[dwb]

Fascinating.

Your sketch reminds me of a goods depot I passed by when cycling between Bregenz and Lindau along the banks of Lake Konstanz during the summer.

David

 

[TheAndrian]

Hi David,

The more I look at photos of the older classic buildings found along railway lines in general (Austria and its neighbouring countries) it's amazing how one notices a generic design. The goods shed kit (Dacice) which I propose to alter would also be literally identical to the goods shed found at Hohenberg station further up the line towards Traisen. Looking at an old photo postcard of St. Aegyd station the goods shed was half the length and must have been extended at a later date, hence a possible reason why the existing structure's roof tiling differs in one half. However the one at Hohenberg has one roof tile type.

I really need to take a vacation at some point and visit this line and take my own detailed photos for reference.

Cheers

 

[TheAndrian]

Hi,

Another update again. I've been exploring SketchUp quite heavily of late; in particular the terrain tools and various extensions. Even though this project's modules will be fairly flat, I'd wanted to see what can be done in SketchUp to help generate prototypical terrain based on real locations and providing a means of creating "contoured" terrain for the modeller to work from.

The caveat is that the GeoLocation feature is only available in the Pro version; earlier free versions of SketchUp did have this feature under Google. Also the other problem is the accuracy of the terrain data which I will touch on later.

I'll present a brief sequence of photos of how one can generate layout elements used with two free extensions by Fredo6 (available via SketchUcation). A more detailed approach would have to be put in a separate topic.

Firstly a flat GeoLocation of the area was grabbed and formed into a terrain sheet;

Using Fredo6's JointPushPull Extension in the vertical direction I created a "skirt" for the terrain, this is required in order to use the next extension;

Next using Fredo6's Slicer extension we have two options for creating a workable model maker's terrain.

First is two slice the terrain into horizontal sheets for say Styrofoam sheets (I've set here a 3mm styrofoam sheet in real world scale from N i.e. 3*160=480mm). I also set an option to flatten the sheets with reference numbers.

The other method is to generate transverse members again slicing but with gaps to form the terrain (dimensional lumber of 25mm real world scale is 25*160=4000mm with gaps of 200mm real world scale 200*160=32,000mm).

From the flattened pieces generated from the above two methods one can export these as png or dxf files which can then be scaled down for printing (3d, laser cutting or ink print), or even with some file conversion to other CAD packages.

As can be seen in the transverse pieces two corners don't extend fully which may need editing.

To illustrate the problem of geographic data accuracy I dropped my 3D model of the warehouse on to the terrain sheet, and it can be seen closely that the terrain is a bit bumpy;

Higher resolution terrain is obtainable but at a price and can be imported into SketchUp via suitable Ruby scripts. I have found a place to purchase XYZ files (Cloud Point) for separate features i.e. terrain, buildings and roads in Austria; an option for very accurate modelling.

A final piece of advice if anyone does intend to use SketchUp for model making don't use model scale dimensions (SketchUp is not good at small values) work in real world scales and then for a final output i.e. 3D printing etc scale it down to your model scale.

Hope this was of interest.

Cheers

 

[TheAndrian]

Hi again,

I'd mentioned in my previous post about the use of Point Cloud files for generating terrain in SketchUp and I was curious to see how far one can push SketchUp with suitable extensions to create a more detailed/accurate terrain. I downloaded a sample Point Cloud file covering over a square kilometre. I had to firstly import the data via a suitable script extension to generate guide points into SketchUp. Point Cloud files are basically a list of points in a grid (in this case 10m x10m) with an elevation value and these files can get pretty big since since they are CSV/ASCII files. As I understand SketchUp can just about handle 1 million points before its starts to slow down and misbehave.

Below shows the sample file imported in as guide points:

These were then converted into a triangulated terrain mesh, this took about 30 minutes to generate using Fredo6's Toposhaper extension:

To further process the above result with JointPushPull and Slicer, I had to get rid of the skirt generated by Toposhaper and delete all the artifact guide points.

The final result of the terrain sliced is shown below (again with scaled up 3mm slices representing styrofoam sheets). I deliberately sliced the model this way to see if my computer and the extensions could handle such a vast amount of data and surprisingly it worked but I had to leave the computer on over night to generate this:

Well I don't know where this data model is based on (definitely a river is shown), but the detail and accuracy is considerably higher than using SketchUp's native terrain data.

Cheers

 

[dwb]

Fascinating !

