| OPT Creation Modelling General advice on modelling Once you have your preparations completed and are confident that you know what you want to do, and how you want to do it, then it is time to actually begin work on the proper model. You would generally start with the main hull/fuselage as this would provide something to arrange the other meshes on and, in arranging them correctly, can help you to decide on some of the size relationships (you may want to create simple placeholder parts in some cases to help work out the relative sizes and positions). One thing that is definitely easier to do as you go, rather than all at once, is to optimise the model by removing unseen faces and unnecessary vertices. Again this is especially true if there will be more than one copy of a particular mesh. This will also mean that you will have a better idea of how many faces you have per mesh, and in total, and so know more accurately how well you are doing in keeping to the level of detail you want. While you are making each mesh or section of model
remember that although 30 or 40 faces isn't a lot in
terms of the total facecount of most OPTs that this can
be the difference between a mesh producing the
triangling-to-infinity effect and needing to be split or
the mesh working correctly and not needing to be split.
Extra meshes mean more work in setting up the hitzones in
OPTech, as there are more meshes to be dealt with, and if
a distinct part (like a bridge or hull) has been split
then these meshes will need to be grouped under one
target ID in OPTech and also XWA can produce poor
lighting effects between sections of mesh. As you continue to work take the time to view your model from different angles as sometimes things can look fine in the top/bottom, left/right, or front/back views but a problem or flaw can become apparent at a different angle. One of the reasons for redoing Home One for a XWAU version was that the tip of the nose looked to be tapering nicely when viewed from the side or the top but it looked rather less good when looked at from an angle. Also take the time to look at your reference pictures again as you build the model up and check not only that the model is following your plans but that the result of your plans does look correct compared with these references. Sometimes, especially if there has been a delay between your making the sketches and when you started the model, you can find yourself making a good model which matches your plans perfectly but find that you have made a small mistake at the planning stage and therefore the model is actually inaccurate. Finally when building the larger meshes consider how
you are going to texture them. If you plan to use multiple textures then you
may find this easier if you create the mesh so groups of
faces for different textures will be divided along
straight lines (all faces ending at the same point)
rather than being interlocking (a face or faces in one
group ending further along than the others, and so being
partially into the next group). The more the groups of
meshes interlock the more the textures may have to
overlap to allow for this. More specific advice When modelling I generally deform and shape simple objects (aka primitives) but whether this is the best method for you depends on your own preferences and the modelling package used. What might be relatively simple in one package can be more difficult in another and different ways of visualising how a models structure can be built up suit different people. Obviously I can only advise on how I do things so if your preferred method, or your modelling packages preferred method, is to not use this approach than you may want to skip to Optimising the Model or even Textures and Texturing. There are a lot of modelling tutorials available which can be of help but trial and error is a good teacher and many of the tutorials are meant for (much much much) higher facecount models than would be appropriate for an OPT. The following mostly applies to either 3D Studio MAX (I use MAX 2, which is sufficient for my purposes) or, from what I have seen in my experimenting, to if you are using gMAX. The arrangement of the features in MAX 2 and gMAX is different but the same methods can be used. With an approach based on deforming simple objects by
moving or scaling vertices, groups of vertices, or faces
within the object you will find that there can be a
degree of overlap between the different sorts of object. Spheres. These can be used to produce
small hull elements such as bulges (1/2 a sphere
stretched) or larger hull elements like gravity-well
domes (1/2 a sphere) or engine nacelles (1/4 of a sphere
stretched lengthways). A simple teardrop shape is easy to
produce by selecting all vertices ahead or behind those
around the widest point and either stretching or
squashing them lengthways and then moving this group of
vertices. Cylinders. These are very flexible
and can be used as the basis for the main hull of some
ships (lots of segments lengthways), as wings where
applicable (squash lower vertices to provide a flatter
underside and stretch rear vertices to provide a more
tapered rear), and most of the time for the engines
(simple engine would be three segments, first segment
would be barrel, second segments end vertices would be
shrunk together slightly for thickness of barrel, third
segments end vertices would be shrunk together to a
greater degree and then moved so that they are inside the
