Thomm

Sounds very sexy ! But ... (1) Doesn't this cause problems with additional animations ? Will there be a rigid connection between the gun and the towing machine ? (2) When the gun is coupled to the truck, are they treated as seperate objects or one (with regard to the targetting/hitting/damaging algorithm) ? (3) Can a truck tow any gun or just the one "assigned to it" ? (4) Different topic: How will wheeled vehicles turn ? On the spot (like only the Königstiger could) or in an arc ? (5) Yet another topic: Will the user interface display hit and kill probabilities (preferably in percent ) when I aim at an enemy vehicle ? Thanks, Thomm

To post or not to be ! Okay, here goes ... What are the visions you have with regard to the future of Combat Mission. In lengthy discussions we have clarified the principles behind the game engine, and - since it is all well thought through - it does not leave too much to improve (within the scope of the game). It just seems to be very good as it already is. So, how will you enhance Combat Mission ? <ul> [*]By adding new hardware and scenarios (e. g. Eastern Front). [*]By adding new gameplay elements (river crossing in boats). [*]By improving the graphics. [*]By changing the engine (can it be made better ?) Or will you go in a completely different direction ? We all (?) have seen what became of Close Combat and therefore I would like to know if you still have any dreams regarding the future of CM and what they look like ! Thanks a lot, Thomm P.S.:Yeah, I know it is an irrelevant post, but it is Vodka-induced, okay ?!

Dear Fionn ! ... comes from my academic background: Publish at all costs ! I also think that there is a huge difference between posting formulas, algorithms or even ideas on the one hand and to actually implement them ! How many competitors are out there who could abuse the mathematics, especially without the data to bring it to life ? Nobody can take the months of fine-tuning and debugging from BTS ! So why bother ? Seems that they are going to rule their niche for years to come, and the next game on the subject will for sure try to be "different" at all costs. So what harm would it do to distribute some distributions (haha) ? Regards, A very naive Thomm

Again very convincing ! Thanks for taking so much time ! Also good that you draw a line with regard to info disclosure. So that I know when to stop asking for more ! Actually I was close to asking for the source code By the way: Why do you ALWAYS have to kill a poor STUART ? Even worse: in front of your wife, tss tss Regards, Thomm

Sorry ! I did not want to behave like a spoiled child ! You know what: I wish you would put all your formulae on a webpage, one after the other, with all modifiers, standard deviations, everything and put an end to these discussions ! Oh, here goes the spoiled chikd again !

Yes, I think I forgot that the tank graphics will usually not be drawn to scale, which brings up a visual issue: I assume that AT shells will have tracer characteristics. Does that mean that they will just go through the enlarged tank without leaving any trace (in the replay phase), if they do not hit the physical tank ? I just realize that my assumptions on the shooting stuff were all wrong: Again it is statistics instead of vector algebra . Whatever. <BLOCKQUOTE>quote:</font><HR> Even if we could afford to make a very accurate polygonal representation of each vehicle (got a super computer to run it on?) <HR></BLOCKQUOTE> I beg you pardon, but since you can move detailed 3D polygon models (yes, they ARE beautiful) around in real time at e.g. 15 f.p.s you should be able to do a shell hit calculation every let say 10 seconds. Who will notice ? It would happen in the turn evaluation phase, anyway ! And the armored storage bin would not bother me the least ! And what special code do you need for the polygons ? You just had to store a value for armour thickness of the polygon, and maybe what group it belongs to (turret, superstructure, hull, ...) for damage evaluation. Okay you are right: Thou shallst not be immodest ! Lets wait for the supercomputers ! But: Do the kill probabilities change when the turret turns ?

To Lokesa: If I have understood it right then the center of the tank is used for LOS calculation because this calculation has to be done much more often than the shell impact evaluation. I think that the LOS picture with the Sherman in it gives the impression that the attacker could even spot a protruding gun barrel ! On the other hand this is not stated anywhere. If I recall the text correctly it says: "If it can be spotted it can be fired at" ... a redundant, though correct statement. Nevertheless you have to consider that while units are looking around all the time, tracing shots is a task which has to be done much less often, thus it can be done more accurately. You could for example cut a bounding sphere of the tank with the line segment between the position of the shell at the beginning and at the end of the time increment.If you find an intersection point then you start to check what particular polygon of the tank was hit (see Panzer Elite). I think this is done in Combat Mission, too. Pretty cool, if you think about it. Unfortunately I do not know how the CM damage model works, but if they really calculate where the shell strikes, then the rest should be fairly easy ! Regards, Thomm [This message has been edited by Thomm (edited 05-28-99).]