David

 

[TheAndrian]

Hi David,

For your information, I just found out that this piece of terrain is actually in Villach. The river is the Drava, the main land mass by the river is Villach Hauptbahnhof .... I had a suspicion that the nearest corner in the images posted must be a bridge of some sort and it is in fact the Steinwendestrasse crossing the Drava along with the railway line. Here's a link to Google maps of the location for convenience.

Cheers

 

[TheAndrian]

Hi,

It's been over a year since I last posted on this project. A lot has been going on in the background in particular with the research of the the buildings that need to be made. I spent quite a considerable amount of time searching various architectural forums (German/Austrian) trying to understand the historical development of the Austrohungarian railway building architecture as well even going down the rabbit hole of the humble development/history of the brick industry in Austria (a surprisingly fascinating subject) which I will discuss at a later date regarding the model making of the loco shed and the staff/administration building at St. Aegyd am Neuwalde.

In this update it is more a list of books I have recently purchased which has helped in my research due to the recommended books in the forums I visited.

Recently Railway-Media-Group published a book in cooperation with Roth-Technik Austria about the steel industry history at St. Aegyd am Neuwalde titled "St. Aegyd am Neuwalde - Eisenbahnen, Rollbahnen und die Neue Traisentalbahn". A very interesting book with plenty of detailed drawings of the rollbahnen at the various steel production facilities in the Traisen valley. The book's ISBN is 978-3-902894-66-3 and the author is Manfred Hohn.

Probably the most important author on railway building architecture in and around Austria is the Hungarian architect Dr. Mihály Kubinszky; he had plenty of books published and they are worth their weight in gold. The information in his books is mind boggling and anyone interested in modelling Austrian and Hungarian station buildings should pick these books up (these days used books are still available).

I was lucky enough to pick up a brand new copy of "Bahnhöfe in Alt-Ã-sterreich" published in 2009 which details the station buildings surrounding Austria (Hungary, Italy etc), a good reference for those interested and the book has128 sides. ISBN 978-3-85416-197-4.

The real gem by Mihály is "Bahnhöfe in Ã-sterreich" published in 1986 and comes in at a hefty 272 sides, ISBN 3-85416-077-1. This is a very comprehensive book, doesn't have drawings of every station building in Austria but impressive. This book is highly recommended, the drawings are very detailed along with plenty of photographs.

I've just ordered another book by Mihály titled "Architektur an der Semmeringbahn" ISBN 978-3854161561 which I am sure will be just as detailed but specific to the Semmeringbahn.

Hope this of interest for those into Austrian railways.

Cheers

 

[TheAndrian]

Hi again,

Having found books and various scans describing the typical buildings used by the Austro-Hungarian (private and national railway) from the late 1890s and onwards, I am in a position to prepare CAD drawings of the station buildings.

My first main focus is the water station which is unusual in that the main tower holding the 56 cubic metre tank has two wing buildings as well having exposed cyclopean stonework.

Here follows images of the drawings so far done in CAD, these are not by any means complete but a generalized framework before the details like stonework, windows, doors etc are integrated into the drawings. Once satisfied with the detailed drawings the final process of compromises will be made chiefly for the timber work which will be laser cut and engraved using plywood or basswood.

With respect to the lower part of the structures stonework, I'd be playing around with ideas of how to reproduce this. I decided that 3d printing the parts for this should be done using a HP Multi Jet Fusion printer with their PA12 material - the texture is very good for a rough surface i.e. stonework. I trialed a 3d print of the cornice to see how it would appear and see if the printer was capable of capturing the detail - see below.

Cheers

View of top:

View of bottom:

I am very pleased with effect that a rough "hewn" stone effect can be created using this material. The cornice is only 2.5mm high. On the 4th photo you can see the telltale sign of the powdery effect of this material especially the fine "hairs" on the edges. This was done in raw material print without any dye added i.e. black.

I will have another print made of the cornice made with more added detail. If I am happy with that then I will proceed with the main part with the cyclopean stonework.

