[mjn.mixael]

I know what is going on, or at least part of it, and you can thank Axem.

In 3.6.12 all our background nebula graphics were horribly compressed to the point of ugly. Axem pointed that out and in MediaVPs 2014, the background graphics have been reverted to less horribly compressed images. That means they have more color data, less banding, and generally are more full. Ala, bigger bloom.

I will look further into this later this day... but I think your answers are not sufficient, The_E. The nebula graphics do look exactly the same, but the 2014 version are burn out... it does seem like bloom effect. The ships also look a lot more vibrant, saturated, etc. (I know they are different models, but still).

I do think he is detecting something here.

Ships being more vibrant is insignificant. Just about every ship in the screens posted are Shivans, which by the way, got complete makeovers from retail models and near retail textures. Of course they will be more vibrant.

[mjn.mixael]

2. Apparently the subsystem "Crystal" of the new Shivan Comm Nodes is now invulnerable. I tried attacking it from every direction, but it still remains at 100 %. I tried the same mission soon after with the old mediavps, and the crystal was vulnerable again. This does not prevent me from completing the mission, but it takes 5x more time to destroy Comm Nodes without destroying their Crystal first.

Retail says no.

Played the retail mission 3 times to test this theory and I never could even damage the Crystal without first destroying some of the other subsystems first. Functionality is the same on the HTL model as far as I can tell.

[Lykurgos88]

Played the retail mission 3 times to test this theory and I never could even damage the Crystal without first destroying some of the other subsystems first. Functionality is the same on the HTL model as far as I can tell.

You are doing it wrong. It takes a bit of trying and error; you have to get close and aim the upper or lower part of the crystal. No other subsystems are needed to damage. "Into the lion's Den" is apparently scripted so that once the Crystal is destroyed, the whole Comm Node self-destructs. It saves a lot of time. I remember finding this weak point as a kid playing FS2 retail and being exited as hell. Don't you dare to say I remember it wrong

I captured a FRAPS video of me destroying the crystal (and crystal ONLY) in MediaVP_3612. It will probably take a lot of time to upload it though because of my ****ty upload speed.

It's a shame really that this trick doesn't work anymore in the MediaVP_2014_beta. It's not a game-breaking restriction by all means, but it's still a shame from a gameplay mechanics perspective.

[Lykurgos88]

In 3.6.12 all our background nebula graphics were horribly compressed to the point of ugly. Axem pointed that out and in MediaVPs 2014, the background graphics have been reverted to less horribly compressed images. That means they have more color data, less banding, and generally are more full. Ala, bigger bloom.

Right. I suggest that the default Bloom intensity value should be lowered then. 75 is way too much for new nebula graphics.

[mjn.mixael]

Yeah.. that sounds more like taking advantage of a specific POF bug that I'm not going to bother to try and recreate. I can tell you that based on the POF data of retail, it should be completely (or apparently very near) impossible to just damage 'Crystal'.

It's pretty clearly a difference in the very locations of vertices and not anything else.

[Goober5000]

You have the cause and effect reversed.  Volition intended the crystal to be destroyable, and if the retail POF is buggy, it's buggy in the sense that shooting directly at the crystal doesn't damage it, not that shooting at its edges does.  I myself have used the crystal trick many times.

Have a look at this Volition Bugfix of the Day (which was an unfortunate case of an issue being closed without properly investigating it)...

8.30.99
Item #: 3181
Original Date: 8/20/1999
From: Jim Boone to John Enricco

[Freespace 2] I have trouble causing damage to the crystal in the Shivan comm node. Damage does not seem to consistently be applied when I hit the crystal. It seems like I have more luck hitting the crystal if I aim for its edge rather than the center of the crystal.

John Enricco: What Jasen has told me is that sub-objects are damaged by shooting at any part of their geometry and that no areas are immune. Closing this item.

James Tsai: Closed then.

Crystal? What crystal? What Shivan commnode?

(from http://web.archive.org/web/20080317015244/http://www.freespace2.com/bfod.cfm)

[Lykurgos88]

Here is the demonstration video for non-believers:

As you can see, shooting at the middle of crystal does nothing. However, if you aim at the top, the Crystal gets quickly destroyed even though other subsystems are barely touched. It has always been like this, ever since the retail FS2 days. And for some reason, you also have to be relatively close to succeed. Long range sniping is out of the question.

Even though this is apparently a bug, it becomes an awesome unintended feature, that is worth preserving. It rewards the player for good aiming. Actually I think that keeping this feature would allow the mission "Into the Lion's den" to be more faithful to the Volition's original vision, since Crystal subsystem was MEANT to be the "critical hit point".

I don't know if you have closed the development of the new HTL Comm node, but if you haven't, it would be a nice little detail to enable the destruction of Crystal subsystem the same way as in the video.

[mjn.mixael]

Well, we can't fix it without completely changing how the subsystems are placed in the POF which is generally an FSU no no. So in this case, mission takes the fall as I already stated.

