The Ship Model Forum

Lower fiberglass superstructure openings have been added to accommodate the mid-sphere forward-facing view ports, house the manipulator, forward and aft pan-and-tilt cameras. There isn't much reference material for the manip and cameras but I'll build what I can and add it in the next installment.

Once the DSRV is completed, all external parts not molded to the superstructure are segregated and oriented for 3D printing. Creating a 3D model that looks accurate when rendered is totally different than creating a 3D model that lends itself to 3D printing. Having the ability to render shots like this helps to ensure the accuracy of the model for later stages of development.

CC

[attachment=1]DSRV Descent BnW.jpg[/attachment]

[attachment=0]Port Side.jpg[/attachment]

Had to jump over to a couple of different models, but found time last night to replace the Splitter Keel, (located behind the Transfer Skirt) create the opening for the remote manipulator arm forward of the Transfer Skirt, installed forward light recesses and then threw some lights at it to see how it looks. This is the result. More to follow.

CC

Front panel complete. The shelf panel is next.

CC

[attachment=0]Frame 30-03.jpg[/attachment]

Back to the forward (Control) sphere "cockpit."

A rough layout is constructed before modeling individual panels. The center-mounted periscope dominates the shelf panel, so I built that first to aid in keeping the surrounding objects in scale as I compare the view with reference photos.

[attachment=4]Shell Outline 01.jpg[/attachment]

[attachment=5]Shell Outline 02.jpg[/attachment]

[attachment=3]Shell Outline 03 - Panels Extruded.jpg[/attachment]

[attachment=0]01-1.jpg[/attachment]

[attachment=2]02-1.jpg[/attachment]

[attachment=1]04.jpg[/attachment]

Another Sub-D (Sub-patch / NURBs) comparison example.

This is the basic hull of a MK-VIII SEAL Delivery Vehicle; I threw it together last night to practice with a couple of tools I've let gather dust, and it came together remarkably fast. All of it is sub-divisional modeling and all polys are quads. There are 666 (Yikes!) of them. If this were a hard surface model, this section would have thousands of polygons.


[attachment=6]SDV Bow - Texture.jpg[/attachment]

[attachment=5]SDV Bow - Textured Wire.jpg[/attachment]

[attachment=4]SDV Stern - Texture.jpg[/attachment]

[attachment=0]SDV Stern - Textured Wires.jpg[/attachment]

This is a MK-VIII I built in Dec (a little further along with a detailed interior) using hard surface modeling; It has 477,000 polys! Once the technique is learned, a Sub-D model can be built faster with far less polys.

[attachment=3]Build Progress 06.jpg[/attachment]

[attachment=2]Build Progress 07.jpg[/attachment]

[attachment=1]Hard Surface SDV.jpg[/attachment]

Transfer Skirt Modeling: Part 3

16. There’s just one thing left to do to wrap this lesson up: Add the fasteners we spent so much time preparing the skirt for. This requires six-sided disks for the bolt head and lower skirt nuts. One set is placed in position and a total of 18 are added around the Y axis using the Radial Array command.

To add a little extra detail, I like to go to the center of each fastener group and rotate them a few degrees so the result doesn’t look like a bunch of cg-created clones. Details like this will make your work stand out under close-up scrutiny.

And of course, the fasteners get a bevel too.

[attachment=2]Step 16 A.jpg[/attachment]

[attachment=1]Step 16B.jpg[/attachment]

[attachment=0]Final Transfer Skirt.jpg[/attachment]

So there you have it: a quick hard surface tutorial to create symmetrical objects. Using the techniques described, the entire DSRV outer base exterior can be created by Lathing a single poly around the Z-axis, then detailed afterwards. I’ll post more when I complete some additional modeling on this project. While I wish I could post an entire tutorial devoted to every aspect of re-designing this model, it would take so long to do properly, that I would fall behind on my modeling schedule with several looming deadlines.

Even though this is about a 15-minute modeling job, writing and editing a tutorial and organizing screenshots takes hours!

If one person learns something new, the effort was worth it.

Keep modeling and have fun!

