USF Confederacy in 3D | Blender

[3DShipWright]

Hey Folks, I am both honored and humbled to announce my next Blender build:

 

USF Confederacy is my 4th ship in Blender, yet will be my first attempt at a full historical recreation. While ‘Rose’ will always be my first love, I plan on taking many lessons-learned from my little-brig-that-could and apply them to a project on a grander scale.

 

Goals of USF Confederacy Build:

 

Accurate

With Rose, I proved to myself that it is possible to do comprehensive rigging and sails on a 3D model. I had seen some absolutely stunning 3D builds on MSW before I joined, but I noticed that 99.9% of the threads stop during the rigging process (I’m talking about full digital builds; not those done to 3d print parts later). Why did so much excellent work not get completed? Well, because in the 3D world rigging is actually where the overwhelming majority of the work takes place. I hoped to be among the first, because visually, it’s the yards, spars, sails and rigging that occupy the majority of the space taken up by any model ship.

Anyway, with Confederacy I want to bring the same level of realism to the keel, frames, planking, carvings and fittings – and have it be historically accurate this time. Thus, any deviation or parts taken from the designs of other ships is to be well planned out and documented up front.

 

Light

Rose clocks in at 6.2 million vertices and requires 14Gb of RAM and 7.5Gb of VRAM to render… without water, clouds, or anything else in the scene. Even if I never sell or distribute her (I’m not in this for the money lol) I do plan on one day doing renders of Rose duking it out with other ships in an epic naval battle, or perhaps sailing up the Thames in Victorian London. Ergo, I must do more with way, way less.

I’ve long been imagining/theorizing techniques to reduce geometry/improve performance and I am genuinely excited to try them out.

 

Beautiful

Photoreal or bust, ‘nuff said.

 

Excited to share – will post steps from the planning phase later today.

 

 

Best,

-Nate

[Montaigne]

Looking forward to following this, as Rose was an impressive build. 14 Gb RAM/7.5 Gb VRAM, that’s some major render requirements. I’m happy I’ve never reached those, as it would rule out my laptop, the Ol’ Faithful. 
Wish you fair seas with your new build. 

[wernerweiss]

Hello Nate,

as one of many builders of the Confederacy kit from Model shipways I look forward to follow your build of the 3D model.

 

Best regards 

Werner 

[3DShipWright]

Okay, here we go...

 

So, the obvious first step was to identify suitable plans. I first greedily gathered up everything I could find on Confederacy, good, bad, or ugly. Then I began the arduous process of sifting through them piece by piece. I ultimately decided upon Harold M. Mann's reconstructed admiralty plates circa 1979 for the National Maritime Museum. His plates are not completely without distortion, but it's minimal, and most importantly, they are complementary to one another.

 

So I imported them, used blender skew and rotate them ever so slightly (so the parallel lines are in fact parallel), then finally I scaled them.

 

 

Then I aligned each plan in the set with the appropriate axis from which I'd be working on them.

Next, I drew out the station lines.

 

 

I flipped/mirrored the bow station lines to the same side as the others, then positioned each one front to back using the half-breadth plan.

Then I used the aft most station line to shape the edge of the planking as it would appear inside of the transom and quarter gallery.

 

[3DShipWright]

STEP #2 Form the shape of the hull

 

Put simply, in this step we will create a continuous, flat (palinar) mesh that forms the shape of the hull. This is both my favorite and least favorite part of any build. If done right it will yield stunning results, and make the planking process a hell of a lot easier down the road. If done wrong, it will lead to major shading issues that appear as 'dents' in the hull or it can lead to insurmountable topology problems, such as I experienced with rose.

 

So as not to seem arrogant let me say up front that I have never gotten it right. That said, this is where I have a new approach I want to try out:

 

I want to have the mesh not only conform to the shape of the hull, but to also have the topology so each strake can be easily modelled. On a real ship, as the strakes curve around the hull they are wider in some spots, narrower in others, and are even cut as diagonal scarfs in some areas. Put another way, what we really need to care about is the latitudinal 'arc' at key points like the main wale, gunwale, channel strake and any cap rails at different levels (ie. the forecastle, quarter deck).