It has nothing to do with just 'enabling destruction of the crystal.'

Actually, I think Goober should try to fix it. With all the complaining he has done to me about the MediaVPs being a moving target and POFs not being reliable because they change, etc. It should be interesting to see the solution.

[mjn.mixael]

Despite my already being confident in knowing it's a nearly unfixable problem (at least to fix within the rules of FSU), I spent my entire afternoon on it because clearly you all know better...

Investigation shows that this definitely not fixable in that it's a flaw in the design of the ship itself. There are too may bounding boxes, all rotating around and the mixture of all those bounding boxes into FSO's (shall I note here, rewritten) collision code makes it nigh impossible to create an HTL model with the exact same hitbox exploits.

Code: [Select]

<MjnMixael> i do not understand why the comm node is such a piece of ****
<MjnMixael> it must have something to do with all the overlapping bounding boxes
<MjnMixael> or SOMETHING
<MjnMixael> don't we have a wiki page with all the pcs2 subsystem properties listed?
<MjnMixael> i can't seem to find it
<MjnMixael> aha
<MjnMixael> http://www.hard-light.net/wiki/index.php/Subobject_and_subsystem_properties#.24lod0_name
<MjnMixael> so helpfully without POF or PCS2.. well... anywhere on the page, making it so intuitive to find
<Oddgrim> well..
*** AdmiralNelson has joined #scp-fsu
*** The_E has quit IRC: Quit: Leaving.
<MjnMixael> screw it.. how about we hack a workaround?
<MjnMixael> yeah, that didn't work either
<Zacam-Work> K, what's exactly the issue?
<Oddgrim> :c
<MjnMixael> ok, yeah
<MjnMixael> the comm node is a mess of bounding boxes
<MjnMixael> and that is why you can't ever kill the subsystem you are targetting
<MjnMixael> it was like this in the original and it's still like that now, just differently
<Zacam-Work> Hmmm. One idea: Open it in PCS, "Purge BSP Cache" and save as a PMF, open the PMF and save that as a new POF and see if it's still borked?
<Zacam-Work> Convoluted, I know, and not likely to work, but worth a shot.
<MjnMixael> no, that's not what I mean
<MjnMixael> what I mean is that each subsystem has it's own spinning bounding box and there's a lot of them at unusually large sizes all mixing together
<MjnMixael> I know FSO uses that data to calculate damage
<MjnMixael> add in that Lykergos (and I remember this) says you need to be close to the Crystal in retail in order to hit it...
<MjnMixael> that tells me you need to be within a certain bounding box, while firing through to another or some mix therein of wierdness that finally allowed in the retail POF to actually hit the Crystal subsystem
<niffiwan|afk> gah - sounds damn complicated
<MjnMixael> which explains why I'm killing 'Blade 1' while shooting at the 'Upper Shell'
*** niffiwan|afk is now known as niffiwan|work
<MjnMixael> which also explains why it seems to be easier to hit the crystal after other subsystems are destroyed
<niffiwan|work> iirc there is some wierdness in applying damage to subsystems, I recall seeing a fix that would let multipart turret barrels properly take damage to the turret subsystem
<MjnMixael> what I would -need- to fix it properly is a way to tell FSO's collsion code.. 'hit this poly, this one RIGHT HERE, and ignore all bounding box decisions, and this will damage Subsystem X'
<niffiwan|work> I wonder which part of the code that's all in...
<MjnMixael> honestly, based on all my experimentation this afternoon.. I think the problem is much deeper... I expect you'd have to delve into all the collision code as well
<MjnMixael> the retail POF got lucky in having an exposed part of the crystal bounding box or somesuch
<niffiwan|work> urgh
<Zacam-Work> So, does it (the model) work on an older build or with the old_collision commandline (or whatever that was)?
<MjnMixael> i've no idea
<MjnMixael> let's find out
<MjnMixael> no
<MjnMixael> I think I could hack a different solution.. but it would be a hack for sure
<niffiwan|work> are there a bunch of overlapping subsystems in the commnode?
<Zacam-Work> Hit VA up on that idea of collision box model deal to simplify the hitbox detection?
<niffiwan|work> (I should probably just open the VP and see if I can drive PCS to figure it out...)
<MjnMixael> there are lots of overlapping subsystems
<MjnMixael> well
<MjnMixael> let's be clear
<MjnMixael> there's only one Special Point style subsystem that completely surrounds the Crystal subsystem
<MjnMixael> but then the bounding boxes for the Shell pieces and the blades also overlap each other and the crystal
<MjnMixael> ok, my hack works....
<Zacam-Work> \o/
<niffiwan|work> what was the hack?
<MjnMixael> a large Special Point type subsystem to replace the subobject type
<Zacam-Work> Oh, nice. Me likey.
<Oddgrim> brute force it in :p
<MjnMixael> that's what she said
<BotenAnna> yes, yes it is
<Oddgrim> who let _that_ thing in
<MjnMixael> it's a large enough sphere that I can get enough hits on it to actually destroy the subsystem
<Oddgrim> well it works so yay
In order to make something work like it used to, I had to break every single FSU rule we have about model replacement. I have no idea how that will affect other campaigns and missions, either.

