Best Practices for using the Bridge

[WIZARD]

Hi all

Not really a tutorial or even a particularly comprehensive tip but I thought I'd post here some of the tips I've learned for crossing the bridge between tS WS and Model side.

Note that I have not yet fully explored animation transfers across the bridge with 7.61 Beta 8.

This process allows me to use the far superior modeling environment of the Workspace and still have access to Model side tools and features - which for my uses are the LUUV plugin, the Modelside SDS, and often the fillet tool

Here's what I have been doing - I've been modeling a series of reasonably detailed medieval-style houses, upwards of 50,000 poly's intended for eventual use in an external program (the program I'm using is Vue 6 Pro Studio but this should apply to any prog loading .obj format files).

The optimal workflow I've discovered (and if there's a better ones please let me know!) goes like this:
First build your model in Workspace - Note: you can do this with the bridge turned on or off - off will give you better performance particularly with slower computers. Personally I model with the bridge on. This makes it easier should I need to use a tool that is modelside only (usually the fillet tool).
I find it helpful consider my UV coordinates as I build. I will only give any given sub-object a small number of different textures (this makes the re-texturing in Vue that much easier. I aim to balance the construction so I can use the simplest UV arrangements and minimum no. of individual textures with aiming for a practical number of sub-objects with a singe level object hierarchy. trueSpace handles a shallow (ie 1 level) hierarchy of complex sub-objects far more efficiently than it does a deeper hierarchy of simpler objects.
To facilitate this I (for example) may build all my window frames separately then boolean union them together (after having set each window frame's UV's). This means I have all my window frames as one object with one texture. I do the same with the glass in the windows, one glass object, one glass texture and so on. Obviously each modeling task will have it's own specific requirements so you'd factor that in as well.

Now we've built our model, covered it with placeholder textures - I usually use DX9 realtime shaders for this - and applied our UV coordinates. We've saved our scene as we've progressed and we now have a model that's perfectly usable in a real-time scenario or easily retextured with shaders appropriate to your renderer of choice within or outside of trueSpace.

Note: If you've used boolean operations on objects that touch each other it's important that you closely examine the geometry where the two objects met. tS boolean ops are likely to have added a bunch of extra vertices on the conjoining edges. This will cause problems later so ensure that you tidy up any geometry involved after using a Boolean operation.

Note: As my workflow will probably require fine tuning work to the geometry I do NOT group my sub-objects yet

Next step - Cross the bridge - Up till now tS will probably have been perfectly stable but the Modelside is more prone to crashes. Here is a key step - before crossing the bridge from Workspace to modelside save your scene to a WS library.
Cross to a model side view and reload the scene from your WS library. This will update the modelside copy of the scene. Using this process is the secret to success with the bridge.
Now we have a Model Side (MS) scene. trueSpace will have created Lightworks shaders that are reletively anologous to the WS Real time shaders. If you are rendering using Lightworks you can adjust any textures necessary as required and render away with volumetrics and other Lightworks features.

I want to export my model as an .obj file so I have some more work to do. Once I'm across the bridge into the old part of town I select a component of my model and apply one layer of Model Side SDS. One of three things will happen; the SDS layer will be added without problem, trueSpace will be unable to apply the SDS and will tell you there is a problem with the objects geometry (likely if you've used WS boolean ops without tidying your resulting geometry); or (rarely), trueSpace will crash.
If your SDS works OK (which is the most likely scenario) you can then remove the SDS layer. This step will have optimised your geometry for later export with the LUUV plug in.

If SDS fails or tS crashes go back to WS and fix any dodgy geometry. With a complex object it can take trial and error to locate the offending faulty geometry. I generally use the Separate Selection tool in PE mode to break my object up into smaller pieces, Save a copy of my WS scene in a WS library, cross the bridge, reload scene from WS Library and apply MS sds to narrow down the search for the offending geometry. Repeat this process repairing any faulty geometry until it's possible to apply Model Side SDS with out a problem. Continue until all your as yet ungrouped objects can be subdivided and returned to no subdivision without problem.

Return to Workspace view, save your scene to a Workspace library. Select all your required objects (I draw a marquee in the LE for this) and 'glue' or 'Encapsulate in 3D' to create one object with one layer of subobjects. Save your scene to a Workspace library, move to a MS view and reload your WS scene (Right Mouse in Library Stack View, select Load). Now you have a model that will export without problems.

Just for added piece of mind I usually apply a layer of Model Side SDS to my grouped object just to be sure. With a large object this can take a while but any remaining faulty geometry will be revealed by this step. Assuming the SDS layer applies OK you can then remove the SDS layer - again this can take a while. Once your model has returned to it's original non-SDS state it will export from LUUV with no problems at all, the model will not have holes in it (the SDS will have fixed that) and there should be no need to triangulate the mesh as it is already optimised for export.