CC

(PS: This goes without saying, but I'll do it anyway: [b]Save your progress often![/b])

Transfer Skirt Modeling: Part 2

9. The lower flange is finished the same way. The final point spread for the lower skirt looks like this:

[attachment=9]Lower Skirt Point Profile.jpg[/attachment]

10. The points are selected sequentially and when “P” is pressed, a polygon is created matching the lines on the reference profile.

For a sanity check, the positions of the two bottom inner flange points are verified to be in alignment with the bounding box edges at the correct position, (X=134.62cm / 2) since the bounding box is centered and only half are needed to establish the proper radius (r=1/2 the diameter.) There isn’t much math required for 3D modeling as opposed to CAD work, but a little basic geometry knowledge saves a lot of headaches later and ensures accuracy.

[attachment=8]Lower Skirt Poly Profile - Annotated.jpg[/attachment]

11. Building the skirt mesh is easy. Every 3D app has a different name for the next tool, but they all perform the same function.

A profile is used to generate a symmetrical mesh around an axis. In this case, the tool is called Lathe, because that’s what it does – it lathes a shape around (in this example) the Y axis. Creating a symmetrical shape around the Y axis is the reason for centering the bounding box in step 1. Other apps have descriptive names like Spin, etc.

[attachment=7]Lathed Lower Skirt Views.jpg[/attachment]

12. This is a good spot to point out the location of the intersection of [b]Potential [/b]and [b]Reality[/b]. With a drawing that contains some, (but not all) measurements, we have a [i][b]potentially[/b][/i] accurate depiction of the lower skirt dimensions. But in[b][i] reality[/i][/b], the upper and lower flanges need a little more “meat” on them to attach fasteners. This is why it pays to eyeball reference material in minute detail. I ended up extending the upper flange in each section of the skirt outwards just a little more than the reference drawing to allow placement of the mating flange fasteners later on.

This drawing – as helpful as it is, is only a guide, but it's prudent to avoid sole-source tunnel vision by referring to other sources of reference while modeling. This helps to minimize bigger problems that are discovered further along and require extensive, time-consuming repairs. When it comes to drawings, (that aren't blueprints) [b]Trust - but verify![/b]

[attachment=6]Revised Lower Skirt Upper Flange - Annotated.jpg[/attachment]

13. With the lower skirt shape complete, the upper skirt is built the same way.

[attachment=5]Upper and Lower Skirt Shapes Complete.jpg[/attachment]


14. Using my keen powers of observation honed to a narrow focal point by years of mistakes, several reference photos were used to determine the mating flange has 18 circular cut-outs. The flange polys of the upper section of the skirt are selected, copied, and pasted into a blank layer (or workspace) as a reference point to use in the background so an extruded disk of the proper size can be positioned, before creating a radial array of 18 of them. The disks will be used as “cutting tools” to subtract their shape from the upper and lower mating flanges on the top portion of the skirt. The cutting tools must penetrate all the way through the geometry in order to be subtracted properly. When the fastener seating surfaces of the flange polys are cut and pasted in a blank layer (or workspace) they retain their original position, so they make an excellent reference template. By selecting just the polys I need for reference purposes, it's easier to see what's going on, rather than look down in a top view at a mesh with hundreds, (or thousands!) of polys.

To recap, the basic workflow is:

a) Copy the flange geometry for use as a reference into an empty layer.

[attachment=4]Step 14A.jpg[/attachment]

b) Determine the size of the disk required to get the right spacing when 18 are needed, by making the array and going backwards to re-size and re-position the base disk. Trial-and-error are used until everything falls into place. That's one of the cool aspects of 3D modeling: you can go forward or backwards without trashing the base mesh when working in several layers to create cutting tools.

[attachment=3]Step 14B.jpg[/attachment]

c) After trial-and0error to get the positioning of the first (small disk) cutting tool correct, a final radial array of 18 equally-spaced disks is placed around the flange perimeter.

[attachment=2]Step 14C.jpg[/attachment]

d) Subtract the volume of the disks from the upper and lower skirt sections.

[attachment=1]Step 14D.jpg[/attachment]

15. And now . . . I’ll share one of the secrets of 3D modeling:

In the real world, very few edges, (including corners) are really flat.

