You seem to be jumping to conclusions here which, unless you have seen the model mesh, you cannot be sure about.

9000 polys does seem high but it may be because the modeller has used them to include higher levels of detail than is found on the stock models. Without seeing the model mesh you cannot be sure.

Experiments by others have shown that the number of polys in the model has very little effect on the fps in game. What seems to be more important is the size of the textures used to skin the model.

I'm looking forward to the D.520, whether modded or official.

Ashe

For example
swordfish kate Ju 52 C47 fliable any news about avengers problem ?

True, but looking at screenshots the model is very smooth in the same way that I've gotten using "sub-surface mesh." Also, the D.520 has very smooth lines. It would be hard to add that many polygons modeling airscoop, tailwheel, engine exhausts and so forth. So, my semi-ignorant guess is that most of the polys come from smoothing out the lines of the wing and fuselage.

Don't get me wrong, the D.520 model is very nice looking and I look forward to flying it, but I think that had the modeler been willing to work a bit harder he could have made it to DT's standards. After all the MS.406 is a very similar looking plane and it was modeled to IL2 standards back in the IL2 Sturmovik era.

This is very interesting indeed. Do you have a link to details of these experiments?

Fair comment.

The original ones I am thinking of related to in-game testing of a scratch built Lancaster model. The LOD0 model was, from memory, greater than 20000 polys.

Very recently I have been trialling an IJN Kongo BB made by others.
The LOD0 model for this was around 160000 polys !!!
I didn't expect it to work, but it did. FPS was 65 on my mid level system compared with 62 for the stock HMS KGV model. Beta version available here: http://www.sas1946.com/main/index.ph...html#msg207259

The high polyon models may work with the game (although it wasn't programmed to necessarly do so). The difference will get clear, if you have lots of that models in one scene.

But your statement, that large textures have a bigger impact than more polygons, is generally correct.

axis little problem.
 

Sorry if this was reported before. Could someone check the engine one axis and engine one prop pitch: if it is set to simetric it runs ok, but if the simetric switch is turned of, the values start from 10 or 11%. When i am throttling up it stays on 0%, i keep mooving the lever... and suddenly BANG 11%!
It is not serious stuff because we can use the simetric switch, but if it is not very dificult to solve it i woul appreciate.

Thanks to DT, and the community.

Presumably, this was a closed beta-test of Oceanic Team's Lanc.

Do you know if there were any other factors which could have contributed to better or poorer fps, like map or visibility levels? (In a cleverly-modeled sim, I could see a Lancaster model with 20k polys being a lot easier on FPS at night.)

Maybe this has to do with differences between ships and planes in the game. For one thing, a fair bit of a ship model is hidden below the waterline. Another factor is that ships don't move or maneuver nearly as fast as aircraft.

Mind you, I'm not trying to dispute you, I'm wondering if the IL2 engine handles ships (limited AI and maneuverability, often seen at lower LOD levels) compared to aircraft.

If you're right, then it makes a whole lot more sense for TD to cap the total number of pixels in textures, rather than the total number of polygons in the model. IIRC, currently texture size is 125 pixels squared for ground objects, 250 pixels squared for vehicles, and 1026 pixels squared for vehicles. I don't know what it is for ships, but I could see total pixels being huge for a big ship.

Perhaps the limit on polygons for models is more part of the "non-compete with CloD" part of TD's agreement with 1c than an actual limit of the game engine.

In fact its correctly:

128, 256, 512, 1024 pxl square (I guess, because of the 32bit game structure), while for ground objects 256pxl or below is still standard, but can be increased to 512 pxl square, if reasonably.

1024pxl square is max, preferibly 512pxl square... you can have as well one large and few smaller textures for one model though.
Polylimit for larger ships is ~5000 to ~8000 tris, depending on object size and complexity.



No. That has nothing in common. Its rather, that the engine is build for it, so its the best to keep it, so its on the save side. And also because the game has already so much content, that it would go inharmonic looking, if new content would be made with higher standards. And its maybe understandable, that we are not able to remodel every existing content in game to higher standards. That would mean a new game (what CoD is in fact).

But the stock model, HMS King George V BB has 20016 polys for the LOD0.

and the stock model, HMS Illustrious CV has a LOD0 of 16815 polys.

-- So would it not be reasonable for modellers to take these as a guide rather than be limited to 8000 for a capital ship?

In the work I have been doing I have used HMS KGV as the principal guide to model size and structure with the result:

HMS Warspite BB - LOD0 15998 polys
RM Caio Duilio BB - LOD0 19386 polys

If used correctly by mission builders there would not be lots of these models in a scene and if realistically spaced only one of the models would be displaying at high lod level at a time. -- unless your doing a Pearl Harbour scenario I guess.

When used in the game the biggest fps hit is seen when the AAA all opens up at the same time from multiple warships. This is not related in any way to the complexity of the 3D model itself, just the effects generated by the number of guns.

Ashe

Well, I have different models then. :grin:


Well... most of the ship models (and especially higher poly ones) came with Pacific Fighters and were 3rd party work. So I suppose, the 3rd party people didn't care to orientate on the given limits and the models were nethertheless pressed into the game with one eye shut (due to the well known time and money problems, that Luthier had with PF). But thats just a theory of mine.
As well it could be, that with Pf the limits were raised - without me knowing about it.
However, you are right, current work should orientate on the given and I see no problem with higher polycount for big warships/battleships, as they rarly come in dozens.

However... it must be reasonable! And if you think about it, its expectable, that the King George V is rather high poly, with all the masses of tiny anti-aircraft weapons on deck (the AA guns on the after deck alone are already ~1000 tris). Marat on the other side is 'naked' and can live with a third of the polycount.

While Warspite is still ok, being the same category as KG-V, the Caio Dulio is IMHO over the edge.
The trick is, not to look for the maximum limit, but for the average values - these to orientate on.

See here the numbers of a few of the bigger stock ships (only LoD00, NULL mesh deleted) - in order of polycount decreasing:

King George V. - 15867 tris
http://img838.imageshack.us/img838/5391/kgvi.th.jpg

Illustrious - 11444 tris
http://img856.imageshack.us/img856/8254/illy.th.jpg

Tirpitz - 8020 tris (!)
http://img694.imageshack.us/img694/9297/tirp.th.jpg

Kent - 5961 tris
http://img257.imageshack.us/img257/3420/kentg.th.jpg

Niobe - 4080 tris
http://img641.imageshack.us/img641/8831/niobeb.th.jpg

Marat - 3845 tris
http://img440.imageshack.us/img440/8384/maratp.th.jpg

Illmarinen - 3036 tris
http://img72.imageshack.us/img72/3237/illma.th.jpg

Note, how the rather small destroyer is higher on polycount than the battleship Marat. As I said, its depending on size and complexity, if its really reasonable.