THANKS for the precise answers ( HAPPY )! Slowly the pieces fall into place ! Seems that we think along the same lines (or rather lines of sight, haha )! I like these abstraction games ! And - you won't believe it - I think I am actually running out of questions ! But does not matter, because I am sure that I have already qualified myself for beta-testing . Thanks again, Thomm

Melde gehorsamst, Herr Oberst: It seems that CM traces the path of the shell physically correctly from the moment it leaves the muzzle, that is, the path of the shell is determined at this time. The only influence the AI has on the shell flight path is thus the vector at which the gun barrel points at the moment the shot is fired. The rest is physics. So you would not need any modifiers for enemy speed, facing or such, but just trace the shell to where it hits. This is of course computationally expensive, but due to the turn-based system there seems to be enough ressources for this. Anyway, the problems are shifted from guessing modifiers to a rather complex gunner AI which has to AIM in the original sense, "shaking hand" and lead and all, since afterwards the physics engine takes over and that was it. I hope that Big Time Software will confirm that it works like this !

<BLOCKQUOTE>quote:</font><HR>No tweaking a Jagdpanzer IV's frontal armor from a 3.5 to a 3.65 to simulate several statistics at the same time! We have the correct real world values in there for the top and bottom hull, as well as the mantlet.<HR></BLOCKQUOTE> This is certainly more transparent. Do you also calculate where exactly (geometrically) a shell hits the tank such that one can tell from the visuals where the tank is hit (see also: Panzer Elite). If so, then the only chance to introduce random noise is when and how the shot is fired. I.e. the initial velocity vector of the shell determines its history in the CM environment. If that is the case I am again impressed, because that means that you have to have an AI for the gunner calculating lead and knowing the weak spots of the enemy ! Thomm [This message has been edited by Thomm (edited 05-27-99).]

Dear BTS team ! Lets say that the polygon soldiers represent the exact "center" of a squad. You also now exactly, where a shell hits. When you evaluate the effects of a shell hit, do you take into account that squads e.g. in a field are (usually) scattered over a wider area than the same squad seeking cover in a shell-hole ? That is, will a direct mortar hit in the shell-hole kill the entire squad, but only some soldiers if they are hit in the open (and supposed to keep a distance from each other) ? Thanks in advance, Thomm

<BLOCKQUOTE>quote:</font><HR>I can tell you that I don't put HMGs upstairs as often as I used to.<HR></BLOCKQUOTE>If this is a result of the game engine, then I owe you some respect (AGAIN ?! ). I just want to repeat a Yes / No question of mine (to keep you busy): Can squads "jump floors" everywhere in the house ? And another one: Is there a direct LOS within a house, even if in reality it would be broken by walls ? Or do squads have to get within a certain distance ("average room size") to get an unbroken LOS ? Okay, hope I will not keep you to long with the answers ... Thomm [This message has been edited by Thomm (edited 05-27-99).]

Approximated ? HOW ? What about the examples I gave ? What will happen in the game ! PLEASE ! You always had nice solutions so far ! Tell me ! How does it work ? Thomm WARNING: I recommended this site on cc3.gamestats.com ! This is what happened: <BLOCKQUOTE>quote:</font><HR>Looked at the site and the game does not look all good to serve as competition to CC4. IMO the graphics seem a bit "rough" and it does not appear to be highly detailed. Especially the close-up of the three SS-panzer grenadiers. <HR></BLOCKQUOTE> Get the good stuff out, SOON !

<BLOCKQUOTE>quote:</font><HR>HMGs also won't be able to fire effectively directly down (or up) stairs<HR></BLOCKQUOTE> Does that mean that we will see staircases in the building ? Or can squads "jump floors" everywhere in the house ? Will the AI move squads from the top floor to the ground floor if they are under attack and this breaks LOS ? Are there separate rooms ? Are there doors ? How do you take into account that 10 men cannot use the staircase all at once ? I BET you have good answers for all of these and my restless soul will finally find peace ! Thomm [This message has been edited by Thomm (edited 05-25-99).]

Dear BTS Team ! You did not answer the crucial question whether you bridges can be driven under or not ! For your motivation : Even in CC2 you could ! Thomm

Great features !! But what exactly will happen to my HMG on the top floor if a enemy squad enters the ground floor ? Will there be instant contact (like in CC) ? Or, to be more precise: Will they first fight the infantry squad on the ground floor and THEN the HMG team ? You said that there are two level houses: Yet I have seen four level houses on some screen-shots. Is there anything between ground floor and top floor, then ? Please be aware that strongpoints like bunkers and houses draw a lot of attention and should be modelled very carefully, because everybody will watch very closely when his/her men take a building ! Regards, Thomm P.S.: On Public Relations: Compared to the sreen-shots on the previews the shots on your homepage appear outdated ! You should add more and newer ones, in my opinion ! I would also be *VERY* cool to see a small picture of every tank you have finished in the vehicles list ! Especially since the quality of your tank models can compete with any 1st person simulation ! Give us some eye-candy, 'cause you have got it !! [This message has been edited by Thomm (edited 05-25-99).]