Once exported to .obj your mesh will load just fine in Vue. Vue will not recognise the image files and you'll need to retexture the mesh in Vue but your UV coordinates should be intact and the geometry stable. Occasionally the export process can corrupt the occasional UV settings in which case re-apply UV coordinates in tS - MS or WS is fine and re-export to .obj.

This workflow can involve some back and forth work - the more you do this the better you will get at building elegant geometry that works first time. The first house I built using this methodology had 12 subobjects, 5 of which would not subdivide in MS and required tidying up. The second house had 15 subobjects and 3 faulty bits of mesh. The latest house, with twice the overall polygons of the first one in 14 subobjects only had one subobject that required fine tuning. Almost invariably the dodgy geometry is the result of using Boolean Operations of objects that intersect so avoiding that method, or tidying up the relevant meshes as you go will save much work later on.

Finally, the real secret to success is to save your Workspace Scene to a Workspace Library, then go to your Model Side View and then reload your scene from the WS library. This will update the MS version of your scene and will prevent the loss of textures and other issues that repeated crossings of the bridge can introduce. Since using this process I have had absolutely no problems whatsoever when crossing the bridge.

I can't vouch for the robustness of this method when using WS animation across the bridge but I have found that it definitely helps to maintain the integrity of your scenes when using the bridge often. I used to save using the MS libraries as well but I've found that is unnecceassary - for my outcomes anyway.

Hope this helps and good luck with your projects

All the best

Wizard.

[WIZARD]

Oh yeah - another tip I've found invaluable - before exporting your .obj from Modelside turn off the bridge!

HTH

Wzrd

[blakeo]

Thanks W!Z - I'm reluctant these days to cross back to MS - I'm quite happy in WS now - but this gives me confidence to use MS as needed.

I will be putting some of your tips to use soon

B

[v3rd3]

This is a great post. Thanks Wizard.

[Karthogen]

Quick question on this one. If you are saving your scene on WS and loading on MS, does the bridge need to be turned on at all?

the reason I ask, is because I have a scene that was bogging down my machine on WS so I moved it over to MS and it was quick er for a bit, but as I added more it is starting to bog down model side. I am still modeling on WS and then loading it into the scene on WS. By turning off the bridge I am thinking I could free up some resoures and speed up modelside a bit more.

[WIZARD]

Quick question on this one. If you are saving your scene on WS and loading on MS, does the bridge need to be turned on at all?

Yes - otherwise the scene will only load in the Workspace. Remember the bridge essentially links two separate programs and having the bridge on means both are running - and using CPU cycles.

the reason I ask, is because I have a scene that was bogging down my machine on WS so I moved it over to MS and it was quick er for a bit, but as I added more it is starting to bog down model side. I am still modeling on WS and then loading it into the scene on WS. By turning off the bridge I am thinking I could free up some resoures and speed up modelside a bit more.

I've also come to realise it's usually not 'the scene' that bogs down my machine but the way I use it. Slow scene navigation just means I am probably displaying too much in the various 3d views. I usually hide anything I'm not actively working on and use a single 3d view. I rarely use the open gl functions in the Modelside display. In Workspace I turn off Anti Alias, Bloom, Glow etc.For lighting and animation of complex scenes I use simple placeholders for the big objects. Using these techniques I find scene slowdown to be a thing of the past - assuming you have a reasonably capable PC of course.

I only ever have the entire scene visible when I'm actually rendering - and then of course my placeholders are all hidden.

Hope This Helps

[Karthogen]

Place Holders?

This a feature or just a way of doing something?

[WIZARD]

Place Holders?

This a feature or just a way of doing something?

I use placeholders when fine tuning my lighting or other effects. A placeholder is simply a low res object like a cube or a cylinder that you use instead of a hi poly model for composing, lighting and test rendering.

One way to work with placeholders is to first build your hi-res model, the one you want to render in your final image/sequence. Next, make a simple stand-in object the same size. I often export the 'invisible' attribute from the object render attributes dialog and encapsulate the model and stand-in placeholder together. This makes it very easy to hide or show the hi or lo-res version as needed and leads to much faster renders for setting up purposes.

HTH

[Karthogen]

Nice, thank you for the Tip. Scene slow down was and is always a pain. I don't have the newest PC, but I would think it could cope.

[xtetra]

Kind of a dumb question, but when you are working in WS and you go to save your scene, do you save as the old .scn (you can do that from WS right?) or the new WS side format .rsScn...or whatever it is?

Reason I ask is I recall reading somewhere that saving in the old format made things more stable...

Regarding making objects such as window frames...do you use booleans to create these of just poly modeling etc? For instance, to make just a simple rectangular window I usually create a cube and then boolean subtract a smaller cube to make the frame, then likewise for each pane. (Picture a regular double hung window) Once that's done and the parts are arranged how I want them I've been "gluing as sibling"

Reading this thread something tells me I could be going about it all wrong...

Thanks!