barrel to form the nozzle). Capsules. As a capsule is a hemisphere on each end of a cylinder these can be very useful. You can increase the number of segments lengthways (by adjusting how many segments the "cylinder" portion is split into) without changing how many faces are used for the circumference. It can also save time to have a rounded end already on your "cylinder" rather than having to produce this yourself. Boxes. These are not just for "boxy" elements as if the box is divided into segments then this splits the faces along those edges and gives the box 6 or 8 (or however many) faces around its "circumference". Quite complex cross sections can be produced by making a box with several segments either vertically or horizontally and then scaling or moving these vertices (though this would apply to Cylinders and Capsules as well). General advice would be that if you do have to produce a quite complex cross-section to consider making this by producing a mesh with only one "lengthways" segment and then using the extrude tool (if available) to select the faces making up one end of the mesh and then extruding this as many times as needed. You could produce a primitive with as many "lengthways" segments as required and quite easily then scale / move all the vertices in each "column" but if you are collapsing any vertices together to produce the cross-section then this would be more fiddly to have to do multiple times. Another piece of general advice would be on
Triangle-asymmetry. In MAX at least the left and right
sides of a mesh have the same triangle-pair arrangements
( |\|\| or |/|/| ) rather than this also being mirrored
(to be ( |/|\| or |\|/| ). This can mean that although
the vertices are symmetrical that because different
vertices within each triangle-pair are joined together
that the model can look asymmetrical if the
triangle-pairs are distinctly non-coplanar (i.e. there is
a "bend" along the diagonal between them). Something else I find useful is the extrude tool where
a face or group of faces can be extruded out from or into
the mesh. At its simplest by selecting and extruding one
end of a mesh this can be used to add an extra segment. Sometimes when using the extrude tool and moving faces around (or even when just moving faces) you might find that the faces joining one side of the face you are moving to the rest of the mesh and the faces joining the other side look different. This is another example of triangle-asymmetry which can be solved by deleting the faces on the side that looks worse and rebuilding them so that they mirror those on the other side. The final thing I will mention here is that OPTech does not seem to have a fully useable means to scale the entire model. Some people have had success with dividing target length by the length displayed in OPTech, selecting all meshes,and then entering that percentage into each of the boxes for X/Y/Z scale but I found this caused meshes to be displaced out of position. When I mentioned the trouble I had with this method it was further recommended that once all meshes were selected that they should be centred on 0,0,0 to prevent this displacement but I still had no luck. Assuming you are as unlucky as me there are other
options. A common practice in the special effects industry is to use more than one model (whether CGI or physical) for a ship so that for long shots a less detailed model can be used (If physical then the model is smaller, and so more easily moved about. If CGI then the model will take less processing power to render, and so the footage be quicker to produce) and for close shots either a more detailed model or a model consisting of just (for example) a very detailed bridge section can be used. Computer games use the same principle to try to minimise the amount of your computers power they use and therefore keep the game running smoothly. The further away the player is from a ship the less detail of it he will be able to see and therefore the less detail needs to be shown. The ability to define lower detailed parts for a model and the distance at which these are used is one of the major advantages of OPTech (along with the two methods of getting a model with texture mapping set up into it) over AceDXF. A similar idea is also used for player-flyable fighters where you have three OPT files for three different purposes. There is the "normal" quite low detail Base OPT which is used for most fighters of that type and normally has an opaque canopy, then "zooming in" you have the more detailed Exterior OPT which is only used for the exterior view of the craft that the player is flying and normally has a transparent canopy so you can see the little pilot and a cockpit, and then at "maximum zoom" you have the Cockpit OPT which consists only of a very detailed model of the cockpit of the craft. Level of Detail parts To use Level of Detail settings you will obviously need to make the lower detail parts which will be used to replace your normal ones. How much work you put into this depends on what sort of detail you want your "low detail" model to have and on what sort of ship you are making. Fighters can have very simple low detail meshes as they are small and so at other than close range not much detail can be seen. This is also something that they should have as at the ranges which the low detailed model would be used at it is likely that multiple fighters would be on screen and so it is desirable to minimise the number of faces being used. Starships though can have quite complex low detail meshes, or at least quite complex low detail models, since different parts of the ship can be switching at different times and so each of these parts needs to be separately modelled. Naturally the more complex and larger the ship the more complex even the lower detail version will be. General advice, though it is more true for larger
starships, is to consider how complex you want the low
