265 replies to this topic

[Jyi]

Koniving, on 05 December 2019 - 12:48 PM, said:

Overall, in JYI's recent video with the Atlas, you can see this at work.
The AI can detect being hit, but closer target instantly takes priority so after "AssignAttacker_CombatTarget_C'
It is always running "RunEQSQuery InTargetRangeCone Assign_CombatTarget_C"
due to higher level tree behavior overriding this sub-tree and damaged tree reference.

Well, I think it's the closest target that is shooting at the AI that takes priority. Otherwise, the enemy would switch to player when player walks close, but as you can see from my video, the Awesome doesn't care a crap about me as long as the Atlas is shooting at it.

I think there's also some priority targeting on player, but it only seems to activate once no one else is shooting at the AI. As can also be seen from the encounter with the Awesome. When the lancemate stops shooting at the Awesome, it starts concentrating on me - even when the Atlas is momentarily closer to the enemy than me.

It's a sh*tshow.

Edited by Jyi, 05 December 2019 - 08:15 PM.

[Jackal Noble]

The AI focuses on what is presenting the most damage at the moment.

[Koniving]

Jyi, on 05 December 2019 - 08:14 PM, said:

Well, I think it's the closest target that is shooting at the AI that takes priority. Otherwise, the enemy would switch to player when player walks close, but as you can see from my video, the Awesome doesn't care a crap about me as long as the Atlas is shooting at it.

I think there's also some priority targeting on player, but it only seems to activate once no one else is shooting at the AI. As can also be seen from the encounter with the Awesome. When the lancemate stops shooting at the Awesome, it starts concentrating on me - even when the Atlas is momentarily closer to the enemy than me.

It's a sh*tshow.

Was the helicopter shooting the atlas? Was the thunderbolt? Because it only seemed like the vehicle with the ac was shooting. That's why when I brought up the test it was using AI on hold fire orders.

So the issue really boils down to not evaluating multiple threats or having a focus strongest attacker tidbit.

[Jyi]

By the way, I just stumbled on SaltyPuglord's video of the "hidden files" of MW5 demo. Maybe some of you had already found these, but it was new info for me.

To simply explain this: there are video files called "Codex_something" that show a LOT of the metagame features of the full game. Things like repair costs and weapon prices.

The argument has been made that the difficulty of AI in MW5 demo is "fine" because in the full game there will be repair costs and other upkeep costs.

In the Codexes, there is an example of the repairs of a Centurion. This Centurion has lost an arm, and is down to structure on everything else but head. It has also lost 2 medium lasers and 1 single heat sink and has 1 damaged single heat sink.

Now, the repair costs of structure, armor AND the single heat sink come down to 20,360 cbills and take 4 days. That is excluding the replacement of the 2 medium lasers and the 1 heatsink. So, what are the prices of medium lasers in the game? Going by the info from the Codexes, a standard LRM-20 costs 250,000, which is the same price as on Sarna.net. From this information, we can deduce that the cost of a single, standard medium laser is 40,000. The price of a single heat sink is unknown to me, but I'd assume it's less than 10,000.

With some quick maths, the total repair of this Centurion is 110,360 cbills at best. Now, it doesn't take into calculation a ton of ammunition, but in the example, the Centurion hasn't had any ammo explosions and thus hasn't lost the ammunition. I don't know whether MW5 has the same system as HBStech where you don't need to replace used ammo, and only need to replace ammunition that has been destroyed. But if we were to include the replacement of the 2 tons of AC10 - which is 6000 cbills per ton according to Sarna - the total repair price of the Centurion would be 122,360 cbills. Based on the Codexes, the replacement of a single component takes 1 day per. So replacing 2 ML and 1 SHS would take an extra 3 days.

The Codexes also show that the repair costs and times are increased by a percentage in conflict zones and places where you have bad reputation. It is unknown to me what the maximum limit of this increase can be, but in the Codexes there is an example of up to 75%. This boosts the 122,360 cbill repair bill to 214,130 cbills. And with 75% increased duration of repairs, the 4 day repair + 3 days for replacement of ML & HS would be inflated to 12~ days.

