French Protected Cruiser COSMAO 1889

[taylormade]

Wow, I've been searching for weeks trying to find more photos. These are fantastic. Do you have a source I should know about? Any chance of getting higher resolution files? Thank you for posting these!

[setori]

give me your mail i can send bigger pics

[taylormade]

PM sent. Thank you!

[taylormade]

From the photo setori posted, it's obvious that the plans and the actual ship are quite different. It appears they added on to the pilot house to enclose the steering gear and binnacle and they must have widened the rear of the bridge as there are now doors on the side of the pilot house which would have been impossible according to the plans. There also appears to be railing around the top of the pilot house providing another platform for the crew. There is also a searchlight on the forward mast which is not shown in the plans. And there does appear to be some sort of structure with windows under the bridge. Amazing what one photo can show - and really change things around.

[taylormade]

After further examination of setori's second posted photo of the Cosmao, I believe this shows the ship late in her career. She was the longest serving ship of her class, surviving until after WW1. By that time I sure many changes and improvements had been made and her appearance had changed dramatically. Notice the the lower decorative hull trim is missing. The plans show this feature as being made of wood. As far as I can tell it serves no purpose other than to set off the lines of the ship. I surmise it became damaged and rotten over the years and was eliminated for expediency's sake during a refit. The many other changes were necessitated by her service requirements and improvements in technology over the years. Would they have had electric lights on a ship in 1890? At any rate, I have decided, as mentioned before, to model the ship from the builder's plans and represent it as the builder's model presented to the French Navy Board for approval. Therefore, she will be shiny and new with no aging, rust streaks or other wear.

One question - every photo I have of the ship and her sisters seems to show a gray paint job. When did the French Navy stop painting their ship's hulls black? Would Cosmao upon completion in 1890 have a black hull? Would the upper works be gray? Was any buff paint used in the Mediterranean Squadron in the 1890s? Thanks everyone for your interest in this project.

[taylormade]

Progress on the mid-deck area. Just roughed in for shape and scale.

[taylormade]

I'm having some computer problems, so progress has been halted until I get things working again.

[taylormade]

Back to normal. Work on the middle deck.

[taylormade]

As much as I have enjoyed working on this project, I'm afraid my computer is just not up to the task. The files on the Cosmao have become so large they are crashing my PC at every turn. I will continue my research and will start on the actual model as soon as possible. Thanks for stopping in and checking my progress up to this point. Sorry to disappoint you folks.

[DrPR]

I understand your problem. I built my 3D CAD model of the USS Oklahoma City CLG-5 in four separate files (hull, forward, middle and aft superstructures), each between 200 megabytes and 320 megabytes. The total is just over 1 gigabyte (about 2.8 million objects and 22 million points).

The program became very slow with files over 200 megabytes so I created the separate files. It took about an hour to do an initial render of each file. When I tried to combine all the files the program choked at about 750 megabytes. Don't know why - there was about 28 gigabytes free RAM and half a terabyte free disk space.

I ended up creating renders of each part using identical view angles, view distance and lighting. Then I combined the images in Photoshop.

Phil

[taylormade]

I was at almost 2 gigabytes. My computer normally cooks, but I have too much detail and too many subdivisions for this to work. I was considering your approach, but the real purpose of the model was to be able to move around the ship in a virtual world. The renders were just snapshots and not the final purpose of the project. In the meantime, I'm going to try a Thornycroft second class torpedo boat, that should be a walk in the park after Cosmao.

Beautiful job on your model - I know how much work went into that.

[DrPR]

I had the same goal - a walk around video. I modeled everything 3/16 inch (4.8 mm) and larger (1:1 scale). About a quarter of the model was nuts, bolts, screws and rivets (but I did not model the threads!).

I took a lot of shortcuts to reduce file size. For example, all of the cables strung in the towers and along the superstructure sides were actually octagonal extrusions instead of cylinders. They generate far fewer facets (smaller file size) and render with complex shadow patterns that more resemble braided cables than smooth pipes. Just don't look too close!

And instead of solids much of the structure is just simple zero thickness planes. Again, smaller file sizes. But since I started the project in 2004 3D Printing has come of age and very little of the model will print (2D planes have no thickness). If I want to use parts of the model for 3D printing I will have to redraw them as real solids. That isn't as much work as it seems because the research is already done and the dimensions are there.

Someday computers will catch up and I will find a program that I can use to create a single file, and maybe even create the walk around video! When I started 17 years ago the computers and programs couldn't handle a 100 megabyte file.

Phil

[perotech]

Just found this thread, and thought I could add to the companionway detail mentioned earlier in the thread.

The reason the top half of the curve is thicker, and object "B" protrudes is that it would appear to be a handle for a sliding cover, I assume to enable better headspace when ascending descending in fair weather.

I have attached a photo of HMS Gannet (1878), with a similar looking hatch. Hope this is of some help

[Iceman 29]

The problem is that CAD programs don't use all the cores of the processor, it seems quite complicated for this type of programs (10 cores on a recent Imac).

