering - hidden flaws are flaws that will not be fixed. If they are showing us their 'flaws', it's because they know they are going to be fixed.

 

44 minutes ago, Incarnation of Chaos said:

I really want people if nothing else to caution themselves about taking any definitive conclusions about KSP2 until it actually releases. 

Users do what users do. :)

I would prefer people to openly address any concerns they have here than whispering them somehwere else, so someone can explain them when to worry and when do not - and eventually get an early warning about one detail or two they may had missed.

In the very end, users have the ultimate knowledge about how the product works - as they are the ones using it 24/7 - there's no test team in the world that can match such working force.

[Incarnation of Chaos]

11 hours ago, Lisias said:

The SpaceX's Dragon capsule once exploded on a test, then the video was published and a lot of people had disturbing thoughts.

But let me tell you, that explosion was a bless. They found a inherent flaw on the fuel system, and by knowing that a lot of resources and even lives were saved, as that thing was being considered on some other crafts too (not only by SpaceX).

You will have terrible FPS on code compiled for debugging, as lot of checkings is being made befir3 executing every single task to detect any mishap that could screw up something later.

Fail early, fail often. This is good engineering - hidden flaws are flaws that will not be fixed. If they are showing us their 'flaws', it's because they know they are going to be fixed.

 

Users do what users do. :)

I would prefer people to openly address any concerns they have here than whispering them somehwere else, so someone can explain them when to worry and when do not - and eventually get an early warning about one detail or two they may had missed.

In the very end, users have the ultimate knowledge about how the product works - as they are the ones using it 24/7 - there's no test team in the world that can match such working force.

Users can't test it until it goes live i'm afraid, otherwise i'd 100% agree with you. But i don't doubt KSP2 will have many bugs discovered day 1 by adventurous users.

I DO wish they'd actually say somewhere "Hey, we're running a debugger along with custom exception handling in a developer environment that itself is hooked to unity. We're getting approximately 1/4th the performance you should expect. " with the preview videos though.

[Missingno200]

20 minutes ago, Incarnation of Chaos said:

I DO wish they'd actually say somewhere "Hey, we're running a debugger along with custom exception handling in a developer environment that itself is hooked to unity. We're getting approximately 1/4th the performance you should expect. " with the preview videos though.

"Pre-alpha footage" and "test game footage" does that already though.

[The Aziz]

13 hours ago, Alexoff said:

I am worried about the performance of KSP 2, since recently we were shown a video with a new planet, in which there was a low FPS. The fact that the developers do not hesitate to show us videos with low FPS leads to disturbing thoughts.

 

13 minutes ago, Incarnation of Chaos said:

I DO wish they'd actually say somewhere "Hey, we're running a debugger along with custom exception handling in a developer environment that itself is hooked to unity. We're getting approximately 1/4th the performance you should expect. " with the preview videos though.

"Test scene" "asset preview" "work in progress" "not actual gameplay" "recording straight from Unity editor"

How much of that you people don't get? They don't need 240 fps to show off an engine when it's not even in the game. It's still in alpha state, so go ahead, launch an early KSP version, about 0.9, slap a bunch of boosters, fire them up at once and tell me how your experience was. Because I remember having about 2 seconds per frame at times.

Besides, Unity editor isn't exactly resource friendly, I know from experience that after placing a bunch of light sources, flying around with recording software turned on wasn't what I'd call smooth.

They probably have a test build that is more or less optimized to some point, it's nothing unusual, but it's internal, we don't have to see it.

[Incarnation of Chaos]

4 minutes ago, The Aziz said:

 

"Test scene" "asset preview" "work in progress" "not actual gameplay" "recording straight from Unity editor"

How much of that you people don't get? They don't need 240 fps to show off an engine when it's not even in the game. It's still in alpha state, so go ahead, launch an early KSP version, about 0.9, slap a bunch of boosters, fire them up at once and tell me how your experience was. Because I remember having about 2 seconds per frame at times.