 

 

Note that as I cut away the extra geometry, it still extends beyond the plans in some spots, and that's okay for now: strakes that are too long can be fixed, strakes that are too short cannot be fixed.

 

 

So that's the goal, but how do we get there from our Bezier curves in step #1...?

[Martes]

RMG has a set of deck plans, should you need them, and an original lines taken off her.

 

https://www.rmg.co.uk/collections/object?vessels[0]=Confederate (captured 1781)

[3DShipWright]

Step #2 (Continued)

 

So what I want to do is

1. Draw additional curves that mark the top and bottom of the main wale, the bottom of the channel strakes, and different levels of planking.

Note: I DO care about inflection point s in the overall run of the planking. I DO NOT care about things that are cut into the planking such as gunports or sweep ports.

 

 

2. I added a 2D plane (6in X 6in), rotated it upright with the bottom edge flat at location z=0

 

 

3. I then added an array modifier (z-axis) equal to the number of strakes in that section (My research conflicted a bit here as my sources showed anywhere between 25-30 plank strakes from lapstrake to bottom of the main wale) I went with 30, under the assumption that it's easier to later subtract than add  

 

 

4. Next, I added a curve modifier with the deform axis set to 'z', then I used the eye dropper to choose one of the station line Bezier curves.

 

5. Obviously, it doesn't reach anywhere near the lower edge of the main wale so to fix this I went into edit mode and grabbed the top two verties and moved them upwards until the result reached the appropriate height.

 

 

6. When done with that segment, I simply duplicated the plane and changed the 'curve object' in the curve modifier to be the next station line. Each time I would then go back into edit mode on the new plane and readjust the height of the top edge so that the top of the 30th plane would touch the bottom of the main wale. When done, it looks like this:

 

 

Edited August 4, 2023 by 3DShipWright

[3DShipWright]

Step#2 (Continued again...)

 

Before we can join these together, we need to repeate the process for the upper segments.

My segments are as follows

a. The top of the rabbet (which also the origin of the overall ship , same as it is on plans) to the bottom of main wale (already covered in last post)

 

b. The main wale

 

c. The top of the main wale to the bottom of the channel stringer. FYI - I'm calling this the 'gun wale segment' because within this area runs a gunwale stringer, though interestingly, it neither follows, nor influences, the overall run of the strakes (more research req'd)

 

d. Bottom of the channel strake to the top of the lowest cap rail (forgive me, I don't know the proper name for this piece if there is one)

 

e. The top segment, which is actually multiple segments, but luckily the change in the arc along between the curves is minimal - so I was able to do it as one piece.

 

 

The 'yucky' part is that on ever level above the first one, we need to also align the 'floor' of each plane before we scale it vertically. This, sadly, doubles the workload on each one...

 

The next level up would be the main wale. Now, the models I've seen depict two heavy duty stringers along the top and bottom edges, and the planking in the middle is cut into a series of scarfs, lap-joints, and other assorted jigsaw pieces lol. However, the total width appears to be around 5 strakes, so that's what value I used.

 

If you're interested in following along the steps are:

1. duplicate the plane array from the lower segment along the same station line

2. change the array modifier to the number of strakes in the new segment (5 in this case)

3. In object mode, press 'g' to grab the whole thing, then 'z' to move it upward to align the bottom of the new array with the top of the array segment below it.

4. Go into edit mode and move the top edge up or down so the highest strake aligns with the next curve up (same as we did for the previous section)

 

 

 

 

After the main wale was done it was simply wash, rinse, repeat until all station line curves within each hull segment had an array of planes on it.

 

 

And, thankfully, I was finally ready to join the hull into a single object...

[3DShipWright]

Step #2 (Finale)

 

So joining the pieces together is actually fairly simple, but I do want to call out a couple nuances and "got-ya's" as I go along...

