PDA

View Full Version : Update: Prop Collision modelling



RAF74_Buzzsaw
Aug-29-2017, 02:12
Salute All

There was a complaint not too long ago about the way some aircraft collision models were implemented... the suggestion was these planes, when hitting first with their propellors, were invulnerable in collisions. Thanks to the persons who brought it to our attention. :thumbsup:

Below is a example of the problem... it only affected the Spitfire:


https://www.youtube.com/watch?v=gcvkw6zu2RU

Because we obviously take bug issues seriously we investigated the problem.

What in fact is happening is not the prop causing the damage, its the nose of the aircraft... in the case of the Spitfire, for some reason the original developers did not make the nose detachable... therefore it is invulnerable.

The propellors are not causing the damage. In fact the first aspect which came to light was the fact there is no collision model for propellors versus other objects... only versus the ground.

Here is a sample systems collision model for the Spitfire.

https://ibb.co/h27cK5

You can see there are no collision boxes for the prop.

What is in place for CLIFFS OF DOVER is a collision model for the prop contacting the ground, shown below. Props have 'clips' on the end of the prop models, which when contacting the ground, signal to the game engine to change the visual model and also cause engine damage. But only if the prop contacts the ground. There is no effect for contacting other objects.

https://ibb.co/ceoTRk

Someone looking quickly at the collision modelling might think the 'clips' also interact with objects, but not the case.

What is actually needed is to build a complete set of collision hit boxes for the prop and write a new system into the code. The graphics models are relatively simple and have been built.

https://ibb.co/iuZGe5

But the real issue is the code work and that is not possible for TF 4.5... because we do not want to delay the release any more than necessary.

We have fixed the bug with the non-detachable Spit nose section... and that provides a temporary fix which will eliminate the issue.


https://www.youtube.com/watch?v=HQScvJiprCo

But the real fix which will allow props to cause damage to objects and to be damaged by them, i.e. the 'Taran' effect, will have to wait.

By the way, we have to mention another fact.

There has been a considerable amount of hysteria about this collision bug... but in fact it was not a big issue. While it is true a player in SP may be able to slow down his computer to 1/8th speed and creep up on another friendly flying aircraft, in real practice at normal flying speed, there is almost zero chance of a collision occurring which would leave a Spitfire without damage. And yes, you saw I could do it on the ground... (and notice the Spit took wing damage) but CLIFFS OF DOVER is not a lawnmower simulation. There is almost no chance to creep up on an enemy aircraft... if that aircraft is a bomber, its rear gunner will shoot you to pieces long before your prop gets anywhere near it. And a single engined type will maneuver out of the way 99% of the time.

Most of the collisions which occur online on the servers and which see one aircraft damaged but the other unaffected are due to connectivity and lag.

Thanks for everyone's patience. ;)

PreyStalker
Aug-29-2017, 04:44
Great work Buzzsaw !

Thanks for sharing the hitbox image too. Reminded me of the months I spent once reworking and resizing all the cargo ship hitboxes in Silent Hunter 3.

What a can of worms that was !

Appreciate all TFS efforts to iron out these pesky bugs.

Prey.

:salute:

ATAG_Ribbs
Aug-29-2017, 07:54
Thanks for the videos and explanation Buzz! :thumbsup:

Erpr.Gr.210_Mölders
Aug-29-2017, 08:25
Thanks for the report! :thumbsup:

A curiosity...does this applies also to incoming bullets? I mean when the bullets hit the nose\prop do they provoke any damage or not?

9./JG52 Sturm
Aug-29-2017, 08:27
Thanks for the update, I don't think it is anywhere near as big a problem as some have made out: Correcting the invulnerable Spit nose hopefully solves most of the aggravation where after a collision one aircraft could explode and the other be merrily on its way with very little damage.


Thanks for the report! :thumbsup:

A curiosity...does this applies also to incoming bullets? I mean when the bullets hit the nose\prop do they provoke any damage or not?

Nose I'd say definitely because you can wreck the engine entirely with hits to the nose or engine cowling. Prop I'd suggest not, never experienced prop damage unless contacting the ground which tallies with Buzz's explanation of only having points at the ends of the prop modeled.

ZG26_Sir_Racha
Aug-29-2017, 12:24
Thank you for the information Buzz! much appreciated :)

Ekko
Aug-29-2017, 13:38
Thanks for the update TFS:thumbsup:

RAF74_Buzzsaw
Aug-29-2017, 14:18
Thanks for the report! :thumbsup:

A curiosity...does this applies also to incoming bullets? I mean when the bullets hit the nose\prop do they provoke any damage or not?

None.

There are no Prop hit boxes for bullets to contact... there are hitboxes for the Prop governor, engine, oil tanks, etc. etc. You can see that with the Spit model shown.