Besides, Unity editor isn't exactly resource friendly, I know from experience that after placing a bunch of light sources, flying around with recording software turned on wasn't what I'd call smooth.

They probably have a test build that is more or less optimized to some point, it's nothing unusual, but it's internal, we don't have to see it.

I mean i get it, but that's because i'm literally training to be a software engineer. Most other people see those and don't realize to just what degree these tools tank performance sometimes.

[dprostock]

It is clear that the promises to sail in a Trasatlantico have been futile, at most we will have to be content with a raft of tied barrels ...

[MKI]

On 7/18/2021 at 12:08 AM, Incarnation of Chaos said:

I DO wish they'd actually say somewhere "Hey, we're running a debugger along with custom exception handling in a developer environment that itself is hooked to unity. We're getting approximately 1/4th the performance you should expect. " with the preview videos though.

 

On 7/18/2021 at 12:44 AM, Incarnation of Chaos said:

I mean i get it, but that's because i'm literally training to be a software engineer. Most other people see those and don't realize to just what degree these tools tank performance sometimes.

So the "literally trained software engineer" wants the video to specifically state all the technical reason why it look so unoptimized, instead of just "pre-alpha footage". Seems kinda backwards. All my non-programmer friends thinks a debugger is a fly trap XD

 

Regardless of what "fine-print" you slap on the video, I'd hope the average consumer realizes the video they are seeing of in-game footage a year from release isn't an indicator of the final product. This goes for literally anything you could buy, especially something as easily changeable as software. If not, and the average consumer assumes what they are seeing is the final product, just release more specific gameplay closer to the release when things are smoother and more optimized. Meanwhile keep the jank-builds coming now and build up the hype, and keep whatever fine print you want. 

 

I just played COH3 pre-alpha public release, had it crashed every single time, and still enjoyed the experience and understand it will be more stable in the future.  And that game also has basically no release date, and the game was announced the day the pre-alpha dropped. I'm not saying KSP needs to do this, but the fact KSP is throwing out hints about development shouldn't hinder its release success, because I'm pretty sure everyone understands its still being worked on. 

 

On the topic of PC resources. I think its rather fair the new KSP 2 will naturally ask for higher amounts of resources than the previous version. Any decently optimized software should be fine working on slightly older hardware. But lets be serious, KSP was never an "easy game to run" I don't know if its heavy physics calculations can be optimized to run on anything and everything. 

I'll wait to see how the game works later at scale to make any determinations in that regard. In the mean time if your worried about playing this game... upgrade your stuff! ;D

 

Edited July 19, 2021 by MKI

[Incarnation of Chaos]

4 hours ago, MKI said:

 

So the "literally trained software engineer" wants the video to specifically state all the technical reason why it look so unoptimized, instead of just "pre-alpha footage". Seems kinda backwards. All my non-programmer friends things a debugger is a fly trap XD

 

Regardless of what "fine-print" you slap on the video, I'd hope the average consumer realizes the video they are seeing of in-game footage a year from release isn't an indicator of the final product. This goes for literally anything you could buy, especially something as easily changeable as software. If not, and the average consumer assumes what they are seeing is the final product, just release more specific gameplay closer to the release when things are smoother and more optimized. Meanwhile keep the jank-builds coming now and build up the hype, and keep whatever fine print you want. 

 

I just played COH3 pre-alpha public release, had it crashed every single time, and still enjoyed the experience and understand it will be more stable in the future.  And that game also has basically no release date, and the game was announced the day the pre-alpha dropped. I'm not saying KSP needs to do this, but the fact KSP is throwing out hints about development shouldn't hinder its release success, because I'm pretty sure everyone understands its still being worked on. 

 

On the topic of PC resources. I think its rather fair the new KSP 2 will naturally ask for higher amounts of resources than the previous version. Any decently optimized software should be fine working on slightly older hardware. But lets be serious, KSP was never an "easy game to run" I don't know if its heavy physics calculations can be optimized to run on anything and everything. 