 

1. I first want to apply the modifiers for the plane arrays along each station line curve, and once again, do this segment by segment. Ergo:

• I select all planes along the rabbet to bottom of main wale segment

• Second, I duplicate and apply all modifiers - I duplicated them so I could retain a backup (which is actually what allows me to present this as a tutorial) ['Shift' + 'D', then 'Enter'] for blender users. VERY IMPORTANT - I must apply the modifiers on each object BEFORE I join them. Fortunately, Blender has an add-on that allows modifiers to be applied to every object you have selected. It's called 'Interface: Modifier Tools'. It can be found it under the main menu bar. Select 'Edit' -> 'Preferences' then a box will pop-up. Then select 'Add-ons' on the left-side menu then scroll down till you find 'Interface: Modifier Tools' Click the check box and close the window. There will now be 4 extra buttons at the top of the modifier stack, and the very first one just happens to be 'Apply All'. So with all duplicated planes selected, press 'Apply All', the join them into a single object by pressing ['Ctrl' + 'J']. I then hid the backups from view.

Fig 2. The add-ons menu in Blender.

 

Fig3: Screenshot of the plane arrays joined as one object.

 

2. Now that I have the plane arrays as one object, I want to connect them into one continuous mesh. There are a few ways to do this, but I did it by going into edit mode and loop cutting [Ctrl +R] each array down the middle like so:

Once I've bisected at least two plane arrays, I can then select the edges nearest one another and bridge the edge loops. TIP - make sure x-ray is turned on or not all verts may be selected.

 

IMPORTANT - the 'Bridge edge loops' command has several options in the resulting pop-up window. Make sure that 'Connect Loops' is set to 'Open Loop' and that you have the 'Merge' checkbox checked.

 

This may sound strange, but I then need to dissolve the resulting edge so that the only vertical lines in the mesh are that of the station lines. I will be adding interpolated cuts later, but I'm speaking from many of my past mistakes when I say that now is not the time to add them  

Anyway with the edge still highlighted just press 'Delete' and choose 'Dissolve Edges' from the pop-up box.

 

Caveat - the bookends: When I got to work on the arrays nearest the bow and stern, I have an extra edge of vertices (because there's not an array beyond that to which a connection can be made). Simply delete these vertices.

 

3. Lastly, I just repeat the above steps for each plane array, then do the same for the upper hull segments. Note that each hull segment will be it's own object for now, and that is fine. (FYI - I'll explain the use of the subdivision and why the foremost array is off the ground in the next step.

 

 

Smooth sailing to you all,

- Nate

 

 

[3DShipWright]

Basic Hull shape complete, starting to separate the keel 'stencil' into the various components adjoined with their correct wood joints (Nibbed scarfs, hook and butts, and rabbet joints - not to be confused with the rabbet line lol). 

Here a couple screenshots. From the images below, please note that I have yet to:

• Make the rabbet line twist correctly curve moving aft.
• Properly secure the main frames using the rising wood stringer and alternating sets of cross-chocks and rebates; i.e. the makeup of the keelson. 
• Add the final cant frame or hawse pieces to the bow. Interestingly, the plans I'm following don't show Confederacy having any dedicated hawse pieces, only an additional two cant frames that continue to walk up the gripe/boxing of the stem... If you have any knowledge of this, I'd highly appreciate it

[Martes]

Are you absolutely sure about the frame structure? Do those frames, depicted on the plan, represent single frames or pairs, and are you sure there should not be another frame between them that was omitted for model style purpose?

 

I have seen at least one model of the Confederacy that shows very different set of frames:

 

 

Taken from here:

https://www.shaffers-ships.com/usf-confereracy.html

 

The structure depicted here is British style, with pairs separated by filling frames. It may be possible (I don't know) that the ship could have been built by French style, where there are only pairs separated by slightly wider gaps, but it is very unlikely to be as you have on your model now.