MightyBouff
Aug-29-2017, 14:40
None.

There are no Prop hit boxes for bullets to contact... there are hitboxes for the Prop governor, engine, oil tanks, etc. etc. You can see that with the Spit model shown.

One for the future then, roll on
5.5!

=AN=Lobo
Sep-01-2017, 21:24
One for the future then, roll on
5.5!

This will be too much for netcode. Individual spinning blades with collision meshes and bullets interaction will need something I believe we don't have tecnology for MP games to be accurate and with good performance. In this scenario ping will do a mess.

=AN=Lobo
Sep-02-2017, 01:16
Thanks for the update, I don't think it is anywhere near as big a problem as some have made out: Correcting the invulnerable Spit nose hopefully solves most of the aggravation where after a collision one aircraft could explode and the other be merrily on its way with very little damage.

Someday people will remember me as the voice that said the obvious, a long long time ago...


https://www.youtube.com/watch?v=hu4BspYhLPo&feature=youtu.be

Single player, AI 109 and player Hurricane. Five HtH collisions recorded in a row. No netcode involved. Hurricane nose detach perfectly.

All rigid body dynamics in Cliffs of Dover (collisions included) demands A LOT of work. It's totally unreliable.

And it's a fact. Never was a netcode/lag issue. And I really don't believe, after all I showed, that a living soul still thinking this is a minor issue.

Maybe if people stop talking nonsense like "it's ping", and do some serious testings, this mess one day will be adressed. Cliffs of Dover have the worse rigid body dynamics in modern serious combat flight sims.

And is really sad that I have to waste so much time showing this to people that fly this sim and others that are working with all tools available and testers, and have all this denial in exchange.

Instead, people maybe need to stop the fanboism and start to test things and think why some issues happens, and TFS listen more seriously who points HUGE issues(hes, all rigidy body in CloD is a HUGE issue).

But keep the good work, all of you. Netcode, graphics, FM, DM (due bullets) are great in CloD, best in market. Just stop the denial about bugs and solve AI questions and lack of rain and REAL dynamic weather. This sim will be great.

=AN=Lobo
Sep-02-2017, 02:23
By the way, we have to mention another fact.

There has been a considerable amount of hysteria about this collision bug... but in fact it was not a big issue. While it is true a player in SP may be able to slow down his computer to 1/8th speed and creep up on another friendly flying aircraft, in real practice at normal flying speed, there is almost zero chance of a collision occurring which would leave a Spitfire without damage. And yes, you saw I could do it on the ground... (and notice the Spit took wing damage) but CLIFFS OF DOVER is not a lawnmower simulation. There is almost no chance to creep up on an enemy aircraft... if that aircraft is a bomber, its rear gunner will shoot you to pieces long before your prop gets anywhere near it. And a single engined type will maneuver out of the way 99% of the time.

Most of the collisions which occur online on the servers and which see one aircraft damaged but the other unaffected are due to connectivity and lag.

Thanks for everyone's patience. ;)


https://www.youtube.com/watch?v=xA2MceOT7uc&feature=youtu.be

A chance of this happens in CloD MP, Buzzsaw? Or just in a lawnmower simulation?

Tips for pilots in CloD: when engaging a bomber and winchester, go for the CF part... As usual, all recorder in SP, no "lag" or "ping" involved.

Now will you take this seriously or people will need to start flying fighters as missiles to you wake up? I already propose a simple solution, and as usual, mockery in response, because I don't have the SC, etc. At least TFS can put a real explosion event, with damage radius, when a plane explodes, as any other flight sim.

But all rigid body dynamics in CloD needs to be reworked or revised, searching bugs. It's unreliable.

♣_Spiritus_♣
Sep-02-2017, 02:36
.And is really sad that I have to waste so much time showing this to people that fly this sim and others that are working with all tools available and testers, and have all this denial in exchange.


I guess I didn't stop work on three other 3d projects for the last month to investigate this issue?

Anyways, while we appreciate everyone digging, finding, and wanting CloD to improve, another video showing the same thing as the last twenty isn't going to speed up any corrections we can make.

We are aware and we will continue to improve CloD in every way we can.

Thanks.

RAF74_Buzzsaw
Sep-02-2017, 02:53
Hello Lobo

We have already thanked you for pointing out the issue... and pointing out the issue... and pointing out the issue... ;)

As mentioned, we will be addressing the problem in TF 4.5 with temporary fixes which should solve most of the problems.

And in TF 5.0 we do plan to implement, (if it is reasonably possible) a prop collision model to provide the accuracy CoD is known for.

Despite what Lobo says, CoD does not have the worst collision model of any game out there... the opposite.

Thread is closed.