Text
12.7.12D Image Creation/Editing
Two-dimensional image data is a vital component of any AR application development. Images might be used as backdrop or for other environmental uses, or to create texture maps to enhance the visual quality of 3D geometry models.
Adobe Photoshop is the industry-standard image creation & editing software tool. Photoshop is one component of a larger suite of media development software. Those that are most relevant to AR development include:
Adobe Photoshop
|
|
Adobe Illustrator
|
|
Adobe Animate
|
|
Adobe Audition
|
|
Adobe AfterEffects
|
|
Adobe Premiere
|
|
http://creative.adobe.com/
12.7.23D Model Optimization/Curation
Optimization is the action of making the best or most effective use of a situation or resource.
Curation is the active and on-going management of data through its lifecycle of interest and usefulness. Curation activities enable data discovery and retrieval, maintain quality, add value, and provide for re-use over time.
Both of these activities are essential to the process of developing AR applications. The creation of suitable Environment, Facility and Asset models involves a large part of any development effort. This makes the model optimization process a key enabler to any cost-effective AR effort. In addition, proper curation of these models for future re-use is also critical. Curation is enabled by the data repositories described above in Section 12.4. However, data repositories don’t populate themselves, so curation should be incorporated as a close-out activity in any project plan.
The following sections describe in detail the attributes that may apply to 3D models used in AR applications. These may be used to compare a project’s source model data and the desired/required qualities of models to be used in the finished project. If these attributes happen to coincide, no editing, manipulation or tailoring of the models will be needed. Unfortunately, this is very seldom the case.
Parameter
|
Significance
|
Attributes
|
Mesh data size
|
Primary driver of application’s CPU/GPU hardware performance
|
File size in Mb; Triangular Facet count
|
Number of meshes in model
|
Directly related to number of draw calls that an image generator must execute, also a performance driver
|
Number
|
Component completeness
|
Individual model meshes may or may not correspond to actual components, may need to be combined or sub-divided
|
Partial/Full/Multiple
|
Redundant coincident geometry
|
Often found lurking in CAD models. Causes anomalies in rendered images
|
Redundancies Present/Clean Model
|
Component nomenclature
|
Importance varies widely per Use Case: Crucial for design review applications
|
Per Authoritative Model/Arbitrary Names
|
Model hierarchy
|
Importance varies widely per Use Case: Crucial for design review applications, may need to be altered for kinematics
|
Per Product Data Structure/Per rigged functionality/ Absent
|
Part local axes/orientations
|
Valid part axis definition supports component replacement and articulation
|
As-Modeled/Absent
|
12.7.2.2Model Appearance
Any modifications that enhance the realistic appearance of 3D geometry models must be considered in light of their associated effort. The importance of realism varies widely with different Customers and Use Cases and needs to be clearly communicated and understood.
Parameter
|
Significance
|
Attributes
|
Coloration of components
|
Importance depends on application’s Use Case(s). Application of realistic materials is potentially time-consuming.
|
Realistic/Arbitrary
|
Mesh Smoothing
|
Smoothing enables use of coarser meshes to describe curved surfaces.
|
Almost always desired
|
Image texturing
|
Allows surface features to be ‘painted-on’ instead of ‘modeled-in’. Potential benefits to appearance & performance
|
Worth it/Not worth it
|
Transparency mapping
|
For architectural or interior imagery, or to make internal components visible. May be present in some source CAD models.
|
Present/Absent
|
Normal mapping
|
Allows local variations in surface to be rendered over low-resolution meshes. Potential benefits to appearance & performance
|
Present/Absent
|
Unwrap texturing, Light/shadow baking
|
Gives most-realistic appearance to models Potential benefits to appearance & performance
|
Not found in CAD source models
| 12.7.2.3Geometry Data Size Reduction [Decimation]
Decimate:
1. To destroy or kill a large part of (a group of people or organisms).
2. To inflict great destruction or damage on
3. To reduce markedly in amount
Decimation replaces 3D mesh geometry models that exhibit high polygonal facet counts with 3D mesh geometry models that exhibit much lower facet counts. Properly done, decimation produces models of virtually identical appearance, at data sizes that are smaller by factors ranging from 2 to as high as 100.