Look at the edges of the objects in the room you’re in and you’ll see a slight glint where the edges catch the light. Adding a slight bevel or chamfer to an edge is the key to making photo-realistic 3D objects. While we aren’t striving for photo-realism in this tutorial, the mating surface of the upper and lower skirt flanges need to be chamfered to emphasize the joint. This small detail will pay off later when the 3D printed part is built in two pieces and bonded together.

Modeling chamfered edges to accentuate detail separates a beginner from a more experienced modeler.


Every corner/edge that is visible to an observer is chamfered to catch the light. The skirt sections started out with 120 sides (that was the Lathing sides number) before the flange cut-outs were made in Step 14D increased the polys (and their associated edges and points) even more. The chamfering command is applied to these external points manually, but with so many, it's easy to miss one and end up with a messed-up polygon. I might not see a trashed poly until later when it’s harder to repair.

The solution is to divide and conquer!

To save time and make it easier to see a missed point or edge, (especially since this part is symmetrical) the entire mesh is quartered within the Top viewport. The two open sides, (90 degrees apart) are capped with an end poly by selecting the exposed points and pressing the "P" key to generate a poly. This is repeated on the other open end to "seal" up the quartered skirt. A "watertight" mesh accepts commands easier. Some commands will not work at all on an open mesh.

After the edges are chamfered, the two end faces are deleted, (so "junk" polys aren't left inside the mesh) and the upper and lower skirt sections are mirrored in the X and then the Z axis, restoring them to their original form as shown below.

[attachment=0]Step 15.jpg[/attachment]

End of Part 2.

Modeling The Transfer Skirt: Part 1

[attachment=9]Transfer Skirt Reference.jpg[/attachment]

Sometimes we get really lucky when modeling and a drawing with measurements becomes available. The Transfer Skirt is one such example and it will be used to demonstrate a couple of tools common to hard surface modeling.

This is the drawing complete with metric measurements. The stair-step marked areas inside the skirt are Keep-Out zones to ensure the hatch on the downed submarine can open properly. The skirt is disassembled at a flange (located at the 89.865 cm height) for air and heavy trailer transport, so this will be a two-piece part. The upper section is part of the mid-ships pressure sphere.

An added bonus of this drawing is it shows the external shape of the lower hatch, it’s angular orientation when fully opened, the diameter of the hatch’s seating surface, and the pivot point location used by the hinges.

There’s a lot of useful information here!

[attachment=8]Skirt Reference - Metric.jpg[/attachment]

First, if it’s crooked, the reference drawing must be straightened; for this I use Photoshop. (Crooked drawings are common when they are scanned.) The straightened drawing is used as a template in a following step so it needs to be as straight as possible.

1. Using two of the drawing dimensions, a bounding box of 134.62 cm (X axis) and 89.865 cm (Y axis) is created to establish a “calibrated reference measurement”. Note that the box is resting on the origin of the Y axis (Y=0) and centered on the X axis. This model will be made to exact scale and scaled to 1/15 before printing.

[attachment=7]Bounding Box.jpg[/attachment]

2. Working within an X/Y viewport, the reference drawing is imported into the background and then scaled and positioned in X and Y so the two bounding box measurements from the drawing are aligned with the box built in step 1. (The bounding box edges are highlighted in green.) This display setup is saved as a configuration file named “Skirt” so I don’t have to import the picture and perform the alignment again when I switch to a different viewport.

[attachment=6]Reference Drawing Re-Sized.jpg[/attachment]

3. Now we’re prepared to create geometry! Using the on-screen template as a guide, the Disk tool is used to make the exterior outline, (basically a 2D "slice") of the lower skirt. The disk has 60 sides so the skirt won’t end up with unsightly faceting later – this is really noticeable when 3D printing and ensures most of the (avoidable) time-consuming post-printing sanding is eliminated. Less than half of the disk's 60 points will be needed to establish the arc, but this should be more than sufficient.