This is a real problem, because the processor is underused, so as the project grows, it gets slower.

[DrPR]

Pascal.

The low price DesignCAD 3D Max that I use for CAD work does use up to five of the six cores (10 tasks) on my i7 desktop (Windows uses the sixth). I was surprised to see it doing this about 10 years ago but I guess it isn't too hard to accomplish with Windows 7 and later.

One thing to keep in mind about multiple core processors is that they generate a lot of heat when all are running 100% duty cycle, and things like rendering a 3D design can push them to the limit. I have a dual fan liquid cooler on the 4.2 GHz processor, and when all cores are running the fans start to roar!

The reason I mention this is that laptops do not have enough cooling capacity to cool the processor when all cores are running hot. For a while Apple issued a warning that it's laptops should not be held in the lap because you might be burned! When they get hot they slow down and the program really drags.

The Intel i5 and i7 processors have an additional embedded small CPU (one of the guys I worked with wrote part of the software for it) that monitors the temperature of all the cores. If the temperature approaches 60C this processor shuts off the clock to the core until it cools. It switches the clock on and off as necessary to control the heat. If that doesn't work the monitor processor drops the voltage in increments to reduce power. If it lowers the voltage too much the core crashes - blue screen of death (BSOD). But the CPU doesn't melt!

This caused a lot of laptops to suddenly stop working with the BSOD, and no one knew why.

On my machine with the liquid cooler the machine would suddenly start roaring like an airplane taking off, but it kept running. It it was strange - it usually happened on Wednesdays at about 10 AM, and only when the machine was running idle! I installed some debugging software and finally figured out that Microsoft scheduled disk defragmentation on the same day each week, at the same time, and it used all six cores at 100% duty cycle. But they only started this when the machine had been idle 30 minutes.

I contacted engineers I knew at Intel and told them what I had discovered, and I called Microsoft and let them know. They changed the way they defragmented disks (short bursts slipped in all the time). And no more BSOD. Because I had thermally stressed my CPU the fellow at Intel offered to send me a new $700 chip!

Laptops are a bad choice for demanding CAD jobs (any program that uses all the CPU cycles for an extended time).

Phil

[Iceman 29]

Thanks for this interesting informations Phil! Good to Know..

I use Fusion 360 from Autocad.

Nice project taylormade!

[taylormade]

I may attempt to restart this project. I just replaced the C drive in my Microsoft Surface Studio 2 with a Samsung 970 EVOPlus with NVMeM.2 technology. It is beyond fast and I find that Blender, which is the CAD program I use, is now instantaneous. When I typed in object perimeters in the past it would often take five or ten seconds for the program to respond. Now it happens instantly when I hit the button. I also keep the project on the C drive and that seems to make the program even more responsive. I'll keep you posted.

[DrPR]

Before I restart my CAD model work I guess I will have to get a solid state drive - if they are compatible with ordinary hard drive SATA interfaces (I haven't investigated hardware compatibility with my workstation). I see a 2 Terabyte version is available. The CAD model is on a 2 Tb disk drive now.

With rotating hard drives I always placed the programs on Drive C and the data on Drives D, E and F. The reason was head movement latency. This is the greatest part of the time lag for reading and writing. If the program and data are on the same drive, as the program loads new modules and accesses data the heads are constantly moving back and forth between different regions on the drive. By using different drives the occasional program module reloads on Drive C are to the same regions on the drive and therefore the head moves a shorter distance and access is faster. The same is true of data access on a different dive. Both accesses can be happening simultaneously.

Since there is nothing moving on the solid state drives head motion delay is not a problem. Program and data can be on the same SSD with no penalty.

Now the question is whether I can replace Drive C on my Windows 7 machine without having Windows throw a fit and stop working?

Phil

[Iceman 29]

Good news! You will be able to continue your project.

@DrPR : Yes, SSD is a real booster. I�ve install one in a very old IBM T30 with Windows XP, It is now very fast, Windows start very quickly.

You can use Clonezilla program to made a real image of the hard disk. The SSD should have an equal or superior capacity of course. I did not have a problem with the Windows license as it was the same computer.

https://clonezilla.org/

I may attempt to restart this project. I just replaced the C drive in my Microsoft Surface Studio 2 with a Samsung 970 EVOPlus with NVMeM.2 technology. It is beyond fast and I find that Blender, which is the CAD program I use, is now instantaneous. When I typed in object perimeters in the past it would often take five or ten seconds for the program to respond. Now it happens instantly when I hit the button. I also keep the project on the C drive and that seems to make the program even more responsive. I'll keep you posted.

[taylormade]

This project may be back on. Just got a new computer with a 4090 Graphics card and an I9 processor and it makes short work of Blender. It now handles my old existing files with ease. The fix I thought would work four years ago didn't. I'll see what happens when I get back into this project.