Dear Big Time Software Team ! Can you tell me how you intend to model houses, street-fighting and house-to-house fighting ? How much abstraction will be involved ? Will it be possible to shoot at the top armour of tanks from the top floors of multi-storey buildings ? Can you hide in the basement during an artillery barrage ? Will it be possible to place a HMG on the top floor and a infantry squad at the ground floor ? Thanks for your answer ! Thomm

Dear Big Time Software Team ! Thanks for the nice discussion on LOS we had, I hope you did not get too bored answering me ! Congratulations on your algorithm ! Now I think I understand the principles and innovations of your engine and you made me really curious of how it works in reality ! I already started to subconsciously measure LOS lengths when strolling through the streets and the woods; time to take it more easy, it seems . I wish you would have time to compile our discussion in a simple sketch showing the terrain mesh along with some tiles, subtiles and a exemplary LOS, because it is the knowledge of those beautiful technology details that makes ME buy games - and maybe others, too ... Thomm the Convert (looking for his Diners') P.S.: How fine do you subdivide the 60 second moves, i. e. what is the smallest time increment during the move evaluation ? One second ?

Hi Thomm, Yup, looks like you are just about ready to go back to your work Few minor corrections and I think this thread will indeed have run its course: <BLOCKQUOTE>quote:</font><HR> You use a grid of points to describe the terrain elevation. Distance between the points is 20 m. Elevation between grid points is calculated by interpolation.<HR></BLOCKQUOTE> Nearly on target. The tile itself is 20x20m, but the "distance between points" for elevation is actually smaller than that. Hard to give an exact number because a single 20x20 tile can have lots of different heights and patterns of heights within it. <BLOCKQUOTE>quote:</font><HR> Within the same 20 x 20 m grid you define "tiles" which you can assign a default terrain type (e.g. forest).<HR></BLOCKQUOTE> Correct. Each 20x20m tile is a specific terrain element, with possible additions (see below). <BLOCKQUOTE>quote:</font><HR> Additionally you use a grid with a ten times higher resolution (2x2 m) to add details like walls and hedges. They are abstracted as prisms with a 2x2 m base and a certain relative height. Absolute height of the (top of the) object is calculated by adding the interpolated terrain elevation. <HR></BLOCKQUOTE> Correct. Things like walls, building walls, roads, foxholes, shell craters, etc. are all done at this small 2x2m scale. Each has a defined height, which can (of course) be higher or lower than the one next to it. <BLOCKQUOTE>quote:</font><HR> Contrary to "walking" from (sub)tile to (sub)tile you use true VECTORS to calculate LOS. That is you calculate the actual intersection of the LOS with the actual terrain (I suppose one tile is split up into two triangles to do this). Some integer algorithm is used to determine which tiles (and subtiles) are crossed by the projection of the LOS on the horizontal plane such that you minimize the number of tiles you have to perform these calculations for, BUT the calculation itself is done in pure 3D geometry. <HR></BLOCKQUOTE> Right on the money. This is an area that 2D games simply can not touch. They use approximations for height and shape, not direct mathematical models. The smaller the scale of the 2D game the more that this matters (i.e. a squad level game is more affected by this than a divisional level game is). <BLOCKQUOTE>quote:</font><HR>Am I finally there ? Can we close this thread ? Can I go back to work ? Can I start the thread on pathfinding ?<HR></BLOCKQUOTE> Yup, time to go back to work! I would hold off on the pathfinding thing for a couple of weeks. We really haven't spent that much time on it at this point, so there isn't alot we can add to the discussion right now. Steve [This message has been edited by Big Time Software (edited 05-21-99).]

Hi again ! Okay, one last round before I go to sleep ! 1) My 2.5D statement and my column analogon were based on your emphasis on a 3D world. So, if I would use a tile system and I want to add a 3rd dimension I would interpret a wood tile as a 2 m times 2 m times e.g. 10 m prism. Thus I would know that this 2 x 2 area is covered with 10 m high trees. That means that I could look over the tile only if I climbed a very high hill. This is how I understand 3D LOS in combination with a tile system. 2) About the difference between "best" and "100 % realistic" you are right and I was wrong. From now on I will think of it as "the best LOS algorithm of a tile based strategy game of CMs scale". (Do we agree ?) 3) Quote: "Do not forget that CC is at a whole scale lower than Combat Mission". I have understood this and I think that I have mentioned it myself. 4) Quote: "CC is flat, 2D" Well aren't there elevation levels in CC ?! Can't you shoot over a two-storey building from a three storey building ? I would call this 3D LOS calculation ! What am I misunderstanding here ? I know about the flaws that Lokesa pointed out and I agree that Atomic was careless with the data and the implementation of their algorithm. I just want to make sure that I understand correctly how it is supposed to work. I would be shocked to learn that CC does not use the 3D LOS technique I thought it does, though buggy ! Of course you will not face problems like non-existing hills because of the 3D visualization of your terrain. To Moon: Your reply was very clear and understandable ! You made your (=BTS') point very well ! Thanks to everybody, maybe you will even write me more despite of the little misunderstandings we seem to have, though I think we are converging quite well ... Regards, Thomm P.S.: Do you use the A* algorithm as a pathfinder ?