[attachment=5]Exterior Disk.jpg[/attachment]

4. With the disk sized proportionately and positioned to align with the lower skirt exterior, the 60-sided poly has served its purpose. Only a fraction of the disk's points, (those that fall on the exterior shape) are needed, so the poly is “killed” off --leaving just the points. The points that fall outside of the shape are deleted. I call this "Point Harvesting".

[attachment=4]Exterior Disk and Point Harvest.jpg[/attachment]

5. The same process is repeated on the interior of the transfer skirt with a slightly smaller diameter disk.

[attachment=3]Interior Disk Point Harvest.jpg[/attachment]

6. Three rounded radius locations are left to add. For these, I’ll position a 24-sided disk at each, and harvest the points I need to complete the shape.

[attachment=2]Radius Locations.jpg[/attachment]

7. This is the completed upper flange radius. The points of the poly that don’t contribute to the outline have been deleted.

[attachment=1]Upper Radius Close-up.jpg[/attachment]

8. Using the Box tool, a 2D box is placed over the points that mark the upper flange corners. the box poly is killed off, leaving four points in perfect alignment. This completes the upper flange profile.

[attachment=0]Upper Flange.jpg[/attachment]

(End of Part 1)

CC,

I certainly agree with what you said about learning to use a program. I have used thee DesignCAD program since 1988 (it was ProDesign back then) and have been one of the "gurus" on the DesignCAD Forum.

But every now and then a newbie would post a simple question that would force me to rethink how I used the program, and I learned something new! After32 years I am still learning.

Like you I use only a fraction of the commands that are available. For me the reason is that back in the dark ages I learned ways to get the program to do what I wanted to do. Many new functions have been added since then, but because I had learned old ways to do things I often didn't try to learn the new tricks. And in many cases the new tricks were simpler, faster and more effective than my old ways.

So I will be interested to see what you post and see if you can teach this old dog some new tricks!

Phil

Thank you, but I owe a big part of my skills to my teacher, William Vaughan. (He's worth a Google.)

William has taught thousands of students, produced hundreds of tutorials, and contributed to beaucoup magazines and books, --with a few of his own as well. He's forgotten more than I'll ever learn about 3D modeling.

This DSRV project has forced me to re-evaluate my standard way of modeling and think outside the box, and
use a few techniques that challenge the way I've grown accustomed to modeling. No matter how long one does this, there's always something new to learn (or re-learn.)

MS Word is a fairly common program with tons of capability that you may be familiar with. Honestly, what percentage of those capabilities do you actually know backwards and forwards? For me, maybe 25%. It's gotten me by thus far, and 3D software is no different.

There are well over a hundred commands in my modeling software and I use about two-dozen consistently to get results.

I reviewed one of William's books today and discovered a few really useful commands that have been staring me in the face for 15 years. These are what I call the Epiphany Days, and they're pretty cool to me because I can appreciate the added capabilities of what they bring to the table so much.

Once you have a certain level of experience, applications of a tool you may not use often become blatantly obvious. You can read about one in a book, but how you might use it as a time-saving device is what really makes it jump out and laugh at you for ignoring it. This is the point where education and experience collide to provide tangible, time-saving results.

To condense my long-winded post, there are plenty of virtual modeling posts online that proclaim, "Look what I made!" --Not that there's anything wrong with that, but the goal of this thread isn't about what I've made but to incentivize anyone who puts in the sweat equity, (meaning time and effort) with a 3D app. To hopefully motivate anyone getting started in 3D with an interest to create something they have an interest in, (in this case, ships) and demonstrate how I approach building a mesh. (There are multiple ways, so I'm just showing what works for me, based on countless mistakes.)

As a bonus, I'll include some observations I've learned along the way regarding 3D printing; specifically how to optimize a mesh so it prints well. I've paid for meshes that are absolute crap, and then spent days re-building them to the level they should have been at when the author put them up for sale. Shape isn't enough- optimization / efficiency of the mesh is just as important. It renders faster and prints better.

My intent is to verbally grease the skids for a longish thread that hopefully generates enthusiasm for newcomers and offers food for thought in the way of alternative techniques for experienced Modelers as well. It is geared more to 3D modelers than CAD jockeys. (I've inhabited both worlds.) I'll display some finished work and zero in on how I created specific parts of the meshes. It's easier to create complex things when you've mastered the techniques to build the little ones.