Pretty steep, huh? Well, the Codexes ALSO reveal the base rewards of contracts, and how much you can negotiate those up. For example, a mission of "difficulty 5" (which I assume is a cakewalk, as there are "difficulty 11" and "difficulty 15" examples in the Codexes) and a tonnage limit of 30t (solo mission) rewards 250,000 cbills, gives a "salvage shares" of 2 (which I assume is the amount of weapons / components / mech parts you can take) and has a damage coverage, which is 0 cbills. You also have 1 negotiation point, which can be used to boost any of the rewards. And these rewards are on neutral reputation, by the way. The Codex shows that by using that one negotiation point, you can boost the cbill reward from 250,000 to 365,00.

This EASILY covers the repairs and rearming of the example Centurion even in a conflict zone with heavily inflated costs.

Furthermore, there are other examples of "standard contracts" in the Codexes. Going by the names and types of these contracts, I'm 100% certain they are going to be identical, randomly generated versions of the "instant action" -missions we've seen in the demo. In the Codexes, a mission of difficulty 8 with a tonnage limit of 145t rewards 265,000 cbills, 2 salvage shares and 0 cbill damage coverage on neutral reputation. And you get 2 negotiation points to boost that up. The Codex shows that by using 1 neg. point, you can boost salvage shares from 2 to 4, and based on information I provided above, a single neg. point will boost the cbill reward of 250k -> 365k, so I'd assume it boosts 265k even more.

So, if you pick the right salvage, like high cost weapons, you can easily make 1,000,000 cbills worth in a mission that is obviously easier than any of the demo missions we got (145t limit). And if you have a better than neutral reputation, I assume that boosts the amount of negotiation points even further, which means even more money and salvage.

Considering that, as I have demonstrated, you can easily do higher difficulty missions with a light or a medium mech without dying or even losing any weapons or all of your armor, it's not far fetched at all to assume you could SOLO these example missions without ANY PROBLEMS. This means even smaller repair costs than the example Centurion in the Codex videos. Repair costs in the ball park of 10k-20k, which would be inflated to 35k MAXIMUM in a conflict zone.

It would barely be 10% of the cbill reward. Not to mention all the salvage you'd get.

This would make the game a cakewalk from start to finish.

"But! There are other costs, like mech maintenance and traveling", I hear you say. Well, I'm ahead of you on these as well. The Codexes have examples of them, you see. Mech maintenance for a 50t Centurion per MONTH is 30,000 cbills, and both Javelin and Jenner are 20,000 cbills (so I'd assume it's not even based on chassis or tonnage, but only weight class). Traveling 11 jumps is 550,000 cbills and takes 83 days.

So, lets say you do 4 missions solo and rotate 4 mechs, 2 of which are mediums and 2 of which are lights. Lets assume the monetary BASE REWARD of these missions is 250k and you won't use any negotiation points and we exclude salvage from the calculations (for simplicity's sake). The total amount of rewards is exactly 1 mil. On each mission you get somewhat damaged and the average repair costs come down to 50k per mech, totaling 200k. Lets assume you're in a conflict zone where the repair costs are increased by 50%, which would make the totals 300k. You'd still have 700k left for other things.

Lets then assume you travel 11 jumps. The maintenance costs for these 4 mechs per month are 100k - which would mean 200k-300k for 83 days (depending on which part of the month you started your travel), so the maximum would be 850k. That's only 150k in minus. But we all know already the total rewards would be MUCH MUCH higher than 1 mil. and you wouldn't be traveling 11 jumps if you didn't have the money for it.

Oh, and yes, the lancemates have salary, but if you just kick them out of your merc corp or only hire the worst ones and never take them on missions, those costs are negligible (for example, a rookie pilot has a salary of 1,000 cbills a month).

Bottomline is: if you solo missions in MW5 and the AI is at the level it is in the demo, it will be borderline impossible for you to lose money. Period.

Edited by Jyi, 05 December 2019 - 10:13 PM.

[Koniving]

Jackal Noble, on 05 December 2019 - 08:37 PM, said:

The AI focuses on what is presenting the most damage at the moment.

The AI doesn't have a quantifier for damage delivered.
Just AI_damagesense which is basically "I am taking damage".
When taking damage (any value) the corresponding tree kicks in, identifying who delivered the attack and assigning them as the target, then it jumps to the Hastarget call which instructs the decorator_mech to rotate to face target with torso which kicks the legs in gear once torso reaches it's limit.