 

Edited March 11, 2023 by Martes

[3DShipWright]

@Martes

 

So that's an excellent question, and something I've been stressing over as I do this part.

 

Short Answer:

1. At least as far as the main framing is concerned, Yes, I'm sure.
2. As to whether or not the plans I'm using are complete I don't know. I'll be using the Constitution and Essex to fill in any gaps in documentation, but only as little as possible.

 

Long Answer:

There was only one set of Confederacy frame plans I could find online, but there are several things about it that lead me to believe they aren't 100% accurate.

 

However, unlike my last project I did extensive research before I started this time. So here's what I do know:

 

• The frames depicted in the picture above are completely wrong. I've seen images of that cutaway model before, it's not unique to that museum. Even in the model, those frames are fake. Inside are solid cross section frames like on the Modelship Ways kit. A true tell is the placement of the doubled frames. There were many American frigates that utilized the classic British style, but in the revolutionary time frame those that did aligned them with the gunports and in Confederacy's case, the sweep ports. This meant an uneven spread along the keel, but structurally, a more sound design.

 

• Around this time Americans introduced pieces of southern oak into ship hulls. When they did, they utilized thicker but fewer frames overall. (How Ironic that "Ol' Ironsides" was technically softer than other ships) However, there do need to be either supplemental timbes or at the very least, filler deadwood around the steps, cant frames, hawses, etc. I still have alot left to do with just the primary frames, so I'm not risking anything by pushing the in between stuff off a bit.

 

Best,

-Nate

 

 

 

 

[3DShipWright]

Got the rabbet to twist properly!

 

[Martes]

If by using the term "main frame" you mean each gap is filled by a secondary frame it is possible. Also, each frame is not solid and consists of several parts, main or not.

 

If not, however, consider two factors.

 

First, here is an example of contemporary British frigate (the Triton of 1773), which was considerably smaller:

 

https://www.rmg.co.uk/collections/objects/rmgc-object-83032

 

Or a contemporary British 64-gunner (a ship of a comparable size):

 

 

https://www.rmg.co.uk/collections/objects/rmgc-object-66486

 

The French framing differed in having no cant framing on the ends (they were all vertical) and having each pair placed together (and not 2-1-1-2 or 2-1-1-1-2 as the British did).

 

And look in the thread here.

 

https://modelshipworld.com/topic/33702-frames-on-early-18th-century-french-ships/

 

Even the French construction in very extreme cases had a ratio of 2:1 wood:space.

Usually it would be between 4:1 and 5:1.

Your model shows 1:1. Even brigs did not have such a light structure.

 

Is the room-and-space parameter known for the Confederacy, just in case?

Edited March 12, 2023 by Martes

[3DShipWright]

@Martes

All true, you're just jumping the gun a bit...

 

Here, I went ahead and did a mockup of one of the frames as I want them all to look when I'm done framing.

 

[3DShipWright]

@Martes

 

And yes, I agree and will put additional framing between the ones I already have

[Martes]

28 minutes ago, 3DShipWright said:

All true, you're just jumping the gun a bit...

Just a little bit.

But there is a manner of making framed models with omitting one or several of the frames each iteration, so it may be a little dangerous to fall for that and take that as the real structure. And it's best to notice early.

 

Also, IIRC, the term "main frames" refers to the pairs of frames in British structures, and in that sense all the frames in French structure are "main". The single frames are called "filler frames".

 

But the composition of the two frames in the pair is different. One would have a single floor extending to both sides, the other would look like your last screenshot. Although I am not sure you want to go that deep into the structure

Edited March 12, 2023 by Martes

[Martes]

If Constitution is of any reference, it appears that she has

 

a) cant frames, British style, combined with

b) very densely positioned frame pairs, French style, but with almost no spaces between them

 

And, again, if I remember correctly, the British considered Chesapeake "overbuilt for her size", meaning she had even more wood than a British frigate of similar size would be composed of.

