Advanced modelling in Lightwave 9 from a basic understanding

rsmayze01

a

IRML

there are many ways to start a model, for me it depends on the shape of what I want to model, I'll elaborate on that a bit further down...

it's perfectly fine to create parts seperately, you'd be making life very difficult for yourself if, for example, you modelled an entire trek ship out of the same geometry, the normal way of doing it would be modelling the saucer section, engineering section, and nacelles seperately and merging them later

so going back to how to start a model, I would be starting those parts individually, for a trek ship I would probably start the engineering section first, using a cylinder for that to begin, and shaping sub-d from there, the saucer would be a disc and sub-ds or if it was circular like the 1701-A it would probably be a modelled cross section lathed around, the nacelles would be a box and sub-d

don't stop there though, if you break it down like that you can still be left with some complex shapes, so you could consider each of those a model in it's own right and break it down further

also it's fine to just have intersecting geometry without it being fully merged, you usually only need to merge something if you want to round the join, and it's nearly always easier for selection and editing to have pieces that are seperate

rsmayze01

a

IRML

yeah I used to have that flickering when I first started lightwave, it tended to be when the intersecting geometry was at a shallow angle to each other, I don't see it any more these days other than in the opengl viewports

you can switch your sub-d mode from subpatches to catmull-clark which supports 'n-gons' (polys with more than 4 points), but it's still good practice to try and keep it all 4 point polys, in fact I almost exclusively use catmull-clark because it's better than subpatches, but I still try my best to build only 4 point poly meshes because you will get cleaner wireframes and less pinching that way

rsmayze01

a

IRML

it used to be when catmull-clark was first introduced to lightwave that it was a bit buggy and a lot slower to work with in modeller, but they fixed that up quite quickly, now it's very solid and not that much slower than the original subpatches

there are still reasons to use subpatches, for a start they are a different algorithm so when you freeze them you get a different wireframe which may be more desirable in some situations, subpatch weight maps still work better with them, they freeze at more resolutions so you can be more flexible with your final poly count, and they are a little bit lighter and faster to work with; but for me catmull-clark is better for the kind of things I do (hard surface modelling basically)

I don't think catmull-clark adds to rendering time, but that shouldn't be a factor anyway because you would likely freeze any sub-d mesh long before you were ready to render it - you would only build the basic shape with sub-d and then freeze it and continue adding details

rsmayze01

a

IRML

yes most likely, splines are another technique but I hardly know of anyone that uses them, and even then they use them to make a low resolution cage which is turned into sub-d for further refinement

I would work with both sub-d and standard poly modelling, I would use sub-d to make the basic shape of the hull - or to put it another way the un-detailed hull, no windows, no panels etc - then I would freeze it and carry on poly modelling details into this frozen mesh

rsmayze01

a

IRML

with lightwave 9's default menus it's near the bottom on the construct tab

count23

Another tip (and this really comes down to personal preference) is if your model is high detail and symmetric (like a star trek ship hull, or a battlestar's engine section), then only model half and then mirror it when you're completed. A lot of people don't bother doing this and try to model the whole ship as one solid object. Once you get the hang of merging objects and shapes together, there is no real need to waste your time with double the work for the same result.