I'll wait to see how the game works later at scale to make any determinations in that regard. In the mean time if your worried about playing this game... upgrade your stuff! ;D

 

Person in training, bit of a difference.

And i'd probably just have a link somewhere in the description for those who want a more comprehensive explanation, no reason to spell it all out.

But for the rest, i agree. If you're really that worried, think about upgrades.

[K^2]

3 hours ago, Incarnation of Chaos said:

And i'd probably just have a link somewhere in the description for those who want a more comprehensive explanation, no reason to spell it all out.

I'm not sure the overlap between people who would understand a write-up on that and don't already know all of that is large enough to worry about. Anyone who worked in games knows that Release runs like molasse and Debug doesn't run at all. And anyone who hasn't but has the tech background can at least appreciate the difference between /Od and /O2 and extrapolate from there.

[Kerbart]

Well since you guys have been mentioning me a couple of times now, here I am: the user who is clueless about game development in general, or the inner workings of game development. I still don't understand the benefits of 240 FPS when my monitor has a 30Hz refresh rate,that's how much of a noob I am when it comes to this. But:

• If I'm going to put down serious money for KSP2, I expect the graphics t o be spectacular. That's pretty much the point, innit?
• Better written software will surely offset some of the penalty of more detailed graphics, but I doubt it will compensate everything.
• So, I'm counting on needing a new PC.
• But I doubtthat, just like KSP 1, it's going to require an $800 graphics card.

Remember, we KSP players don't want the devs to know what kind of hardware we're running the game on. The price we pay for that? They'll have to make assumptions on what the average machine is/can handle. That's our choice, not theirs. Subsequently, we'll find out how it performs when it comes out. Until then I'm not going to fret over it.

[Incarnation of Chaos]

5 hours ago, K^2 said:

I'm not sure the overlap between people who would understand a write-up on that and don't already know all of that is large enough to worry about. Anyone who worked in games knows that Release runs like molasse and Debug doesn't run at all. And anyone who hasn't but has the tech background can at least appreciate the difference between /Od and /O2 and extrapolate from there.

Fair enough, but iv'e just felt some attempt to quell the concerns has been needed for a while. This is hardly the first thread on it, nor is it the first person who's thought the pre-alpha build footage lagging was indicative of performance. But everyone above me has brought up good points, so eh.

Not everyone is on KSP fourms for nerds to explain this stuff to them xD

2 hours ago, Kerbart said:

Well since you guys have been mentioning me a couple of times now, here I am: the user who is clueless about game development in general, or the inner workings of game development. I still don't understand the benefits of 240 FPS when my monitor has a 30Hz refresh rate,that's how much of a noob I am when it comes to this. But:

• If I'm going to put down serious money for KSP2, I expect the graphics t o be spectacular. That's pretty much the point, innit?
• Better written software will surely offset some of the penalty of more detailed graphics, but I doubt it will compensate everything.
• So, I'm counting on needing a new PC.
• But I doubtthat, just like KSP 1, it's going to require an $800 graphics card.

Remember, we KSP players don't want the devs to know what kind of hardware we're running the game on. The price we pay for that? They'll have to make assumptions on what the average machine is/can handle. That's our choice, not theirs. Subsequently, we'll find out how it performs when it comes out. Until then I'm not going to fret over it.

Input latency is reduced, since each frame is basically "Closer" to the events on screen than otherwise possible.  Also some games (KSP is one!) tie loading speed to FPS.

 

[runner78]

On 7/18/2021 at 9:37 AM, The Aziz said:

Besides, Unity editor isn't exactly resource friendly, I know from experience that after placing a bunch of light sources, flying around with recording software turned on wasn't what I'd call smooth.

Editor or runtime? Game performance in the editor is always much slower than runtime build due to the editor overhead. There is also a difference whether the editor is in release mode or debug mode. For example, it makes no sense for level designers to run the editor in debug mode, this is only useful when debugging scripts.