 

[Richard Johnson]

*** A fascinating thread. I look forward to each post

Richard

 

[TheAndrian]

Hi again,

Another update on this project. I've recently got distracted by another building for this module and that is the administration/staff building, which I believe is now rented out as apartments by the OeBB. Whilst messing about with SketchUp a while back I made a rough model of the walls of this building. See below:



What is interesting about the walls is the stucco; the borders have been smoothed and the rest is roughed up. To simulate stucco is especially difficult on smaller scales; I've seen various attempts to simulate this surface using textured paints, sand paper etc and in my personal opinion in the majority of cases they don't look convincing. A while back I saw that some model makers used Rust-Oleum Stone effect paints but the texture from that line is too big for N scale even maybe for OO/HO. Another product from Rust-Oleum is their natural effects textured line of paints which have a very subtle effect; this I noticed when experimenting to give a texture for my test piece of track to simulate concrete sleepers. The photo below shows the slight texture after two light coats with the Limestone paint:



I made out of styrene sheet 5 30mm square test pieces for an experiment to simulate the surface effect I am hoping to get for the stucco walls, firstly by spraying one light coat of the Limestone paint on the whole surface area of each piece and then subsequently masked off the edges on each piece. The exposed central areas after masking were subjected to another three layers of the paint in the hope of building up the texture to be more prominent and to give a distinct relief from the flatter surrounding. I sprayed the paint at various distances and angles to get the desired effect. See the following five photos:





The first sample as can be seen was flooded and hence not the effect I was looking for but an interesting surface effect. I think the 2nd piece came out the best, the rest especially No.5 are probably a little too heavy on the texture. What was really successful was the relief difference achieved. This has given me some encouragement to create a masked stencil of some sort for the more complex border around the windows when I eventually get round to building this structure.

I throw in a another photo which shows a closer view of sample 5 showing the difference between the border and the central area:



Two observations I noticed when using this paint; firstly masking tape doesn't rip up the paint like some other brands of paint which is good and secondly I suggest that each time one makes subsequent coats to change or clean the protective surface when spraying this paint because it generates a lot of over spray which can subsequently be like fine dust potentially creating undesired effects on the target surface.

I like this spray paint line so much I have been hoarding them just in case they may one day disappear. The ones I think which are useful are Limestone, Concrete and Terraccota. The terracotta paint would be useful for roof tiles. One of the risks with such fine textures is losing the detail when further painted so using an airbrush with fine thin coats, very thin washes and glazes/filters would maintain the surface texture without obliterating it.

I will be using the 4 better test pieces for the next step in nailing down a colour palette for this surface, which I will go into a bit more detail in another post.

Cheers

P.S. If you don't see the images please let me know, because I have changed my free image hosting provider after getting fed up with Photobucket's recent tactics.

 

[dwb]

Very interesting. Thanks for posting those images

David

 

[TheAndrian]

Hello everyone and Happy New Year!

In 2 posts earlier on this thread, you'd have seen I'd dabbled with SketchUp to model the terrain for the module(s), creating possible wood cross members and styrofoam outlines of the elevations with various extensions on SkecthUps native terrain data and a demo point cloud file. I am abandoning the SketchUp route since to get any meaningful file output you have to pay for the full version software.

The new approach I'm taking is in using two pieces of software MeshLab and FreeCAD which are free to download. In the last few days I've managed to obtain a point cloud file for the terrain for the module(s) from BEV in Austria which I had registered with over 2 years ago and just never got round to ordering from them due to the awkward website arrangement. I'd been using Google Chrome with their website and it just didn't play nice so to get the full benefit of their website I used Microsoft Edge.

I ordered a point cloud in XYZ file format for an amazingly good price, in my honest opinion, for just under 6 euros for the data alone; the elevation grid is 5m x 5m (interpolated from 10m x 10m) covering an area of 1720m by 1120m. Now that is an excessively large area for just the module(s) I'm working on however there is a caveat in selecting areas, polygonal bounding box selections cost more even though easier to work with to fit the module(s) than a North/South by East/West rectangular grid; furthermore if I wanted to extend the base module(s) with an extension module(s) based on the prototype, I have the point cloud data available ready at hand. Maybe in the future I could be more selective for other projects using longitude and latitude co-ordinates.

The area selected is below:

Here follows images of the point cloud and the generated mesh in MeshLab, the generated mesh will eventually be exported to FreeCAD:

As a comparison to my earlier experiments with SketchUp, I made a rough highlight of the area in MeshLab showing the area I'd experimented with in SketchUp:

Having looked closer at the mesh from the purchased point cloud data it appears a lot smoother and more accurate than the native terrain mesh I'd generated in SketchUp. In my earlier posts I showed the possible import of point clouds (Villach 10m x 10m demo point cloud) into SketchUp but as already mentioned to get a useful output you have to pay.