I can't emphasize this enough: Don't hesitate to ask questions - that's what makes threads interesting and instructive. My intent is to keep this thread software agnostic to enable you to adapt parts of my workflow to obtain similar results for your work.

Above all else: Have Fun! I find modeling to be incredibly relaxing when I have a few hours of uninterrupted time to devote to it.

3D modeling is nothing more than problem-solving --using a set of tools to create a finished piece of work in a logical order. Yeah, it's really that easy.

If you've learned how to ride a motorcycle, then you appreciate in the beginning how much deliberate effort goes into operating the controls so you get the bike to do what you want without killing yourself. At some point, you become proficient enough to just enjoy the ride and are one with the machine. The same goes for skiing or any other number of activities. 3D modeling is no different. Once you know the tools and develop the muscle memory to activate the controls without much thought, you're able to concentrate on not just the problem-solving, but the goal. It's a whole different experience, and for me, a lot of fun. It might be for you too.

All of us start at the beginning, so don't ever feel like what I'm demonstrating in this thread is beyond your skill level (today.) Practice productively and you will see tangible results. In this way, I can begin to repay a debt of gratitude to William --who taught me. When you figure something out, pay it forward and show someone else; that's what the 3D community is (or should be) all about.

Roll your sleeves up and prepare to excel!

CC

Impressive modeling skills illustrated in an easy to understand way.

Very interesting! :thumbs_up_1:

Part 2 (Can only load ten pics per post.)

Disregard the last two pics above; I don't know how they were tied into the end because they didn't show on my editable post.

Here are the hard surface and sub-d comparo boxes:
Do you see any major differences?

[attachment=3]Box with Rounded Edges - Texture Mode.jpg[/attachment]
[attachment=2]Sub - Patched Box - Textured Mode.jpg[/attachment]

The hard surface version with rounded edges has 266 polygons, the Sub-D version has 54! This makes a huge difference in Open GL computer performance when manipulating the model, (flipping it around, rotating, etc.) and when rendering in a lighted environment. Each object’s polygons reflect the computer-generated lighting in the scene and those computations add to the render time.

Properly built, sub-D models use less memory and render faster. For animation, they can be bent or twisted without deforming the mesh - something that cannot be done with a hard surface model. Character models are prime examples of organic, sub-d applications. Sub-d objects are infinitely easier to form into complex, organic shapes, (like a hand, for example.) In Lightwave, geometry is manipulated at the point, (vertex) line (edge) or poly level. Moving any of these in a sub-d model is similar to sculpting clay. Control lines are added to force the shape into the desired form. Sub-d models are usually made of quad, (four-sided) polys, though tris (three-sided) sometimes are unavoidable. Ultimately, all polygons are transformed into triangles, (sub-divided quads) when transformed to OBJ or STL format.

Okay, now that you’re familiar with two modeling techniques, let’s go back to the DSRV. This cushion, (one of six located in the mid and aft spheres) was created using Sub-D modeling. It’s got rounded piping on the top and bottom perimeters (made of four-sided segments that make it look rounded) and a zipper inside the center contour.

[attachment=1]DSRV Cushion - Textured Display.jpg[/attachment]
[attachment=0]DSRV Cushion - Textured Wire Display.jpg[/attachment]

It uses only 7903 polygons to achieve this shape; as a hard surface model it would be two or three times more. As a model increases in parts and complexity it becomes a PITA to work with if the poly count gets out of control. CAD models are built similarly but using different techniques and polys are not a consideration. They too suffer from performance lag as the model increases in complexity.

With all of this newly-acquired knowledge under your belt, you should be able to tell when I’ll using hard surface and sub-D techniques -just by looking at the wires. Test on Friday!

Stay tuned, this is going to get more immersive!

CC

For those of you who are advanced modelers, the following is geared toward newcomers, so please disregard. If you’re strictly a CAD modeler, this will be very different than what you’re used to.

Up until now, the majority of the modeling you’ve seen in this thread is of the hard surface variety. There is another major type of modeling called organic, which opens up unlimited possibilities to create complex shapes but it requires completely different modeling techniques. I’ll go over the basics as I proceed with this build and show the “wires” or mesh; see if you can pick out which style is being used.