[Lisias]

5 hours ago, Incarnation of Chaos said:

Input latency is reduced, since each frame is basically "Closer" to the events on screen than otherwise possible.  Also some games (KSP is one!) tie loading speed to FPS.

but the physics engine works a 50Hz, so any input you give between will just sit there until the next physics tick. If you are really in need 50Hz level precision,  go Niquist and set your FPS to 100.

The loading can be speedup settting the fps temporarily to unlimited (it's what MM does).

[Jack Mcslay]

On 7/16/2021 at 3:31 PM, K^2 said:

Purely in context of KSP2, I don't see any engine on the market that would work better than Unity. Rolling their own, making good use of middleware, would be my second choice for it. And having maintained in-house engines, knowing how much work and effort goes into it, I'm not suggesting it lightly. Unreal comes in at solid third, but it'd probably force KSP2 into EGS exclusivity, potentially reduce ability to mod, and no guarantees that it'd end up performing any better. More uniformly across platforms, for sure, but I don't think Intercept would be able to avoid heavy use of bad Blueprints, which would tank any gains from a better engine architecture.

Option 2 has a bad track record in terms of modding ability. Games with in-house engines such as GTAV and Skyrim are often mostly stuck to plugins to add weapons, vehicles and characters since the engine is so heavily focused on the game at hand without much room to make really big changes. I guess building an engine on top of Ogre could help a little with graphical mods by enabling it's python bindings but in any case I don't see a realistic scenario they could implement a modding framework as good as sticking with Unity.

[runner78]

3 hours ago, Lisias said:

but the physics engine works a 50Hz, so any input you give between will just sit there until the next physics tick. If you are really in need 50Hz level precision

By the way, that's just Unity's old default. This can be changed in the project.

[Incarnation of Chaos]

19 hours ago, Lisias said:

but the physics engine works a 50Hz, so any input you give between will just sit there until the next physics tick. If you are really in need 50Hz level precision,  go Niquist and set your FPS to 100.

The loading can be speedup settting the fps temporarily to unlimited (it's what MM does).

Yeah, it was a general answer for a general question tbh.

But i don't cap my FPS anyway, it's not necessary.

[pandaman]

Am I one of the few that's not unduly concerned about PC Performance?

I don't have a high spec machine, bought refurbished about 4 years ago  and KSP1 runs ok, not amazing by a long way, but not bad compared to what many seem to be experiencing.

I may be being over optimistic, but due to its better code optimisation, I expect KSP2 to run 'about the same' overall, but with far better graphics.

I certainly don't feel as if I will 'need' to upgrade to play it.  I plan to wait and see and then upgrade if/when I can.

[The Aziz]

26 minutes ago, pandaman said:

Am I one of the few that's not unduly concerned about PC Performance?

I don't have a high spec machine, bought refurbished about 4 years ago  and KSP1 runs ok, not amazing by a long way, but not bad compared to what many seem to be experiencing.

I may be being over optimistic, but due to its better code optimisation, I expect KSP2 to run 'about the same' overall, but with far better graphics.

I certainly don't feel as if I will 'need' to upgrade to play it.  I plan to wait and see and then upgrade if/when I can.

I'm with you. And not just because I just did an upgrade that would allow me to run two instances of KSP at the same time; I'm fairly certain that my 5 year old rig that I'm about to sell, would still be easily capable of running both games without much hiccup.

[Lisias]

On 7/19/2021 at 8:28 PM, K^2 said:

I'm not sure the overlap between people who would understand a write-up on that and don't already know all of that is large enough to worry about. Anyone who worked in games knows that Release runs like molasse and Debug doesn't run at all. And anyone who hasn't but has the tech background can at least appreciate the difference between /Od and /O2 and extrapolate from there.

The only problem on this rationale is that about 90 to 95% of humankind don't have the slightest clue about any of that, but we still need to convince them to give us their money.

Users do what users do. We need to cope with it (or forget about all this <piiii> and go sell cold sandwiches on the beach - where customers will make similar questions about the mayonnaise or something).  

