Sim9 feedback

[Dr.Sid]

Check out the new version:

http://www.commanders-academy.com/comsubsim

Post feedback related to this version in this thread.

[goldorak]

Great performance, I hope that the mk48 in a future release will be fully textured. Great looking torpedo by the way.
It would be nice to have an option that locks to the torpedo (from behind), so that when it changes course we follow it automatically. :tongue:
The torpedo in comsubsim begs for a fire control station. :biggrin: :biggrin:

[Molon Labe]

And a damage model :biggrin:

Some things that caught my attention:

1.from the periscope view, some "weirdness" took place when going from underwater to above water; a result of the difference in transparency above and below the water with a discrete linear boundary.

2. At flank speed, it takes about -2 to -3 degrees bow planes to maintain level attitude in the Akula. Is this intended?

3. There seems to be very little angular momentum, the subs (and ADCAP) continue straight almost immediately after the rudder is straightened instead of continuing to turn until a little opposite rudder is applied; the same goes for pitching, after the planes are leveled the sub stops rotating almost immediately.

4. The ADCAP only loses 1 knot of speed in a full rudder turn. There is also no roll that occurs for the ADCAP in a full turn, unlike the subs.

5. If the ADCAP broaches and reenters the water (tail first), it immediately snaps nose down. Is this intended? (I guess it does make sense because of the impact of the water on the rear of the torpedo).

[Dr.Sid]

Nice testing !

1) Near surface glitches. Yes .. this is problem of most games with water. I have some trick up my sleeve yet and I will try some approaches to fix it.
2) Pitching up at high speeds. That is caused by the sail drag. At the moment I don't know about effect which could counter it.
3) Angular momentum. You're probably right. Will tune.
4) Will look why it looses so little speed. As for the banking, it behaves neutral on purpose (center of gravity (COG) is in center), and also it uses active stabilization (not much needed anyway). Not sure how real torpedoes behave.
5) It should snap nose down. Why not ? It's not that heavy and at high speeds the waters is pretty tough.
There is however deficiency. The hull itself (or any part alone) now generates no angular acceleration (it does not induce rotation). All stabilization is now done by planes and fact they act away from center of gravity. Hull itself should lie on the water if it is light enough (case of surfaced sub or log of a wood) and it should rotate if just one part of it is in the water (case of the torpedo falling tail first). At the moment the torpedo turns down only because it's rear planes hits the water.
I have nice and quite simple way how to compute both static and dynamic forces on single body much better, I'll put it in the sim soon. Anyway, as I said, I think if torpedo hits water tail first it should turn forward. Even if the hull had COG right in the middle, there are planes even if small and when just the end is in the water there is big unbalance in drag.

[Molon Labe]

As for two, that seems like a realistic effect and not a glitch of any sort. The planesman should just have to compensate as a real planesman would.

[Dr.Sid]

I have one real planesman arranged for testing so I'll squeeze him.

[Red Ocktober]

really nice "cavitation bubbles" coming off the props, and nice underwater "fogging"...

night vision looks cool, but should be restricted to binoc or scope view only (?)...

i think sub "physics" is good enough to go with right now... small anomolies are acceptable, and may even add to the sense of realism (if in fact they are not)...

transition from above to below water can be handled by (1) obscurring view with bubbles during transition... (2) keep scope above and below a range that gaurantees the view doesn't intersect the water plane...

coming along verrry nicely... don't stop :biggrin:

--Mike

[Dr.Sid]

I now have nightvision everywhere because of testing. Sometimes I test so long that night comes or I move sub to other half of the planet and it is dark there. So NV comes handy. Btw. check out how the noise increases with amplification ;-)

Which reminds me .. I've already found major bug in this version. Do not use Teleport function, or it will try to generate wake bubbles from your last location to your current location which can take even hours depending where you are teleporting to.
I will make 9b with fixes as I get more feedback.

[XabbaRus]

OK had a quick spin. I like it. I take it at the moment you are limited to 2 ADCAPs as I couldn't get any more out of the 688 or Akula...

I like how you can crash into the seabed though I beached my Akula and couldn't get out.

Is it me or do the props seem to spin too fast? I also need to fix the pivots on the ADCAP.

[Dr.Sid]

I will get into thorough prop simulation soon, I will tell you then. Anyway it is quite clear that DW has unrealistically slow props. Also in DW props turn at constant rate so they had to choose something which would look acceptable at both flank and rest.