It does not have an entry to judge damage delivered by the new threat compared to the old. The AI does not hold more than one target in memory at a time.

In essence, imagine playing mwo. Now every time you are shot you will change to the new target...and completely forget about the previous target. In essence only the enemy you target is rendered and everything else is invisible. Nothing else exists but rectangles, cones, and the guy you targeted.

[Jyi]

Koniving, on 05 December 2019 - 09:55 PM, said:

Was the helicopter shooting the atlas? Was the thunderbolt? Because it only seemed like the vehicle with the ac was shooting. That's why when I brought up the test it was using AI on hold fire orders. So the issue really boils down to not evaluating multiple threats or having a focus strongest attacker tidbit.

I think the helicopter (and multiple tanks at multiple points) were shooting at the Atlas. The Thunderbolt definitely was, but as the lasers are difficult to see (because of their silly looking, wannabe realistic design), it can be hard to spot in the video.

And yea, I think one of the biggest problems with the AI is that it seems to have no threat assesment of any kind. It doesn't differentiate between more damaging weapons or targets with more weapons. It just tries to shoot at the latest attacker.

[Koniving]

So it only targets the last thing to shoot it. If the other targets in JYI's video were shooting (didn't see any effects but using a 20 dollar smart phone), then the shuffle of derp was from cycling between shots from several directions. There's a delay in target changes accounted for by a wait time (war stories mw5 video has a mech rapidly twitching the arm it is rapidly swapping targets without a delay).

[Koniving]

Welp. Got my 360 controller out. Time to have some fun.

Btw will someone jump back a few pages to the one with infantry swarming a thunderbolt, open the spoiler and pick which art I should use for the omicron 3000 (Stinger's) laser?

[Jyi]

By the way, other fun stuff:

If you don't like the PGI logo or the other intro screens, you can rename the files "piranha-logo-sound.mp4" and "legal-screen.mp4" in "...\MechWarrior5Beta\MW5Mercs\Content\Movies" -folder. For some people this folder seems to be on "C:\Program Files", but for me it was in "C:\Games". This location is also where all the Codexes are.

[FRAGTAST1C]

"If you solo missions...."

Why?

Plus, there are higher tier weapons that you might need to buy, higher tier pilots to hire and pay their salary, there could be other money sinks like needing a lot of c-bills to actually assemble a mech from the parts you've salvaged.

Again, I'm not making excuses. Any amount of difficulty can be modded. The AI can be tweaked. But all I get from forum posts is outrage.

[Jyi]

FRAGTAST1C, on 05 December 2019 - 10:20 PM, said:

"If you solo missions...."

Why?

You'd solo missions, because the AI - including your lancemates - is a potato. No, scratch that. A potato in MWO is someone who can not do anything. In MW5, the AI is worse than that, they are detrimental to their friends. They are worse than potatoes.

This is why you'd solo missions. Because the enemy AI is so dumb that it's possible, and because lancemates are useless and would shoot you in the back.

Quote

Plus, there are higher tier weapons that you might need to buy, higher tier pilots to hire and pay their salary, there could be other money sinks like needing a lot of c-bills to actually assemble a mech from the parts you've salvaged.

Again, I'm not making excuses. Any amount of difficulty can be modded. The AI can be tweaked. But all I get from forum posts is outrage.

Higher tier weapons cost more, yes, but as you are already making a huge bank by doing the missions solo, it's irrelevant. You wouldn't hire higher tier pilots, because as I explained, you wouldn't need or want them. And assembling the mechs can't be that expensive, because you can also buy complete mechs in the game, and their prices seem to be around similar as HBStech. So from several hundreds of thousands to tens of millions at best. Actually, I'm not even certain there are going to be mech parts to salvage instead of complete mechs. MW5 might just have complete mechs you repair and rearm.

We don't exactly have any kind of proof how much the difficulty can be modded. Koniving is planning on improving the AI, but we don't know how much he'll actually be able to do once the game releases. It might be more restricted than we realized or he might not be able to do as much as he hopes (even though he seems to have somewhat realistic plans and seems capable to me).