Dear Big Time Software Team ! First of all thank you for this thread ! Maybe the best and most informative I have ever read ! Now my unavoidable comments: #2 In the case of trees and buildings I think the "tiles" have to be envisaged as COLUMNS with a certain extension in the vertical direction, thus being 2.5D. I cannot think of an actual advantage of a cube system in the kind of environment we live in ! About the vehicle bounding box: Thanks for making clear that there is none. But I am afraid that you have to be more humble with regard to your "best LOS algorithm ever" attitude. You might raise false expectations. About your comparison with CC: Please be aware that you avoid a lot of the problems the CC engine faces. Well in fact the problems are just transferred to another abstraction level, meaning that instead of single soldier objects you move around squad objects. So there is really not that much of a difference in the technology behind it. But I have to pay some respect to CC here because the engine has to tie the soldiers together as squads, something you entirely avoid. However, do not misunderstand me: In the course of the discussion you made it quite clear to me WHY you are doing the things the way you do, and I find it very well thought through, indeed. But, just for matters of political correctness: Do not say that your TECHNOLOGY is better than CC's because, well, it is pretty much the same. What you can rightfully state is that the SCALE of your game is more suitable for a WW II strategy game (esp. involving tank warfare). Thanks for your attention and for letting me contribute ! It is very pleasant to watch your project evolve ! Thomm

Pssssst, for the math: (3000/2)^2 = 2,250,000, not 225,000 ! Well, it is just ONE order of magnitude ... How does this link to 7 MB. If I take one Byte for terrain type and one for elevation I end up with approx. 4.5 MByte for all subtiles. And even less with the clever homogeneous tiles you proposed. Regards, Thomm.

Dear BTS team ! I do not see the difference in your tile system and the one of the CC series ! Is it just the higher resolution ? Why should it be more of a problem to have LOS checks for every single soldier ? CC has problems, but I do not think this is one of them. In your LOS walk-through you give a tank hiding behind a house as an example how accurate your algorithm works: "If you can see it, the game can see it". This is a dangerous statement, because the minimum requirement for it to be true would be to check the bounding box of the tank ! Are you doing this ? All in all I think that the tiling method is (still) a very elegant way of discretizing reality, but it is far from being a revolutionary approach, and while in 3D you have the advantage of an easy visualization of height levels, it might appear strange that soldiers are hidden by undergrowth yet you cannot visualize this undergrowth for obvious reasons. I hope you have some comments for me ! It is always a pleasure to discuss with you ! Thomm

Hi Guys ! Thanks for your replies ! I am sorry if I stepped on your feet, but, as you surely know, topics like sprites vs. polygons are a little bit religious and I am still under the impression of games like Myth II or even Syndicate Wars, where sprites just delivered, pretty or not ! Same goes for Panzer Elite ! But I want to stop here, because I should do my homework and read the previous posts before ! Yes, you are right that I did not read all of the posts here, the more I admire you to answering me in this depth and detail, thank you. Just be hard on me, since my previous post was bitchy enough to deserve a good flaming ! Mr. Fionn Kelly mentioned your attention to detail and I have to agree. But do not worry about my attitude, because you have already won my heart with those brilliant shock waves ! I think they alone are worth buying this game ! Sounds are dramatic, also ! Just an additional comment: I watched the videos (day.mov) from the shoot and noticed that after the salvo both the target and the guns were obscured by smoke and dust ! I am really looking forward to see this kind of environmental interaction in future games, what about you ? Thanks again for your answers and accepting the apology for my ignorance ! Regards and good luck, Thomm

I do not see to much visual effects in CM that would actually need a 3D acceleration ! Not even the smoke is transparent, which I think is really unacceptable and outdated ! Fire effects are years behind, too ! This might be a great game, but the graphics just cannot compete in these respects ! On the other hand there are these beautiful tanks and the *really* great atmospheric shock waves ! If you need more CPU power then use sprites for the infantry (I still consider polygon infantry an unforgiveable waste of resources, especially if it means to have three men symbolizing ten). By the way, in the videos their legs do not seem to be synchronized with their speed too well). So, please do not just satisfy the hardcore strategy players but also add enough eye-candy to attract ordinary people like me ! Good luck, Thomm [This message has been edited by Thomm (edited 05-11-99).]