North America had no shortage of wood, but in several, especially early, cases it was just not seasoned enough and prone to rot.

 

 

 

In short, I tend to think that the frames you have on the plans (if they have any connection to real structure _at all_, since they look as based on British "as taken" plan, and the British did not record the plan stations as correlated with actual frame positioning, rather simply measuring the shape of the ship at certain distances from the perpendiculars) represent pairs (i.e should be divided to two different, but adjacent structures), shown with pairs between them omitted for modelling purposes.

 

Additionally, note that Harold Hahn himself writes in his book 'Ships of the American Revolution and their models' that his plans are highly stylised, primarily in intention to build models faster and for aesthetics.

 

This leaves the question of hull ends.

 

 

Because the spacing between the cant frames is also totally unrealistic and has to be, unfortunately, completely reworked - either referencing British contemporary models or (if you find it) a structure from Constitution.

 

Also, note that Hahn's reconstruction has also economized on the transoms:

 

comparing to both to British examples and the Constitution.

Edited March 12, 2023 by Martes

[3DShipWright]

Getting there - Frame by frame, piece by piece

 

[Martes]

Pair by pair, as each frame on your screenshot is in fact two frames combined together.

 

Note about the cant frames forward.

 

At that time it appears the standard was to use them up to the beakhead (Intrepid - and remember, your ship is +/- the size of a 64 with unarmed lower deck)

 

 

 

Or the catheads, which is the same when the bow is round (Constitution)

 

 

So I would suggest to rotate and compress the pairs depicted forward like this:

 

 

It would not be 100% accurate (it can never be, we don't know the real parameters of the frames), but it will resemble real structure closely enough.

 

 

 

[3DShipWright]

@Martes

 

Yep, good call and that's precisely what I'm doing now.

 

I've added all filler frames aft of the bow cant frames and before the stern cant frames.

 

 

 

I actually added narrow filler frames to sandwich in each pair of double around the ports. I'll end up joining these 'narrow' filler pieces with their neighbors and resplicing them to create the true doubles look, but even now the ratio and overall frame plan feels way better than it did before.

 

Finally, I've added the hawse pieces and cut them short to accommodate the bow deck (name?). I'm keeping the cant frames from the 45-90 degree radials, but I will thicken and nudge them back on one another to form a more authentic shape and close in some gap space.

 

 

[Montaigne]

Looking good, and I should add that I’m impressed by how fast it’s coming along. Not that speed matters when enjoying creative work, only that I wouldn’t have been able to keep up lol

[Martes]

1 hour ago, 3DShipWright said:

bow deck (name?)

 

Beakhead, IIRC. And the wall behind it is the beakhead bulkhead.

 

1 hour ago, 3DShipWright said:

I'm keeping the cant frames from the 45-90 degree radials, but I will thicken and nudge them back on one another to form a more authentic shape and close in some gap space.

 

As each of them is actually a pair (see Intrepid again), you can divide each of them to two and spread them so they cover as much area as possible. Also their thickness is slightly less near the keel and can be slightly wider towards the wale.

 

The French did not use this method, they had straight vertical pairs up to the hawse pieces at the time, so you can rely on the British sources at this place.

 

One thing of note - the Constitution has her straight vertical pairs until the first gunport (as opposed the Intrepid that had them starting from the forward end of the second gunport of the lower deck), and you can, I think, safely follow that example and make at least two of the cant pairs actually straight.

 

The rest of the hull until the last gunport should therefore be evenly distributed with pair-gap, pair-gap. Very small gaps.

 

But yes, it already begins to look much better

 

Edited March 15, 2023 by Martes

[3DShipWright]

I'm trying out a new method of planking for the Confederacy, and gotta say - so far it's working out rather well. No overlapping geometry and equal gaps between the boards regardless of the twist of the hull...