Editor’s note: Everyone has their own workflow, and there are many ways to create geometry, so it’s important to remember that whatever methodology gets you the results you want, that’s the best. I use multiple techniques based on what I’m trying to achieve, in the shortest, most accurate way possible. 3D modeling is a tremendous time suck, so learning the quickest routes saves me time and customers money. For anyone interested, I’m using Lightwave3D, an Emmy award-winning 3D modeling and animation program used in television and full-feature movies that is intuitive and very powerful. When the model is finished, it is exported as an STL file for 3D printing.

For organic shapes, I use sub-divisional modeling (Sub-patched or Sub-D for short.) With a hard surface model, I would create a primitive (say, a box) of a set size, and maybe round the edges to match a particular radius. That’s great on a box, but with organic shapes a different technique is required.

I’ll start with a humble box to illustrate my point. The box as it’s modified will have a pair of pictures in two display modes, Texture and Wireframe Texture to highlight the details of the basic exterior and the mesh that it is based on.

I start with a 1m x 1m x 1m default box. Nothing fancy - six sides.

[attachment=9]Box - Texture Mode.jpg[/attachment]
[attachment=8]Box - Textured Wire Mode.jpg[/attachment]

The same box, but with rounded edges:

[attachment=7]Box with Rounded Edges - Texture Mode.jpg[/attachment]
[attachment=6]Box with Rounded Edges - Textured Wire Mode.jpg[/attachment]

Here's the same box turned into a Sub-D object; it still has six sides, but they are all curved toward one another.

[attachment=5]Sub-D Box - Texture Mode.jpg[/attachment]
[attachment=3]Sub-D Box - Textured Wires.jpg[/attachment]

In the next set, (Sub-patched Box) the box has an extra line added near each edge; these control the outer curves to approximate the earlier rounded-edge look of the hard surface version. (Note the glint on the upper left edge, catching the light.)

[attachment=2]Sub - Patched Box - Textured Mode.jpg[/attachment]
[attachment=1]Sub-Patched Box - Textured Wire Mode.jpg[/attachment]

End of Part 1.

Thanks! Both of you have inspired me with your detailed modeling skills.

As you are well aware, there is no such thing as "too much" reference material. When we're lucky, we get blueprints. After that, we have physical articles to measure, and at the bottom of the reference hierarchy, we end up with photos and sketches. While the major dimensions for USN ships are well known, the majority of the drawings I have were culled from navy manuals and other publications. I probably spend as much time researching as I do modeling, because I don't like to make things up.

When a part is ready for inspection, I'll align it with a photo placed in the background using the same orientation to ensure accuracy.

The mid and aft sphere cushions are placeholders. The completed parts will have zippers modeled into the mid-sections.

With as much fine detail this model will feature, I'm leaning toward investing in a Sonic Mighty 4K resin printer to capture the details the FDM Raise3D Pro2 printers I use can't come close to matching.

Attached - Two renders of standard (oval) submarine watertight doors with a little texturing applied.

Great work! :thumbs_up_1:

Nice work. I had seen nothing about how the DSRVs were built before your posts.

Where did you get the plans?

Phil

Finished the upper and lower hatches and animated their mechanisms as well. Both use a rotating hub to move linkages that move locking pins. Simple and quick-acting; the quick-throw handle on the bottom only needs half a turn to secure the hatch, and the wheel on the upper is about the same.

The lower hatch was more challenging because it uses a hydraulic actuator to swing the hatch 135 degrees from it's shut position - at least that's what was called out in my reference drawing. That's so the crippled submarine can open it's hatch and not interfere with the DSRV lower hatch.

Other than the manipulator arm, the mechanized components are complete and the overall pace of the modeling should go faster.

CC

I found some additional reference material, (including the lower hatch hydraulic actuator) which never hurts, but it forced me to buff up the WT doors shown below.

Mid-Sphere Upper Hatch - In between shoveling or snow-blowing into a driving wind, and no need to waste a perfectly good day off while stuck inside, I punched this out.