A summary table of decimation techniques is listed below.
Technique
|
Restrictions
|
Cost, Human Effort
|
Cost, Processing
|
Results, Polygon Reduction
|
Results, Fidelity
|
Results, Appearance
|
Small part ID & removal [Procedural]
|
Need separate meshes
|
Low
|
Low
|
High
|
Med
|
High
|
Hidden part ID & removal [Manual]
|
Use case must not feature hidden parts
|
High
|
None
|
High
|
High
|
High
|
Hidden part ID & removal [Procedural]
|
Use case must not feature hidden parts, Need appropriate software tools
|
None
|
High
|
High
|
High
|
High
|
Parametric mesh reduction
|
Need access to file’s native CAD software environment
|
Low
|
Low
|
High
|
Med to High
|
High
|
Part feature replacement via image texturing
|
Need suitable bitmap images for texture creation
|
High
|
Low
|
High
|
Med
|
Med to High
|
Part feature replacement via bump/normal mapping
|
Need suitable bitmap images and software for normal map creation
|
High
|
Low
|
High
|
Med to Low
|
Med to High
|
Feature suppression
|
Need access to native file’s software environment. Most effective on instanced complex parts
|
Med
|
Low
|
High
|
Med to High
|
Med to High
|
Global polygon reduction via algorithm
|
May destroy component nomenclature, hierarchy
|
Low
|
High
|
High
|
Med to Low
|
Med to Low
|
Selective edge removal
|
Good edge flow, regular surface tesselation
|
High
|
Low
|
Med to High
|
High
|
Med to High
|
Shrink – wrapping
|
May destroy component nomenclature, hierarchy
|
Low
|
Low
|
High
|
Med to Low
|
Med to Low
|
Part rebuilding/Remodeling
|
Requires high degree of 3D modeling skill
|
High
|
Low
|
High
|
High
|
High
|
Point-Cloud/proxy generation
|
Advanced rendering engines only
|
High
|
Low
|
High
|
High
|
High
|
12.7.2.3.1Software Tools for 3D Model Decimation
The following table identifies software used in the model decimation. It reflects an incomplete awareness of every tool in use across the Boeing enterprise. It does not include enterprise-standard CAD software, which varies according to site, business unit, program and/or customer.
Name
|
Functionality
|
Source
|
Okino PolyTrans
|
Geometry Translation & Reformatting
|
Purchased
|
SAP Visual Enterprise Authoring
|
Geometry Translation & Reformatting
|
Network-Licensed at Enterprise Level
|
Autodesk 3D Studio Max
|
3D Geometry Modeling, Lighting, Animation, Rendering
|
Purchased
|
Autodesk Maya
|
3D Geometry Modeling, Lighting, Animation, Rendering
|
Purchased
|
Rhino
|
3D Geometry Modeling
|
Purchased
|
Blender
|
3D Geometry Modeling, Lighting, Animation, Rendering
|
Freeware
|
Simplygon
|
3D Geometry Re-meshing, Automated Texture-Map Generation
|
Purchased
|
12.7.2.3.2Need for Decimation
In some cases, the source models for a project are limited enough in number and data size that they can be incorporated into an AR application in their original, verbatim CAD state.
For the other 99.9 percent of the time, it will be necessary to reduce the mesh geometry data size of the models. This is particularly true for applications built for self-contained, wearable AR computers such as the Microsoft HoloLens.
For HoloLens, as a general rule of thumb, the entire polygon count should be limited to be in the range of 1-2 million triangular polygonal mesh facets.
12.7.2.3.3Decimation using Automated Processes
Numerous algorithm-based automated processes are available to reduce the data size and polygonal facet count of 3D mesh geometry.
12.7.2.3.4Decimation in Native CAD Environment
This is usually the most effective process for simplifying CAD models. Ideally, the originator of the source CAD model simplifies in the original software environment. This makes it easy to alter accuracy settings and suppress features in complex parts. Also, small parts that are instanced throughout the model (fasteners, for instance) are most easily identified and eliminated.
Familiarity with the model makes the originator the best candidate for native-format simplification. However, most often the model originator is not available and the model must be simplified by someone else.