Also, it's ridicilous that you're seriously suggesting that taking care of the BASE BALANCE of the game should be up to modders.

And to be honest, it was a stupid argument to begin with that missions should be easy as f because the META GAME is "going to be hard" (which I have now proven it's not going to be). As in, the part of the game that is just supposed to be some extra flavor is the part of the game you want to be a challenge.

Edited by Jyi, 05 December 2019 - 10:34 PM.

[Unleashed3k]

Koniving, on 05 December 2019 - 10:17 PM, said:

Welp. Got my 360 controller out. Time to have some fun.

Btw will someone jump back a few pages to the one with infantry swarming a thunderbolt, open the spoiler and pick which art I should use for the omicron 3000 (Stinger's) laser?

I think the om 3k on the 2nd (the running) mech looks much better, but that's just my opinion.

[Koniving]

Unleashed3k, on 05 December 2019 - 10:33 PM, said:

I think the om 3k on the 2nd (the running) mech looks much better, but that's just my opinion.

Here ya go.



I don't have to worry about how it fits for "many locations" as my mod introduces a lot of weapons but also has limitations in where they can be mounted (while opening "energy", etc. slots to be soft points where you can mount energy, ballistic, or missile where if you try to put a ballistic in an energy slot you just have a longer delay until it's done due to the extra work). So this laser can only be mounted on arms (and maybe the Locust's CT 'turret'). I was afraid the newer image looked too plain and generic, so I included a bit more of the mount.

Would've had it up sooner but my daughter's up and insisted that I be a jungle gym and then play peekaboo with a box (okay Snake), and had to gradually wind her down a bit so I could sit down and put this up. And now she's feeding me skittles and saying the names of the colors.

(She gets super excited when the mechs come on...and keeps getting into my mech models.)

Edited by Koniving, 06 December 2019 - 12:51 AM.

[Koniving]

Needs texture work, but as I said in reply to Reno in my mod page, I'm looking for a texture artist. (I can do 3D modeling, animation, basic scripting, AI trees apparently, but if I were to texture MWO it'd probably look like everything got colored with colored pencils as I'd probably print out the maps to color by hand....and frankly that's too time-consuming to change the whole look, and I'm not suited for proper texture work.)

[Trystan Thorne]

Not easy to track all the AI discussions here! We need a good AI to sort all the threads.

I get where you are coming from Jyi, but you are way too negative mate.
The AI is not as bad, as you make it out to be. Take off your Tier 1 goggles.

Nice work Koniving on trying to find the faults with the AI.

I will remain with what I just posted in the other thread. The AI is fine, could be better, but fine.
And the campaign management diffculty needs to be considered as well. I haven't seen the video, where the Centurion got repaired in 4 days Jyi, but in the first MW5 overview video, it needed 22 days to fix up the Atlas.

Here's the thing. MW5 must have had a troubled development (quite a few delays and needed EGS money) and the fact that we didn't see any material for month was worrying.
The result is much better than I expected and is one of the best things PGI ever did (beside winning the court case against HG, the great Mech designs and getting the core gameplay of MWO right).
I will wait with my opinion when the games releases fully.

[Unleashed3k]

Koniving, on 05 December 2019 - 11:32 PM, said:

Here ya go.



Would've had it up sooner but my daughter's up and insisted that I be a jungle gym and then play peekaboo with a box (okay Snake), and had to gradually wind her down a bit so I could sit down and put this up. And now she's feeding me skittles and saying the names of the colors.

(She gets super excited when the mechs come on...and keeps getting into my mech models.)

Yes, that's the one I meant, really looks good already, I think if PGI ever needs someone with skill and passion, they should call you for future projects.

You're doing a great job and I really like your professional input!
Keep up the great work m8!

@Jyi: Can we maybe stop spitting around that the missions in the campaign will be "easy as f***" until we actually SEE how they play out?
It's too late for major changes anyway, but none of us have played the campaign, stuff found in the beta-install could only be placeholders, we don't know anything about the releasebuild.

I want a challenge myself, but hell, we haven't seen the release and what it will bring.