detail version to be. The simpler the low detail version
is the lower the minimum faces used by the ship will be
and the lower the processor load, but the more
complicated the low detail model is the closer the player
can be to it before the difference becomes apparent and
so the sooner it can be used. Once you have decided how complex you want to make your low detail meshes (from something like the TG VSD which was basically the normal model minus some details to something like many fighters which are almost just a box with pictures of the fighter mapped onto them) then you can start modelling them. Smaller details can be ignored as they would be set to just vanish rather than being replaced. A useful idea for larger "details" (including things like wings and the main hull) is to model them one level down, rather than having two or three sections for a curve only have one so that something which was (say) 8 sections around and 6 along would be 4 (or 3) around by 3 along. This will look quite horrible, and will actually resemble some of my early OPTs where I had been confused over the maximum facecount, but it will suffice for use at longer ranges and (especially with fighters) will avoid some of the jerkiness that can happen when a lot of ships are on screen. Illustration of
Low Detail and Full Detail model. Cockpit, Base, and Exterior on Fighters Although fighters can have the simplest low detail models and level-of-detail settings they more than make up for it by being able to have three separate OPT files. Thankfully you can at least rescale and reuse some of the textures (much of the time the same textures can be used on a base and an exterior OPT, though since the base OPT will not be examined as closely the textures can be rescaled to half-size, and some cockpit textures can be shrunk to an eighth size and/or merged with other textures used on the cockpit of the exterior model) and you can also use one model as a guide to help make another. Normally you would start with either the base or exterior models rather than the cockpit. Some parts of the fighter might be easier to model with more faces as you would have the extra vertices/faces to follow your sketches rather than having to work out a compromise so in that case you might start with the exterior OPT, while other parts might be easier to model with fewer faces so you can get the shape roughly right before using this as a guide for a more complex mesh.so in that case you might start with the base OPT. You will probably find yourself doing a combination of both and working back and forth between the models as you finalise the positions, shapes, and sizes of the meshes. One thing to remember is that any parts (such as wings) which you plan to set to separate from the main part of the fighter when it is shot down (see hitzone settings) need to be at the same position in the list of meshes or you may find that you will have one copy of that part from the one OPT flying off and exploding while another copy from the other OPT remains attached. This was a problem I had with the IRD. Something else to remember is that if you do plan to have a part exploding separately is to consider how the fighter will look once the part separates.The XWAU Supafighter is an excellent OPT but (to me at least) it does look a little strange for the dorsal/ventral surfaces to vanish but for the faces that were around the edge of the hull to still be there. One thing I will say, which some people will disagree with, is to caution against spending too much time on the exterior OPT despite the temptation to take advantage of the fact that you can make this a very complicated and nice looking model. If you have the time then by all means but just bear in mind how little players will see their own fighter from the outside compared with seeing other fighters (base OPT) or the inside of their cockpit. Once you have the exterior OPT model complete or near complete you can start working on the Cockpit. This should be a model of the fighter reduced to just what the pilot can see and so the simplest way to determine this is to place a camera in your 3D modelling package on what would be the bridge of the nose of your pilot. Swivel this camera around a bit and check how much of the nose or the wings (etc.) the player would be able to see. Chop down the exterior OPT to remove the parts the player would not be able to see, being fairly cautious at this point as you can always test the cockpit and delete faces in OPTech, and then use this as a basis for producing the cockpit OPT. Switches and controls should be modelled as should lumps and bumps and rivets and screws and corrugations. This model will be being examined at very close range for the entire time the player is flying this fighter (unless they switch off the cockpit view) and so needs to be of the highest quality (which it can be since it will be unique in the mission). As you work on the cockpit you may find that you make some changes to the design and if so remember if these changes are extreme enough to also modify the cockpit on the exterior OPT. General tidying up The general tidying up is something that you would probably have been doing (or would be best advised to do) while you were making the model rather than something you would do all at once when the model is complete. Tidying up the model is quite important as it will be being "rendered" several times a second and so any extra unnecessary faces / vertices could make a significant difference to whether XWA plays smoothly or not. It can be quite surprising sometimes just how much of a difference in the total number of faces can be made. Look for any hidden faces that can be deleted.