 

[3DShipWright]

Outer planking nearly complete. Rough tree rail complete. Caprails apart from the forecastle are finished. Trim pieces nearly finished. Mouldings will be fine-detailed later to a cohesive style. For anyone interested, the planking offsets from the hull frames are as follows:

 

Rabbet to Wale: 3in

Wale stringers: 5in

Wale Hook-and-Butt blocks: 4in

Top of Wale to Gun Deck Clamp (inboard on the Confederacy): 3in

Gun/Channel Plate/Scuppers Wale: 4in

Planking above Gun/Channel Plate/Scuppers Wale: 3in

Channel Stringer: 5in

Quarter Deck Stringer & Forecastle Stringer 5in

Trim Pieces: +2in, maximum from surface of their respective perch

Rough Tree Rail: 15in maximum

 

[Martes]

Is there some automatic process you use to generate the planking, or it is done manually?

[3DShipWright]

@Martes - Hey Martes, So I don't know whether to call it procedural or manual, but here is how I'm doing hull planking on Confederacy: 

 

1. The underlying topology of the ship has to represent the outer planking to begin with. Refer to the steps at the beginning of this topic for instructions on how I got the underlying topology to match the planks if you're curious.

2. In edit mode use 'mark seams' to visually outline the strakes, then the shape of each plank on the 2D mesh. Edge loop select makes this faster than you might think, I did the whole ship in under 15 mins

 

NOTE: I'm only showing the bottom of the hull for tutorial purposes, but the rest of the planking works the same way.

 

3. Use the solidify modifier to add thickness as desired, HOWEVER, DO NOT FILL THE RIM. Apply the modifier to get two separate 'sheets'.

 

4. In face select edit mode, press 'L' to highlight the front and backs of each plank and separate them using 'P'. You can do this in groups of planks as long as they're not touching

 

5. Go to each separate plank set object, edge select in edit mode, and choose select boundary loops.

6. Then you can bridge edge loops together, just make sure 'Loop Pairs' is selected.

7. Finally, join everything back into one object and change the 'Transform Pivot Point' to 'individual origins' Go into edit mode, select everything and press 'alt+s' then '-.1' to shrink each island along the normals by .1 inches (or whatever value looks good in metric units, just keep it small)

Edited August 4, 2023 by 3DShipWright
spelling

[Martes]

So for this method to work you need to fulfill two requirements: that your initial run of the vertices and edges should coincide with the planking plan and distances and to be sure that the Solidify would give a correct offset.

 

However be careful. It may be very well possible that since Blender 2.79 that I use the Solidify operator has been improved with it's results on complex surfaces, but my results with it, when I attempted to reproduce the surface of the frames and then extend it to planking was very unsatisfactory, especially around the bow and I had to revert back to manual editing (see my first post about the latest iteration of the Colossus).

 

I find it very useful to keep a "water" object, to be able to verify the smoothness of the waterlines so that more or less all intersections of the hull with it would not produce sudden angled surfaces.

 

Also, take a look at what happens around your keel. There should be no dent there. This is exactly the position the planking edge should touch the keel.

Edited April 2, 2023 by Martes

[3DShipWright]

@Martes - Good eye! Actually, that notch has nothing to do with the modifier, it's an intentional marker I made to notate the start of the stepped scarf for the cant frames and the transition of the rabbet of the keel to the rabbet of them stem. I will remove it once I flush out the detail of the gripe and boxing joints

[3DShipWright]

@Martes - The algorithms have improved, yes, but to your point they are not without fault.

Not sure if this will make sense, but what is missing is the ability to solidify while locking a specific axis. Here's a perfect example:

 

Take a look at the framing of the stern, specifically the counter and quarter piece timbers above the wing transom...

These frames should have even thickness along the curve of the lower counter, which they do, but also extend straight forward. However, they actually move towards the centerline because the back of the mesh is arced because the quarter board is arched. Blender doesn't have a good answer for this.

 

As a fix, I will end up deliberately over widening the front edges, then using the knife tool with x-ray on to cut these edges in line with the backs of each frame, but it's a huge pain lol

Edited April 2, 2023 by 3DShipWright