8:00 to 9:30am - "A killer combination - An interplay of reality capture and reverse engineering"
This was an interesting overview about 3D scanning - reality capture. There are currently essentially 2 methods of creating a 3D model of an existing object. The 1st is to draw it from scratch using a solid modeling program. The 2nd is to scan it, then use software to covert it to a 3D model (i.e. reverse engineer). Point discussed:
Nowdays reality capture and reverse engineering is possible for the following reasons:
A main difference with laser capture as compared to photographic capture is that with photography, you need alot of cameras and to get a good "data set" you need many photos. With laser however, the points taken creates a relatively low file size with quite accurate result.
Lasers however had difficulty with some wavelength.
Angle is also important with laser. At steep angles, the scanner will get incorrect information data. This is the biggest "clean up" that is required
The material of the object is also a factor. Shinny, chromed, glass or reflective surfaces don't scan so well. Photography here is more forgiving.
Use of ultra small scanners to satellites are now available. Scanners are referred to as dynamic acquisition (i.e. google maps).
Different methods of capturing include rulers, hand held lasers, radar capture, gps (now have cm range), MRI and CT scanners, photogrammetry, pattern projection (i.e. xbox), hybrids, arms and probes (very accurate for smaller parts), total stations and tachymeters (this is the most used device in industry today), and lastly laser scanners
When looking at laser scanning, it is generally best to take at least 3 scan, however 4 to 5 are better.
Note that each scanner has a different resolution.
In the diagram opposite, the bottom left is the best result - least "noise" (thin lines top and bottom).
The workflow for scanning is:
________________________________________________________________________________
Alex is a story artist at Pixar Animation Studios. During this session he spoke about working on one of his recent animations "Ratatouille", "Wall-E" and "Cars (21906) - unreleased.
- High resolutions can now be captured
- High quality scanners are getting much cheaper
- More skill base is out there in scanning
- Infinite computing is now available - cloud technology
- Infinite storage available - also cloud
- Fusion sensors (i.e. phones, etc)
The method is to somehow capture the reality (laser or photogrammetry), put through a process to enrich the results.
The challenges nowdays are there are too many tools for the task - no multimodal platform, huge data, close to visualization, non intelligent data (pixels don't have any relationships)
Acquisition is either "ACTIVE" (i.e. send out a laser beam which then bounces back), or "PASSIVE" (i.e. photos). These are then either interpreted on triangulated (for distance), or time delayed (time it takes for a beam to bounce back).
A main difference with laser capture as compared to photographic capture is that with photography, you need alot of cameras and to get a good "data set" you need many photos. With laser however, the points taken creates a relatively low file size with quite accurate result.Lasers however had difficulty with some wavelength.
Angle is also important with laser. At steep angles, the scanner will get incorrect information data. This is the biggest "clean up" that is required
Use of ultra small scanners to satellites are now available. Scanners are referred to as dynamic acquisition (i.e. google maps).
Different methods of capturing include rulers, hand held lasers, radar capture, gps (now have cm range), MRI and CT scanners, photogrammetry, pattern projection (i.e. xbox), hybrids, arms and probes (very accurate for smaller parts), total stations and tachymeters (this is the most used device in industry today), and lastly laser scanners
Note that each scanner has a different resolution.
In the diagram opposite, the bottom left is the best result - least "noise" (thin lines top and bottom).
The workflow for scanning is:
- Scan
- Clean the scan
- Align the scans
- Put in reverse engineering software to create the 3D model
Reverse engineering includes:
- Scan the data
- Create polygons of the data (STL file)
- Fit surfaces between polygons
- Send to CAD to create 3D model
_________________________________________________________________
10:00 to 12:00 - "Photos to physical model"
Now this software is quite amazing and it's currently free to test. All you need is a computer, download and install the software and a camera. You could instead download the software on your iPad and use it to take the photos.
How it works:
1. Install and open the software called 123Catch
2. Take photos of an object from different angles (about 30 photos max).
3. Ensure there are no "shiny" surfaces, nor reflections
4. Upload the photos via the software to turn into a 3D model.
If you want to see a bit more, check out this link:
Below are a few of the photos taken of me to create an animation. An odd 30 photos in total were taken then "joined" in the cloud via 123DCatch
This is , 3D photo booth that was used.
It essentially had 30 cameras mounted at different locations, all pointing to the one point. The advantage of this setup over just using a single camera is:
1. It eliminated the problem of the subject moving in between photos
and
2. The light source was constant
The resultant 3D "surface" is shown in the animation I made below:
This software converts photos to physical models. However the following factors need to be considered:
This is , 3D photo booth that was used.
It essentially had 30 cameras mounted at different locations, all pointing to the one point. The advantage of this setup over just using a single camera is:
1. It eliminated the problem of the subject moving in between photos
and
2. The light source was constant
- Blur during taking photos (subject must remain still)
- Exposure and lighting (needs to be even)
- Framing (about 70% of frame size)
- Coverage - cover as much of 3D object as possible in the 30 photos
- Reflective materials don't work so well
- 3 megapixel resolution max. Higher takes to long to process in the cloud
- You can also 3DCatch a bump surface, then install the surface (merge it) on some other 3D object
For more details on using the software, check the "indestructible" site.
From here you could take the model into software such as 123D Make, which creates "slices" of your 3D model that you could cut from say ply via laser, then assemble (see below)
From here you could take the model into software such as 123D Make, which creates "slices" of your 3D model that you could cut from say ply via laser, then assemble (see below)
________________________________________________________________________________
2:00 to 3:30am - "Story is King" - by Alex Woo (Pixar Animation Studios):
Alex is a story artist at Pixar Animation Studios. During this session he spoke about working on one of his recent animations "Ratatouille", "Wall-E" and "Cars (21906) - unreleased.
During his talk I asked him how did he personally find the transition of working on paper as compared to digitally (AutoDesk Sketchbook Designer). He simply said that as he can use a stylus with pen, the "feel" is similar. He also said the main advantage of digital is that he can make countless iterations without re-drawing so it save a stack of time.
Below are some "examples" of drawings drawn by artists at Pixar:
No comments:
Post a Comment