My next step is to cut up the mesh into a rectangular section so that I can create a fascia for the module(s) sides in FreeCAD and then export for laser cutting. The styrofoam approach for the elevations will be a bit long winded in FreeCAD compared to the extensions I used in SketchUp but doable.

Cheers

 

[PaulTrabant]

Has the layout developed ?

 

[TheAndrian]

Hi Paul,

Thanks for inquiring. It is in progress still. I am waiting for my 3d printer to arrive in January (Anycubic Ultra) to do the buildings, given up on 3rd party 3d print vendors due to quality control and shipping costs. Also there has been a significant development in Freecad with a powerful work bench addon which is still in alpha development which has helped in terrain modelling. I will post maybe later today some screenshots of the terrain modelling.

Again thanks for inquiring.

Cheers

 

[PaulTrabant]

One of mu layouts will be narrow gauge OBB and the buildings really give the country away before you see a train. Looking forward to seeing your progress

 

[TheAndrian]

Hi again, a bit more progress in designing the modules and modelling the terrain as mentioned in my previous post. I managed to import the point cloud into Freecad which has allowed me to generate a NURBS surface of the terrain. From that I have been able to create the module fascias. The module fascias will be two pieces of plywood joined together (each 6mm thick) for a total thickness of 12mm, the inner fascia piece will also be offset 6mm from the upper edge of the outer fascia's edge. The purpose of having the inner fascia is for it's top edge to act as an "anchor" for the terrain foundation i.e. a fine aluminium mesh placed on top of stepped counters representing the change in elevation, subsequently the terrain will be built up to the same level of the outer fascia's top edge. There will be some sanding involved when the profiles of the fascias are cut to represent the change in slope going towards the centre of the modules since the laser cuts will be 2D and not 3D.

Here are some images of the block of terrain with the NURBS surfaces created through to creating the fascias for the three 1m long modules. Note I decided to increase the width of the modules from the Fremo standard of 400mm to 500mm.

The basic block:

Modules fascias generated from block:

Module fascias closer view:

Modules with top surface of terrain flush with outer edge:

Modules with offset surface (foundation) flush with inner edge:

Using the offset surface I was able to slice it in incremental heights, in this case I used 0.4mm, which is overkill and would use a lot of thin plywood sheets unless a reduced area is used and carefully layered in a stair case fashion; in some cases whole sheets would still be required. The more likely option is to user thicker sheets.

Here is an image of the sliced terrain in 0.4mm steps of the offset surface:

A closer view showing the laminated layers and a 6mm base to build upon:

Using some geographic data obtained for nearby buildings I was able to manually create my own point cloud files representing them and placed on top of the terrain - these are approximate but good enough. There are some buildings which overlap or outside the boundaries of the modules which will have to be lost/ignored.

A view of the overall terrain with buildings:

A view with the key buildings; going towards the left from the furthest right 1) Goods shed 2) Passenger Station grounds 3) Water Station 4) Admin/Staff Building 5) Locomotive Shed 6) two residential buildings

I will have to chop the sliced terrain to slot into the inner boundary of the modules and at some point to export the track roadbed created in 3rdPlanIt to "carve" it's way through the terrain slices which will alter the contoured slices into smaller pieces/surfaces. I will also have to create the finger joints for each fascia for assembly.

Hope that was of interest.

Cheers

 

[dwb]

It will be interesting to see how the CAD becomes in reality .

David

 

[TheAndrian]

It will be interesting to see how the CAD becomes in reality .

David

Hi David,

Yes it is coming soon to reality; this project has been a long process but worthwhile, since the methodology developed (through trial and error from a CAD point of view) can be easily reiterated for any future projects in a smaller time scale.

Here is a something I designed and developed (nothing really special to be honest, just a bit of fun) for this project; a 4 way DCC bus connector block which can handle 24 AWG to 12 AWG wire using some fancy cage clamps. For the PCB to be mounted onto a baseboard I have used hex drive D type threaded inserts which connect to the PCB spacers. I've got more designs of this in CAD where the connector blocks are configured as 6 way and 8 way as well as variations using larger connector blocks for even heavier gauge wires.

Cheers

 

[TheAndrian]

Hi again everyone,

Well things have been slow for a number of reasons, but I have an update today. I have been lately transferring some of my CAD files into a suitable format for laser cutting. I had one stage considered going to have them cut by an external service, but I bit the bullet earlier in the year to purchase a Longer Ray5 5W laser diode cutter. Recently I upgraded the laser diode module to the 10W version.