In previous version the physics was rigged so the sub could actually move on the ground. I moved it away. Teleport tool should be enough (if it worked in this version).

Yes, there are 2 adcaps per ship at the moment. They are never removed from simulation at the moment and I can see you guys popping 100 of them at a time :-D

Btw. try to set speed to 1 knot, and check how the prop shines ! The effect is ruined as it starts to turn faster.

[XabbaRus]

Also I tried to run it on my PC at work. We only have the onboard Intel GFX chipset. It loads up briefly and then crashes.

[XabbaRus]

Here is the error

AppName: sim.exe AppVer: 0.0.0.0 ModName: ialmgicd.dll
ModVer: 6.14.10.4299 Offset: 00196707

[goldorak]

Here is the error

AppName: sim.exe AppVer: 0.0.0.0 ModName: ialmgicd.dll
ModVer: 6.14.10.4299 Offset: 00196707

Blame intel integrated graphics. :mad:
They have really really poor support for even the most basic games.
You're better off spending 50$/€ and getting a dedicated low end graphics card.

The ialmgicd.dll is part of the graphics driver.

[Dr.Sid]

Did you try to set UseVBOs to 0 in sim.cfg ?

Which reminds me .. I improved Manual section of the wiki, read it, or even make it better if you can.

[OneShot]

One thing I'm still missing is a button to manually stop flooding or venting of the ballast tanks. Right now its either surface or dive ...

[Dr.Sid]

You'll miss it not much longer.

[OneShot]

A few map view related issues (which I hope are easy to solve)

- please add a symbol/icon for own unit
- what do the numbers in the boxes 3 through 5 in the upper left corner mean? Some sort of explanation/tooltip would be helpful
- 3rd box on upper right corner is stuck to -328ft ... whatever that might refer to

Btw. I noticed you added a reference for buttons for the trim tanks in the Wiki ... tho you still miss the reference for a Stop button for the main ballast tanks.

[Dr.Sid]

Uh .. I see .. the boxes in the left corner are just for debug. I could have turned some off. If you are interested it is from top: framerate, time compression, number of triangles drawn, number of terrain plates drawn, and the last one is distance between first and second sub.

As for the 328ft it is used when you are asking about depth on places where you have no terrain tiles. Only few is included in the demo, because of size. On places without altitude data, constant depth 100m is used .. which is 328ft.

Symbols are comming too .. for now use O key to center on ownship. You should find some different depths readings around Mallorca (the island where the subs are at the beginning).

[Dr.Sid]

Any new info about the crashes ? If UseVBOs:0 does not help, I'd like to know what GFX chipset you have, and if possible I'd like to get you on instant messenger to solve the problem.

[XabbaRus]

OK will try it at lunch time.

[XabbaRus]

OK setting that cfg line to 0 works

[Dr.Sid]

I might do 0 as default at the moment. VBOs helps somehow in speed, but only on cards which supports them well, which are new cards, which don't need that speed bonus much.

[goldorak]

I might do 0 as default at the moment. VBOs helps somehow in speed, but only on cards which supports them well, which are new cards, which don't need that speed bonus much.

No, actually all graphics cards going back 3 years pretty much gain performance from VBO.
Integrated graphics cards are another matter entirely.
No one should use an integrated graphics card for playing, thats just asking for trouble (driver-wise and performance-wise).
I say to keep VBO enabled by default, and keep an option in the configuration file to disable it. We need the full hardware performance, moreso when you will give us the option to enable directly AA and AF from the simulation.
Right now I have to enable it explicitely via drivers which is a pain in the ass.

[Dr.Sid]

Ok .. I only tested it on quite old now 7300go, where it had only marginal effect. I will try it on my new 9800.

Edit: on 9800 it has visible effect especially if 'view range' is maxed. Then it gives 88fps compared to 44fps without VBOs. With default view range the difference is 238fps to 234fps.

[goldorak]

Ok .. I only tested it on quite old now 7300go, where it had only marginal effect. I will try it on my new 9800.

Edit: on 9800 it has visible effect especially if 'view range' is maxed. Then it gives 88fps compared to 44fps without VBOs. With default view range the difference is 238fps to 234fps.