Look for any vertices that can be collapsed
together. Look for any very small faces which can be
combined into fewer larger faces Look for any extra vertices created by
Boolean-Merge. Look for any nasty extra shading Use an Optimise tool if available. Combining or splitting meshes As stated before there is a limit on both the faces each mesh can have before it suffers from triangling-to-infinity and there is a limit on how many meshes an OPT can have before a problem arises with the hitzones. Splitting meshes is the simplest decision as there are only two reasons I can think of to do it. The first reason is one of the methods of avoiding extra shading and is detailed there. The second and (most of the time) main reason is to avoid triangling-to-infinity which occurs when a mesh has too many faces. You can either split all meshes over about 200 faces or test the model to see which meshes need to be split. The latter is more work but considering how much the limit on the faces per mesh varies it may be better. With AceDXF you could get quite visible joins from changes in shading (shadows ending at the edge of a mesh) between the sections of a mesh that had been split. OPTech however does check for coincident vertices in different meshes when calculating the vertex normals and although it makes avoiding extra shading more complex it does also have the advantage of preventing the visible joins, which would be a more common problem If you are texturing outside of OPTech then you would normally split the meshes once texture mapping is complete so that all the faces in the original meshes which would have the same texture can be selected for the texture-mapping operations. Combining meshes is more complicated as is requires more though to decide whether to combine things and what things to combine. The first reason is to minimise the total number of meshes in the model as this simplifies some aspects of editing and as stated there is a limit. If for example you have a lot of small bulges (which won't be combined with the main hull/features) which are low in facecount and are not overlapping then you can use Boolean-Union (or the Collapse utility in MAX which can collapse multiple meshes into a single mesh) to combine these into fewer meshes without any increase in facecount. If the smaller details you wish to combine are
overlapping but you don't want these meshes to be altered
as in their case the problem with edge-wobble would be
minor enough to not be worth the increase in facecount to
solve then there are three options. When choosing meshes to combine for this reason to consider how the level-of-detail settings (which meshes are visible at what distance) in XWA are going to work. Details of roughly the same size should be vanishing at roughly the same distance and so should be combined together if possible rather than small details being combined with large (though it is better in some ways to have a small detail combined with a large rather than having two meshes). The second reason to combine meshes is to try to avoid edge-wobble or the venetian-blind effect. Merging meshes together, even if they are split apart
again, means that there are no intersecting faces and
this avoids the problem XWA has with calculating the line
of the intersection. From different angles this
intersection can be calculated differently and so you can
find that as you fly around the ship the "edge"
between one mesh and another wobbles back and forth. This
effect can become worse the more complicated the
intersection is, which means that craft such as Calamari
Cruisers (which are rounded) suffer particularly from it. Combining meshes can unfortunately lead though to a
considerable increase in the number of faces in the model
as the way the meshes are joining and the holes they are
producing in each other need to be defined. This
increases as more faces in each mesh are involved and the
more complicated the join therefore is. It is especially
true when using curved objects as, for example, a rounded
bulge on a rounded hull will produce a round
"hole" which will go across several faces of
the hull. The bulge itself will also be
"trimmed" along the join with the hull and this
new trimmed edge will also be curved and going across,
and splitting, several of the original faces. Remember to keep a backup of the non-combined meshes as some mistakes are easier to correct before meshes are merged together. Remember also that when you use Boolean-Union on intersecting meshes and look at the change in facecount to not be discouraged by the amount of extra faces that have been created as this change would include any small nearly invisible faces making the join perfect and the faces defined by any of the extra vertices that MAX might have created. Tidying up the join and the extra vertices can reduce the figure to a very significant degree. When combining the meshes you can either use
Boolean-Union repeatedly to merge each mesh in turn into
the composite mesh or (if available) in MAX the Collapse
tool could be used with Boolean-Union checked to combine
all the meshes at once. Since it is advisable to tidy up
the joins between the meshes though I would advise that
it might be easier to keep track of which joins have been
tidied if you combine a mesh in, tidy the join, and then
combine the next mesh in rather than doing all the
combining at once and then having to do all the tidying
at once. Remember also that if a component is going to be
individually destroyable that when destroyed it would
vanish, and so if you have combined it with and split it
from the hull (or other larger portion) that you could
end up with a hole in your OPT which the background can
be seen through. |