After completing the animation for Ginetta, we then decided to try and place the car into a game engine and to make it drive able with some degree of suspension.
This was a quite a long process, initially the car had to be put into 3DS max and a bone system had to be constructed around the wheels of the car. The car body then had to be skinned for the suspension to work in the game engine.
After completing this process the model was then put into Unreal Engine, the blueprints of the standard vehicle template are then applied to the model. The bone system was then applied to the blueprint and the pivot points were adjusted. Also a rough collision box was made in Unreal and applied. Once the blueprints were complete the car was then added to a premade level and was drive-able on the terrain.
Due to the models of the campus nearing completion we decided to start texturing the models. This is another fairly long process, it involves unwrapping the model onto a flat surface. After having the flat template this is rendered out and took into Photoshop, the images are then applied to the unwrapped layer.
Additional effects are added by changing the contrast and adding bump and specular maps to get the finished textured model as shown above.
The building above is the Harold Wilson building at the University of Huddersfield
Recently we invited the Mayor of Kirklees to the 3M Buckley Centre to be 3D Scanned. We scanned him with our Artec 3D Spider Scanner. The image above shows the resulting mesh from the scan.
When scanning the mayor we did several scans all around his face, which we could later mesh together to create a more detailed scan. Scanning him took approximately 30 minutes in total.
After scanning him we then put the data into the scanners native scanning software Artec studio, which meshed and aligned the scans together.
An .OBJ was then exported to Z Brush and 3DS Max and the scan data was manipulated and corrected. The whole process took approximately a week with the most time-consuming factor being manipulating the scan data and adding finishing touches.
Recently we have made a lot of progress on the 3D map of Huddersfield project and we are nearing completion of the University which was the first stage of doing the entire town. The render above shows the campus courtyard which is now completed, we have now done 16 of 25 buildings on the campus.
These buildings were created the same was as the initial buildings mentioned on the blog, however, we have now optimised the process and found that it is taking less and less time to complete the campus buildings due to our knowledge of the software becoming more comprehensive and the ability to reuse parts already modelled for previous buildings.
When the University campus model is complete we will export the model as an FBX file and put it into Unreal engine, which will give us the ability to walk around the digitally created campus and eventually the town when it is complete.
We recently decided to paint one of our models to show how well SLS prints can be painted, this rhino was printed approximately 5 months ago in several pieces and was then assembled. As you can see the print has taken the paint very well and has a glossy look, the reason for this is because the excess unsintered powder has been rubbed away over 5 months leaving a smoother finish to the paint.
To paint the model we used an acrylic based spray paint, we didn't use a primer however we would recommend you to use one for a better finish. The main thing to remember when spray painting an SLS printed model to ensure that as much excess powder has been removed, otherwise in a few day the powder rubs away it will also take the paint off with it.
Recently we have been working for the car company Ginetta, we have been producing an animation to showcase their new LMP1 car. The launch event will be at Silverstone racecourse and our animation will be shown at the event. This image above shows a quick frame of the animation, when it is complete we will post the full animation.
Following on from making the 3D Model of the skull we then decided we would try to use this data and input it into the Microsoft Hololens to create an app.
To get the screenshot from the Hololens we had to set up a device portal on the Hololens which tethers the device to a computer which can then record what the user is seeing.
We will shortly be posting an article describing the process of making a Hololens app and the steps required.
We recently did a scan of a colleague at the 3M BIC, this was the results after a quick 10-minute session. This was done on Artec Spider scanner and then edited in Artec studio 11.
We have recently used the scanner on a few high profile clients in West Yorkshire and have created 3D printed statues from the scan data.
The video above shows an artistic representation of the wind tunnel analysis data for an LMP1 data in 3DS Max. The original data was worked on in 3DS Max and was turned into a path, the arrows were then constrained to the path and were set to deform their shape to show how the airflow would be affected when it was in contact with the car.
These two buildings are part of the University of Huddersfield campus, we decided to start the map here because it is local to the centre and we can easily obtain data to model it.
These were also modelled in 3DS Max and were rendered in Mental Ray using a daylight system.
These are the steps in creating the buildings
1. Obtain local 2D plans from the council’s planning website
2. Take or obtain photos of the building, as many references as you can
3. Model the building from the references, and import the 2D plan to work off
4. Render your completed model.
We will not post every renders for every building we have modeled for the town, however, we will provide regular updates at each major stage we reach.
The train station model was relatively easy to reproduce as there were copious amounts of data on the local council site for planning permission throughout the years which contained data for plan, front and side elevations in very good detail. From this the station could be built out, even to the level of accuracy in the iron work which suspends the roof. This model was built in Maya and although the interface is very similar to 3ds max, certain aspects are somewhat more stream-lined for polygonal modelling, such as the edge looping function and multiple extrusions. This programme was chosen for this model because of the ability to use the special duplication command which can duplicate an object in position and transition from the last part, this was very useful when creating the many hundreds of cast iron girders in the roof of the train station, which although exact copies of one another—taper off to one side of the track.