Yes, the difference in performance is staggering.
By disabling VBO you're basically using the opengl equivalent of DX 7 graphics. DX 7 was way limited because almost everything was processed on the cpu instead of the gpu.
And the delta in performance is bigger the more detail the scene has to render.
The same problem (bottleneck performance due to DX7 code path) actually happens with another open source simulator Orbiter.
I think that you should target the opengl equivalent of DX 8/8.1 functionality performance wise. Every graphics cards starting from the Geforce 4 (who rembers that ?) have support for DX 8.

[To be]

I notice that the cavitation bubbles are generated regardless of whether the prop is actually underwater. Additionally, it is possible to surf with the sub (full up planes, front and rear, all ahead flank, tanks blown) and motor along at 25 knots on the surface with nearly half the boat out of the water. I think drag forces are a bit too simplified.

[Dr.Sid]

Yes .. both things. The second is more problem of stability at the surface then drag. Drag should be more or less correct. But in reality the sub would try to lie on the water. If one half is above the water no lift acts on it and the other half would be forced up.
As for the drag, there is another problem with turbulence. Most subs today are slower on surface even if they have smaller frontal area on the surface, because of turbulence. This too is not taken into account at the moment.

[Red Ocktober]

i think that modelling the correct drag coefficients for the various submerged forms is gonna be quite a task... i would approach this from the top down instead...

since you already have speed and other performance data for various subs, wouldn't it be easier to model your subs to fit the existing data...

also... i'm sure that more than 50% of most submarine structures are under the surface at any given time... i think that drag is only one of the factors that you're gonna wind up playing with in order to get the fidelity that i'm hearing... it might be a lil bit overkill for this type of sim/game...

--Mike

[Dr.Sid]

Sure .. I don't want to do that. Although it's not THAT difficult, it's slow and hard to tune.
If I ever make it it will work with simplified shapes, like sphere, ellipsoid or tube.
At the moment I'm working on quite simple method which estimates center of drag&lift for submerged part in very simple manner and it too assumes ellipsoid shape. It's not much correct for some boats but it looks real enough.

[To be]

Sure .. I don't want to do that. Although it's not THAT difficult, it's slow and hard to tune.
If I ever make it it will work with simplified shapes, like sphere, ellipsoid or tube.
At the moment I'm working on quite simple method which estimates center of drag&lift for submerged part in very simple manner and it too assumes ellipsoid shape. It's not much correct for some boats but it looks real enough.

Maybe allow some ability to fudge with the calculated number (for example, making a U-boat with a deck-gun would be very difficult that way.

[Dr.Sid]

Sure .. you will be able to add individual parts and all basic coefficients (front drag, side drag, lifts, buoyancy and so on) are manually set. What I'm talking now is how they change based on movement direction and how much the boat is submerged, and especially how the angular moment acting on the ship changes.
It's practically done, it just showed some weakness in rotation dumping caused by drag, so I must improve it a bit.
I also found some wrong assumptions in angular inertia, it really was a bit weak.
I guess I would be able to make new release this week, with all discussed fixes and with first surface vessel as a bonus (it's a tug, no military beast, but it's fun to play with, and it will be doubly so when the collisions get implemented).

[XabbaRus]

Sounds interesting. I'm assuming you are modelling collisions properly so something like a tug can push around another boat with out incurring damage?

[Dr.Sid]

Yes, I plan to.

[Hitman]

Overall very nice looking already, congratulations :smile: In this early stage, and the graphics and models are already so much better than in DW....

Is it just me or does the periscope view have a bit of fisheye effect? In any case, I love it. Finally someone has decided that looking through a perscope must look like looking through a periscope and not AT it (Like being 30 cm away) :biggrin:

Have you already decided the layout of the interface? I recall having readed somewhere that you wanted a generic one with different graphics but similar layout for any submarine, so the user can get used to it faster. I take it you will be going for 2D interfaces, otherwise it would be a huge amount of work :eek:

I wish I could be of more help, but the only thing I will be able to do is some testing and maybe help with the 2D interfaces when you start them.

Keep the nice work.

[Hitman]

Copule observations (Excuse if someone mentioned them already):

-There seems to be no simulation of derivation (inertia) as of yet

-The roll effect is probably too soft (Even for a low center of gravity as a submarine)

-The bow dives when setting flank speed. May be I'm doing something wrong with the planes?

-When changing rudder violently from side to side, the roll is not natural, the sub centers and doesn't "swing" a bit before becoming stable.

-Response to rudder is nearly instantaneous?

No ranting intended! This is constructive criticism, the sim looks very good.