12.7.2.3.5Decimation in Translator/Reformatter Environments
Certain 3rd-party software tools, such as Okino Polytrans and SAP Visual Enterprise Authoring, incorporate useful and powerful features to aid in the model decimation process.
12.7.2.3.6Decimation in 3D Visualization/Digital Content Creation Environments
Modern 3D modeling software tools [3D Studio Max, Maya, Blender, Rhino, etc.] are most often used to mediate between CAD systems and the Integrated Game Development Environments [Unity, Unreal, CryEngine] that are used to develop and publish AR applications. These tools all offer varying features and capabilities to facilitate model decimation and optimization.
12.7.2.4Hierarchy, Rigging and Kinematics
If discrete components of an asset model in an AR application are being assembled, integrated, articulated or deployed, the file structure of that model must support that functionality. Examples of this would be the landing gear or control surfaces of an aircraft. In order for them to move independently, the following must be true:
Defined as separate objects within the model
Defined by appropriate object hierarchy
Direction(s) of component translation established
Limits of component translation established
Local axes of component rotation established
Limits of component rotation established
Relationships describing motions of connecting components established
The process of tailoring 3D geometry models so as to facilitate kinematic action is called ‘Rigging’ the model.
12.7.3Animation – General Principles
Dynamic visual content is created by sequentially displaying a multitude of images, or ‘Frames’ at a rapid rate.
The number of frames that are displayed per second is referred to as the ‘Frame Rate’. Traditionally, motion-picture film cameras have recorded images at a Frame Rate of 24 frames per second (fps). The broadcast standard for Television was set by the National Television Standards Committee at 30 fps.
These frame rates are important because they correspond approximately to the ability of the human eye to distinguish discrete images from continuously moving images. Thus, frame rates that fall much below 20 fps are no longer perceived as continuously moving.
The creation of these images can occur in advance of their viewing. Motion pictures that incorporate intricate computer-generated imagery are a good example of this. A single frame for such a production could take hours or even days to generate. AR and VR applications rely on imagery being generated as they are being viewed. This is generally referred to as ‘Real-Time’ animation.
Obviously, the processing limitations associated with current computing hardware, operating systems and networks play an important role in limiting what can be displayed in real time.
12.7.4Simple Keyframe Animation
Keyframe animation is accomplished by recording the position, rotation, scale of components or models at certain discrete points (the ‘key’ frames) in a given interval of time.and then generating a multitude of corresponding values for the remainder of the timeline. These are referred to as ‘in-between’ frames.
12.7.5Constraint-Based Animation
Many animation tools provide for the generation of animation data that is based on logical or relational constraints instead of being explicitly defined. For example, a path constraint can be used to cause a geometry model [or a light or a camera] to follow a path defined on a curve or surface.
Although constraints are immensely helpful in accurately animating complex motions, they may not be recognized by the AR application development environment being used. If this is the case, it will be necessary to ‘bake’ the result of a constraint-based animation into explicit frame-by-frame data prior to incorporating it.
12.7.6External Data-Driven Animation
Rather than creating motion or articulation with one or more of the techniques described above, data from external instrumentation, simulation or analysis efforts can be used to drive objects in an AR application.
This approach leverages tools, processes and competencies that have already been developed, validated and accepted as providing ‘True’ data. Incorporation of engineering truth data from validated sources is the most effective way of making visual applications more authoritative, and thus more useful.
To enable this, external data will need to be tailored and formatted to suit the application. Depending on the scope and nature of the data, this may be accomplished manually or programmatically.
12.7.7Incorporation of Non-Geometric Data
AR applications may be developed to incorporate data in the form of text, video, or any type of 2D imagery. This can involve accessing individual files or connecting to continuously-updated databases or other types of streamed data.
Connections to Live Databases
12.7.8Advanced Display Technologies for Virtual/Augmented-Reality Applications
Head-mounted displays – VR displays are now being produced by dozens of companies around the world, at a wide range of price points.
arip-gabs-v1.0-wd01 Working Draft 01 03 November 2016
Non-Standards Track Copyright © OASIS Open 2016. All Rights Reserved. Page of
Share with your friends: |