There were so many casual players that had Major problems with vanilla BT, so many threads about unfair missions, enemies and stuff.
This game is coded to bring as many new players as possible into the BTU, while trying to keep BTU fans happy (i.e. stay within lore, no engine changes, no min-max meta like in MW:O, etc)

I have many friends that played MW:O for a few ours and never came back, even if they still love BT, play TT on a regular base and that were only waiting for a new "real" Mechwarrior game.

Guess what, they all don't care about AI too much, they all want to play MW with friends vs enviroment, they are all hyped for the game, because it is, what they have been waitng for so long. closer to lore, away from pvp-balance, "meta" and other stuff.

You have a somehow valid point, but only to a certain degree, the rest of your writing is simply reapeating how much the AI sucks for your personal taste and some others, but that's about 5-10%? of the upcoming playerbase from MW5.

You made your point clear more then one time, but it's too late anyway and reading the same stuff over and over just gets boring and doesn't change the release in ANY way….

Act adult, take a rest, let release come and then, if campaign is STILL SO BAD next week, go on with your "witchhunt"...

Edited by Unleashed3k, 06 December 2019 - 12:21 AM.

[FRAGTAST1C]

Jyi, on 05 December 2019 - 10:29 PM, said:

You'd solo missions, because the AI - including your lancemates - is a potato. No, scratch that. A potato in MWO is someone who can not do anything. In MW5, the AI is worse than that, they are detrimental to their friends. They are worse than potatoes.

This is why you'd solo missions. Because the enemy AI is so dumb that it's possible, and because lancemates are useless and would shoot you in the back.

Hyperbole. Next...

Jyi, on 05 December 2019 - 10:29 PM, said:

We don't exactly have any kind of proof how much the difficulty can be modded. Koniving is planning on improving the AI, but we don't know how much he'll actually be able to do once the game releases. It might be more restricted than we realized or he might not be able to do as much as he hopes (even though he seems to have somewhat realistic plans and seems capable to me).

Also, it's ridicilous that you're seriously suggesting that taking care of the BASE BALANCE of the game should be up to modders.

And to be honest, it was a stupid argument to begin with that missions should be easy as f because the META GAME is "going to be hard" (which I have now proven it's not going to be). As in, the part of the game that is just supposed to be some extra flavor is the part of the game you want to be a challenge.

So, one guy who is active here is the only guy who can do it? Did you see anyone from MW : LL team or Roguetech team or any vastly experienced modders talk about it?

No one said that the game should be easy. What we have now is a demo. If you want to keep harping on about AI being easy, guess what the counter-argument will be?

So far, you've seen videos, read some forum posts and have created hyperboles. No wonder video game developers don't care about actual customers. People like you have done such a great job in making noise at an unreleased product that nothing else will matter, not the people in the middle, not the fans and certainly not the paying customers 'cause you have made sure that everyone gets painted by a broad brush.

Just shut the hell up and let the developers release a game and we'll change what we don't agree with. Let them create something that their vision perceives and release their product first. If it's easy, we'll change it. If it's too hard, we'll change. It's people like you that climb on any bandwagon to get some fame without actually doing anything. But hey, you're free to use the mods others make.

It's perfectly understandable if MW5 is super easy. Lots of the developers aren't good gamers and if they see that it's too hard for the creators themselves, then they'll be reluctant to make it hard for the paying customer. Just grow a brain first, man. You're acting like a hormonal teenager who doesn't know anything but makes a lot of noise by looking out of the window.

[Unleashed3k]

FRAGTAST1C, on 06 December 2019 - 12:29 AM, said:

So, one guy who is active here is the only guy who can do it? Did you see anyone from MW : LL team or Roguetech team or any vastly experienced modders talk about it?

No one said that the game should be easy. What we have now is a demo. If you want to keep harping on about AI being easy, guess what the counter-argument will be?

So far, you've seen videos, read some forum posts and have created hyperboles. No wonder video game developers don't care about actual customers. People like you have done such a great job in making noise at an unreleased product that nothing else will matter, not the people in the middle, not the fans and certainly not the paying customers 'cause you have made sure that everyone gets painted by a broad brush.

Just shut the hell up and let the developers release a game and we'll change what we don't agree with. Let them create something that their vision perceives and release their product first. If it's easy, we'll change it. If it's too hard, we'll change. It's people like you that climb on any bandwagon to get some fame without actually doing anything. But hey, you're free to use the mods others make.