So, thank you all for talking like I don't know what I'm doing. Here's your hacked fix.

[Luis Dias]

That was very nice of you MJn!

[Goober5000]

Despite my already being confident in knowing it's a nearly unfixable problem (at least to fix within the rules of FSU), I spent my entire afternoon on it because clearly you all know better...

Yes, that is clearly the sentiment we intended to convey...

Investigation shows that this definitely not fixable in that it's a flaw in the design of the ship itself. There are too may bounding boxes, all rotating around and the mixture of all those bounding boxes into FSO's (shall I note here, rewritten) collision code makes it nigh impossible to create an HTL model with the exact same hitbox exploits.

Now that is interesting.  We already knew, from the Volition bug report, that there was some problem that prevented the crystal from taking hits even back in retail.  I suspect that if Volition had had the time to investigate the problem further, they would have made the same discovery.  This is the sort of unintended emergent behavior that happens when multiple complex systems interact with each other.

Lykurgos88, have you tested the retail Shivan Comm Node model with the latest FSO collision code?

In order to make something work like it used to, I had to break every single FSU rule we have about model replacement. I have no idea how that will affect other campaigns and missions, either.

Well, I do know that other campaigns and missions have used the crystal trick, but on the other hand they still allow you to destroy the comm node normally.  I'm curious as to the extent of rule-breaking though.  Is it as simple as adding an extra subsystem to deal with the Special Point overlapping the subobject, or is something else involved?

Regardless, thanks for your diligence, MjnMixael.

[mjn.mixael]

and so ends another episode of "Mjn hates the forums, fixes bugs". I'm going to go read a book. (le wut? I mean watch TV.)

[Rodo]

hahahaha, gotta admire the patience and dedication folks have in this community

We love you mjn, don't ever leave us please!

[Mongoose]

I honestly can't remember ever getting a clean shot on the crystal even with the original retail model.  I'd always wind up taking out the spinning blades first, and by the time I started doing appreciable damage to the crystal, the whole node was pretty much dead anyway.  The hack efforts are awesome, but I don't think any sane past campaign should have been relying on the original behavior in the first place, at least not if it was play-tested.

[Luis Dias]

Now people complain that the hack was totally unnecessary! MJN, I really feel your pain!

[Hellzed]

I remember hitting the crystal (with intact blades) with some Trebuchet missiles in MediaVPs 3610, and the whole Comm node was going down... Not sure if that can help.

[Trivial Psychic]

I recall once being able to Treb the crystal on all 3 Comm nodes in one run-through.  I believe that I WAS playing on Retail at the time.

[mjn.mixael]

I'm going to reiterate that if issues aren't Mantis'd (see sig), then they probably won't be fixed unless I've directly commented on them here. I'm not browsing this thread for bugs.

So...

[blowfish]

I've been playing through the main campaign with the new mediavps and found a few issues:

• On the bombing missions, it looks like the regular (long range) Cyclops is available, not just the Cyclops#Short as in retail.  Was this intentional?  It seems like quite an odd balance change.
• In "As Lightning Fall" I don't seem to be getting the sensor scrambling I usually do.  Did anyone else notice this?

Other things that are probably so minor they aren't even worth worrying about:

• The Sekhmet's lower LODs seem to be mapped improperly
• The Bakha has smoothing issues near the front of the pods (already reported to Mantis)
• LOD 3 on the Ma'at is basically a box with mostly dark textures.  Is this even worth having?

[Lykurgos88]

Lykurgos88, have you tested the retail Shivan Comm Node model with the latest FSO collision code?

I used the build 10222 in the Youtube video. Are there any collision fixes in newer builds?

To mjn.mixael:

Holy hell, I had no idea that this would be so much trouble. Sorry about my insistence, but thanks anyway for your effort I would really like contribute to FSO myself, but so far I haven't learned enough of the mechanics, so reporting potential bugs and problems is the least I can do.

Oh, and by the way, I believe I no longer have the disappearing Sathanas problem. I think it was a Beta 1 issue. I completed the entire in FS2 campaign in Beta2 and had no problems with the new (and gorgeous) Sathanas model

-----

With regards to the overly burnt bloom in MediaVP_2014, here is an in-nebula comparison. So it's not only the background texture that is affected, but nebula fog sprites too. In addition I'm not really sure what to think about the new green tint that pops every now and then in the 2014-version. I would prefer that the blue nebula consists only of different shades of blue like in MediaVP_3612, but I don't know what the majority thinks. However I think I can live the new nebula graphics when I greatly lower my bloom intensity value (these two picture use the default bloom intensity of 75).

MediaVP2014_Beta in-nebula
MediaVP3612 in-nebula