I thought I'd start with tinkering around in cutting a typical goods shed out and fairly happy with what I have assembled so far. I will probably change a few things since assembling is a little bit involved by my own fault in being too intricate and accurate so compromises are required.

You may see some similarity to the goods shed from Igra which is shown in my earlier posts. However these double length goods sheds even though 16m long and 8m wide had their doors and windows in variable locations; sometimes the main shed doors would be offset in one or the other direction. This model so far does not have the doors evenly spaced from the centre but are offset.

Materials wise I used two thicknesses of plywood to cut, 1mm for the structural timber and 0.4mm for the cladding and decking.

Pictures of 0.4mm cutting in removed with cut pieces and part assembly:

Picture of 1.0mm cutting in action:

Assembled model so far:

Never thought I'd have so much fun using a laser cutter. I will be adding some progress photos in due course just waiting for a Neje 3 Max (with extensions) to cut the module profiles.

Cheers

 

[TheAndrian]

Hi again,

I seem to be getting too much hardware these days, and it is time to set up a proper workshop for my various projects; with the recent arrival of my Neje 3 Max laser engraver/cutter with extensions which when set up will occupy an area roughly about 1200mm x 850mm, that doesn't include the waste board which will be marginally bigger.

Furthermore I had earlier this year received the Anycubic Photon Ultra DLP resin 3D printer after backing it on Kickstarter last October; I did some demo prints and since then did not use it any further until recently. To achieve what I wanted to get out of the 3D printer for this project I had to use Blender for getting textured meshes from FreeCAD. I spent many evenings to understand how to get displacement maps to work properly; this resulted in doing some upgrades to my modest PC (Acer Aspire XC-885), involving an upgrade of the RAM from 8Gb to 32Gb as well as replacing the hard disk drive (HDD) to a solid state drive (SSD). Even with these simple upgrades it can be a bit of a struggle since I don't have a dedicated GPU card inside the PC which unfortunately the CPU's internal graphics processor has to take the grunt of the rendering in Blender.

I'd spent many hours looking for suitable displacement maps which were high quality and seamless from various sources and did find some which were ok but not ideal to represent the stone walls of the goods shed. So doing my first own prints of my 3D models were to say the least interesting and time consuming; it took 3 attempts to get a successful print after tinkering with some parameters; I still need to dial in properly which you will see why down below.

So first off the first failed print, the main body got stuck to the FEP and the build plate just had the supports:

The 2nd failed print, here I didn't use supports and built straight from the plate with no supports or raft apart from baking in the first few layers. It didn't hold and and deformed with some visible layers of delamination and sticking to the FEP:

The 3rd attempt with a raft, a success I suppose, this time having changed the base exposure for baking in the first 5 layers and adjusting the normal exposure from the base 2.0s to 2.5s:

The caveat here is that the 2nd print imho was sharper compared to the successful print; the 3rd print exhibited some odd kind of debris and horizontal banding which I suspect maybe due to over exposure or something completely different. I also noted that 3rd print had at one point layer shifted for some reason too. You can see the differences between the two prints here below:

Another 2 photos with my earlier laser cut structure on top of the 3d printed wall:

Dimensionally the Photon Ultra is pretty accurate on the X-Y axis (80um) which is not surprising since it is a projection of light. The z-axis is the dodgy bit since it is mechanical and I am finding that it is short by about 0.3mm; this may seem small and not to complain about but when you want to say lay pieces of 1mm depth into slots of a 3d print which are 0.6mm or 0.7mm deep instead things are not going to be level. The only solution I suppose is to scale the vertical height of the 3D model by a percentage to compensate for the error. For those interested in the slicing layer size I used 20um, antialiasing 16 and grey level 2.

Anycubic have released the Photon Ultra D2 recently which has a slightly larger build plate and an x-y resolution of 50um with better anti aliasing and looks impressive. So far I am pretty happy with the results achieved on this version so far, I just not to do some more dialling in. It amazes me that so many people complain about this printer that the voxels is evident compared to other printers; for highly textured surfaces it is literally invisible and a blast of self levelling primer will just get rid of it.

I'll keep updating as I progress, this is quite time consuming work and these prints here took me nearly over 12 hours of work involving set up cleaning etc.

Cheers