[K^2]

On 7/20/2021 at 5:45 AM, Jack Mcslay said:

Option 2 has a bad track record in terms of modding ability. Games with in-house engines such as GTAV and Skyrim are often mostly stuck to plugins to add weapons, vehicles and characters since the engine is so heavily focused on the game at hand without much room to make really big changes. I guess building an engine on top of Ogre could help a little with graphical mods by enabling it's python bindings but in any case I don't see a realistic scenario they could implement a modding framework as good as sticking with Unity.

I'm not sure you've spent enough time in modding communities of either of the games you are quoting. Back in GTAIV days I was helping out a team trying to port San Andreas to IV engine by writing an air sim, since IV had no planes. The team fell apart, so this was never finished, but I had a working prototype. I've put a video in a spoiler. Aerodynamics ran as a custom component tick on vehicles, read in relevant parameters, computed aerodynamic forces, and applied them to vehicles. It also controlled animation on relevant control surfaces. The mods worked as injected C++ code and could control just about any aspect of any spawned entities or spawn new entities. With GTA V, the modding tools have only become more complete. I can say with confidence that it's easier to remake entire KSP in GTA V than it is to make anything like GTA in KSP. And this is based entirely on community-made tools. R* has never intended for their games to be moddable.

Spoiler

The inputs are coming in as fully on or fully off. I never got around to implementing steering curves, so flying it is a bit janky, but it is fairly accurate to the physics of an aircraft given the inputs. Also, yes, these control surfaces are completely wrong, but that's the mesh I had to work with. It's something a 3D artist would be able to fix relatively easily by re-exporting the model. Pay particular attention to the landing to see how the custom aerodynamics work together with game physics to produce a very reasonable touchdown. Of course, the speeds aren't entirely realistic, which was intentional to fit more with the map's reduced scale.

 

Skyrim's only problem is a really bad, really outdated engine. There is absolutely no limit to what you can do with mods there other than just the engine sucking abysmally. There are complete overhauls of magic, combat, crafting, etc. If you look at what people have done with mods there, it's way more extensive than any other game out there. People have made entirely new games in Skyrim. Although, they do tend to stick to the theme, because if you're not sticking to it, then you might as well work with FO4, which also has plethora of complete overhauls.

And, of course, you have to remember that real mudding started with id Tech games. I learned a whole lot about game programming by making Quake II mods. Later, I've messed with HL2, which is its own engine, but heavily inspired by design and style of id Tech. These were games made on in-house engines that were good enough that these engines ended up getting licensed out. But they were first-party engines first and foremost and they are practically responsible for us having expectations of mods that go beyond simple graphics replacements.

The only thing Unity has going for it as far as mods is that there's already a huge community of people creating tools for modding any Unity game. Which means that even if Intercept isn't going to put in an effort into allowing mods, people will be able to mod KSP2. Which is great. But first party mod support is always better, especially early on. And it's actually easier to give people modding tools as a developer if you have full access to all aspects of the engine. Allowing for easy mod import in Unity requires adding side-loading capability, which is not something Unity is designed to handle gracefully. Unity expects all of your assets neatly packaged, and working around it actually means inventing back doors. You don't have to do that if you have your own engine. Games by Bethesda, id Software, and Valve that are designed to work with mods will specifically search directories for additional archives, scripts, and even dynamic libraries to load because they are designed to be moddable. Which means you can package new assets and game code in the way that engine actually expects it to be formatted making results far more consistent.

4 hours ago, Lisias said:

The only problem on this rationale is that about 90 to 95% of humankind don't have the slightest clue about any of that, but we still need to convince them to give us their money.

Sure, but no explanation of early work is going to do anything for majority of people who might buy KSP2. The percentage of people who are going to read any kind of explanation on a forum is tiny. Most sales will be made on trailers and marketing that will be released in the last couple of months before release, and these will be tightly polished just to show the game from the best angles. None of it has anything to do with how the game might or might not run either now or in the future. It sucks, but games just aren't sold on facts. They're sold on pretty trailers.