In the same way as explained in previous articles the train station was first referenced in 2D with multiple view ports containing the orthographic projections of plan and sides of the building. with constant referral, the building started to take shape first lifting the floor plan into 3D then cutting back into this with more accurate reference making holes for the windows and doors, this was achieved with the edge loop feature, adding in additional cuts for holes and then appending the resulting gap so that a window frame could be added in place.
Some of the information for this model was not available such as the Corinthian column decoration, this model was an estimation of what is in place at the train station, it is a fair representation of the existing column top. A lot of the period features of this building where produced by using the existing plans and estimating the distance between each part by counting the duplication of aspects in the pediment.
We are currently developing a full 3d model of the town of Huddersfield, including the university campus and other points of interest. We will be documenting the progress throughout the coming weeks and updating this blog with images relating to the creation of each model, for this project we will be primarily focusing on architectural modelling principles, using 2d data from the council to build 3d models from 2d plans. Where possible we will be collecting information to help us better understand the structures we are creating so that they are as accurate as they can be. Image reference and photogrammetry will also play a part in the reconstruction process, further increasing the accuracy of our buildings. We will be using 3ds max and Maya as our primary modelling tools but also delving into z brush and Photoshop where necessary for texturing and fine detailing.
3M Buckley innovation centre
The 3m Buckley innovation centre was the obvious place to start, as this is where we are based! We could therefore get very accurate levels of detail in the form of plans, even down to specific electronic plans for each room. The reason it was important to start here at the Buckley innovation centre with this level of detail, is that it would then serves as a basis for accurately matching other buildings (where data was missing) in terms of ergonomics and heights etc. If a door way or window was of one particular height then we could match this to the standard 3MBIC model which we knew to be exactly correct.
Modelling began from the process of introducing a 2d plane into the scene; this is then textured with a 2d map of the top down floor plan for the building in question. the plane is then frozen in place making sure to right click and set the plane as visible when frozen ( by default this option is set to be shown as grey when frozen ) this allows the modeller to build on top of the map without editing the original position of the underlying map. A simple polygonal plan is drawn out in position and the walls are traced making sure to use the correct edge looping techniques. When this is complete the buildings walls are raised to the first floor position. For the 3m Buckley innovation centre we also had the side profile data so the height of the first floor could be accurately re produced. A new edge loop was then added in the mid-section to allow for door ways to have their tops filled leaving the gap for the door bellow.
This process is repeated for the first and second floors with their different aspects being taken into consideration by the modeller, all three floors are then stitched together and detailing can begin doors windows floors and other aspects which add to the level of realism of the building. Below is a render of the finished building.
The next stage in our project will be to create a virtual avatar of the Vice-Chancellor who will (virtually) help national and international students navigate their campus in 3D. This will be a fully accurate representation of the University of Huddersfield and will greatly benefit those on OU courses and overseas students. We will shortly be showing our developments on the virtual campus.
Making the chancellor an avatar 24th February 2017
The image above shows the final 3D scanned avatar of the Vice Chancellor. Below is an explanation of the process of how we made the avatar from scan data.
To scan the Vice- Chancellor we first set up all the equipment and had the subject sit in an open area so we could walk around him and get a complete model of his face. The process took about 20 minutes. During which time the participant must maintain their position with as little movement as possible. As a side note, the scanner does not require an object to stay still that has an unchanging form, however, with an organic form such as a human where muscles and skin can flex and deform, this would create a problem for the camera to identify unifying points and therefore break the model.
The scanner detects skin very well,; however, it can struggle when confronted with hair, which presumably is due to the process of triangulation. As previously mentioned, there are so many angles in hair that the camera has difficulty in locating the data. One way to combat this issue is to recalibrate the device so that the cameras have a narrower band of overlap between each other. Doing this allowed us to capture most of the subject's hair and the rest was then re-sculpted in software such as “Zbrush”.
After scanning the VC we then “fused” the scan data together, creating a 3D model which was then imported into Z-brush. On Z brush we corrected some of the defects where the scans had overlapped and had differential surfacing, errors such as outlying material and removed so that the model was solid and refined. This was the most time-consuming process and it was difficult to get a balance between the file size of the model and the detail required to show a likeness of the original subject.
The model was then imported into 3DS Max. We used a standard biped model and added some clothing to attach to the face model, which gave us a complete model of the Vice Chancellor. The model was then rigged using a free software called Mixamo developed by Autodesk which gave us the skeletal structure of the biped, allowing us to move the character and animate his motion.
This project was a lot of fun and taught us how to develop scan data for multiple uses either for game engines or for 3D printing. The main hurdle to overcome during this process was the poly count obtained from scan data, which was exceedingly high and comprised of triangles, as opposed to four sided polygons. For a 3D model to be used in a game engine the mesh of the model needs to be decimated, and re-topologised. When this was done a texture could then be ripped from the original scan and re-applied to the subdivided mesh which allowed a greater level of detail to be shown.