It's perfectly understandable if MW5 is super easy. Lots of the developers aren't good gamers and if they see that it's too hard for the creators themselves, then they'll be reluctant to make it hard for the paying customer. Just grow a brain first, man. You're acting like a hormonal teenager who doesn't know anything but makes a lot of noise by looking out of the window.

AMEN!!!

[Jyi]

Trystan Thorne, on 06 December 2019 - 12:17 AM, said:

Not easy to track all the AI discussions here! We need a good AI to sort all the threads.

I get where you are coming from Jyi, but you are way too negative mate.
The AI is not as bad, as you make it out to be. Take off your Tier 1 goggles.

I would love to be less negative. But no, the AI is just as bad as I make it out to be.

I have gone and played the MW5 demo dozens and dozens of times, trying to see where other people are coming from, trying to prove to myself that it can't be that bad... and every time I come back with more proof that it is. Almost every time I find a new way to exploit the dumbness of the AI.

Quote

I will remain with what I just posted in the other thread. The AI is fine, could be better, but fine.
And the campaign management diffculty needs to be considered as well. I haven't seen the video, where the Centurion got repaired in 4 days Jyi, but in the first MW5 overview video, it needed 22 days to fix up the Atlas.

No, it's not fine. I'm sorry to say it, but it's not. If it hasn't been improved in the release, you will sooner or later see what I'm talking about. And if it has been improved... well, then we might have a good game.

Like I explained in detail in the post about repair costs and times: the duration is dependant on where you do the repairs. A conflict zone will have longer repair times. In my example, the repair time of the Centurion goes up to 12 days. So, the base repair time of that 22 day Atlas is probably lower.

And finally, it's irrelevant what the repair time for an Atlas even is, because you will never need an Atlas to do well in the game. You can complete almost all missions (in the demo) solo in a light or medium mech. And the warzone -mission in the demo has the highest 400 ton weight limit, so it's supposedly on the same level as the hardest missions in the final product will be.

Quote

Here's the thing. MW5 must have had a troubled development (quite a few delays and needed EGS money) and the fact that we didn't see any material for month was worrying.
The result is much better than I expected and is one of the best things PGI ever did (beside winning the court case against HG, the great Mech designs and getting the core gameplay of MWO right).
I will wait with my opinion when the games releases fully.

That development of MW5 has been troubled is very believeable. It's pretty obvious now, though, that it was partially due to mismanagement and stupid goals. They had time to develop irrelevant features like gravity for projectiles, but didn't bother developing the AI beyond the basic settings of what UE comes with. That's like putting leather seats in a car that has no engine and calling it complete.

And I don't even think they got the core gameplay of MWO right. Don't get me wrong, it's pretty fun, but it could've been much more fun if they hadn't f*cked it up. It was much more fun in the beta. Then they insisted on putting in collision, which made the game horrible. Then they removed the collision, and never talked about it again. Then they put in weapons and jump jets that made poptarting the best gameplay tactic by far. And when they realized that, they nerfed jump jets to oblivion, where they still remain to this day. And even that wasn't enough, so they nerfed Gauss and PPC's, so that poptarting has never been a worthy strategy after that in 99% of situations.

This kind of back-and-forth -development and completely clueless decisions against arguments from top players is what eventually killed MWO's chances. But to this day, do they see that? No, they're happier blaming Steam and selling mechs.

Funnily enough, I remember when I was in DV8 (originally R-sauce), and the developers asked our clan leader if we could start doing testing and give our input on things. As in, they wanted us to work for free to make their game better. I don't know exactly what came of that, as that's about the time I decided to take a hiatus from MWO.

Edited by Jyi, 06 December 2019 - 12:38 AM.

[Unleashed3k]

Jyi, on 06 December 2019 - 12:37 AM, said:

wall of text again only ragefilled

You bought ONE!!! copy, so this is how important your opinion is.
Deal with it.

You are only ssalty and try to spread that upon others, even more when you see that others actually like the game how it is or have minor problems with it.

As I said before, grow up!

Edited by Unleashed3k, 06 December 2019 - 12:48 AM.