Alamo VR/AR project

chippwalters

I tend to always model in SketchUp keeping components and groups solids if at all possible. This allows me model integrity as I go along, and makes editing these later much easier. For instance, just today I received word from Gary I had missed a window in a key place on the Church which was only revealed in drawings done well over a century ago. Because the window would have to pass through a wall AND a vaulted ceiling, it was easy to do using booleans.

As far as booleans go, I really like using both SU Pros Booleans as well as Dale Martens (Whaat)BoolTools 2. His Booleans can be somewhat tempermental, but they work when SketchUp's don't, so I encourage both sets. I also use his Artisan tools quite a bit as well as ThomThom's Vertex Tools.

Here are some more of the early developmental sketches. They were rendered with Fluid Interactive's Ambient Occlusion plugin then composited in Pshop with a SU line drawing export. I've since purchased SketchFX as well to streamline this process. I'll show some images in a later post.

chippwalters

One of the cannons was the first prop I worked on. The goal is to create a photoreal version of it which when rendered in Unity looked like the real thing. I also needed to create a realistic and properly colored version of it for 3D Printing. Just building out the workflow for it took over a weeks time but the results worked great.

Here's the SU model which as you can see the wheels and the support beams were skewed, so I created on-axis component accurate versions of them, then copied them and adjusted in other components.

chippwalters

Back to the Alamo church. After base approval, the model is tweaked and further componentized, keeping mindful of creating solids which could easily be voxelized in 3D Coat for adding further distress.

The workflow for creating the correct textures includes a custom shader I built in Unity, the details can be found at: https://public.3.basecamp.com/p/oTZSntqtnpobGTrE4WmpRq6P

As the shader does a great job with the normal bakes, you can see the hard lines of the underlying low poly geometry belies it's simple shape on profile views. So, my next step was to go into and add some very low poly distress to corners and edges where they might be seen frequently in profile.

Currently, I'm working with my new SketchFX Ex extension to figure out a custom render theme and video workflow for these in progress pictures. Here are a few tests..

chippwalters

Early stuff:
I started off by using the Oob topology plugins to grab a site terrain to begin work with. I then used Gary's notes, drawings and legacy Alamo blueprints to begin to layout the site. One great thing about SU is each Group/Component has it's own local axis and it made it very easy to work in different axis planes for each of the buildings. This allowed me to keep the model true to compass, so I can then later identify accurate lighting and shadows.

nickchun

Fantastic! I will be taking a look in more detail when I get a chance, its interseting to see how you integrate all the different software.
Love the clay and sketchy renders also.

pbacot

This is exciting! Beautiful, beautiful work1 The cannons alone, or the facade alone...are work to be proud of. Hope we get to see the final result!

chippwalters

@nickchun said:

Fantastic! I will be taking a look in more detail when I get a chance, its interseting to see how you integrate all the different software.
Love the clay and sketchy renders also.

Thanks.

@pbacot said:

This is exciting! Beautiful, beautiful work1 The cannons alone, or the facade alone...are work to be proud of. Hope we get to see the final result!

Thank you.
The facade renders are BEFORE I'll be adding the actual textures for the game model. I expect they'll be much better when completed.

Rich O Brien

This is intense! Following along

chippwalters

I then took the SU cannon into 3D Coat and added distress. There were three objects, one for each material (wood, bronze cannon, and metal). OBJs were exported and UVs added automatically in 3D Coat.

The model was exported as bake targets for Substance Painter and there materials applied, then exported and rendered in Unity.

Next the model was tweaked in SU to create a low poly version and baked and mapped again in Substance Designer. Here are the two high poly and low poly versions of the cannon side-by-side in Unity.

Then, the model was voxelized and tweaked in 3DC and made an entire solid for 3D printing. Many of the features had to be 'supersized' in order to meet the material thickness threshold requirements. Here's the under 1M poly finished 3D print model in 3D Coat.

Finally, colors needed to be added for 3D printing in Substance Painter. The same bake targets were added and then adjusted for the supersized details. Also, I only used diffuse maps (no normals, metallic, smoothness) as the color Sandstone material would not reflect the other qualities. The model was mapped, exported to MeshLab and converted x3d for final upload to Shapeways.

Here's what it should look like when I receive it.

Here's a video describing some of this process:

Mike Amos

Looking really good, something to keep an eye on.

chippwalters

A quick set of animation tests with SketchFX Es:

tuna1957

great stuff

chippwalters

Some more front detailing. The statues were quite the challenge as they started off as huge Zbrush files. I dropped them into 3D Coat, voxelized them and began to close all gaps I could find. I then converted them to surfaces, and 3D Coat has this cool tool called Object-ify (separate) which separates a mesh into only contiguous pieces. There's always junk lying around, and this lets me clean it up and then save out a very high (1M+ polys) mesh for baking.

I then use 3DC smart decimation to create a very small mesh (all 4 statues are under 900 vertexes each!) and export into SketchUp. I'll bake the original mesh on in Substance later and you can see the result.

pbacot

Pretty advanced stuff (and a lot of work). So you turn the image of the detailed mesh into a texture to apply to the simplified model?

chippwalters

@pbacot said:

Pretty advanced stuff (and a lot of work). So you turn the image of the detailed mesh into a texture to apply to the simplified model?

I actually turn it into a single normal map, which when applied, adds a lot of detail to the model. It also creates texture targets in the low poly model allowing me to quickly add diffuse maps for parts and pieces of the single OBJ. There's a video on the process: