PhotoScan

PhotoScan
Developer(s) Agisoft LLC
Initial release 2010 (2010)
Stable release 1.2.4 / 18 March 2016
Operating system Microsoft Windows
Linux
MacOS
Type 3D computer graphics software
License Proprietary
Website www.agisoft.com

Agisoft PhotoScan[1][2] (commonly known as PhotoScan) is a professional tool for a photogrammetry pipeline. It is a stand-alone software product that performs photogrammetric processing of digital images and generates 3D spatial data to be used in GIS applications, cultural heritage documentation, and visual effects production as well as for indirect measurements of objects of various scales. The software is available in Standard and Pro versions, the standard version is sufficient for interactive media tasks, while the Pro version is designed for authoring GIS content. The software is developed by Agisoft LLC located in St. Petersburg in Russia.

It is widely used by archaeologists.[3][4][5][6] There are a lot of UAV companies using it.[7][8][9]

Overview

Agisoft PhotoScan is an off-the-shelf photogrammetric kit that allows for photography (both metric and non-metric, aerial and close range) to be

Agisoft PhotoScan is capable of processing up to tens of thousands of photos yielding result products characterized by high degree of accuracy in both the horizontal and vertical dimensions.

The software package has a linear, project-based workflow, that is intuitive and can be easily mastered.

Agisoft PhotoScan Standard edition is a basic program realization that, however, could be used to solve various tasks: reconstruction, modeling, digitalization of objects and scenes as well as in video games creation, etc.

Main Features:

Agisoft PhotoScan Professional edition includes all Standard edition functionality, that is enhanced with the following features:

Source Data: Digital aerial images

Agisoft PhotoScan processes images shot with both metric and non-metric cameras. In the case of digital non-metric cameras, unknown variables such as the focal length and geometric stability of the detector array are modeled mathematically in the software.

Step 1: Image Management

The raw data can be added to PhotoScan in the form of digital imagery in one of the following formats: TIFF, JPEG, BMP, PNG, PPM and JPEG MPO. Generally image resolutions can be whatever one's camera quality allow, however, one should keep in mind that the resolution of the source data naturally affects the accuracy of the results. So it would be reasonable to opt for 12 MPix imagery at least. To cope with huge data sets (>10,000 photos) PhotoScan suggests "chunk" conception, which enables to process the data in parts merging the results later.

At this stage of the workflow, one can assign a camera model to each image. This consists of the focal length in x and y dimensions measured in pixels, principal point coordinates, radial and tangential distortion of the lens and skew transformation coefficients. In case the imagery was shot with "standard" optics, that is with 50 mm focal length (35 mm equivalent), there is no need to calibrate the camera beforehand: PhotoScan will model the system applying Brown model and estimate calibration coefficients itself. This operation is likely to fail if the source data was captured with a "fish eye" or ultra-wide angle lenses. Consequently, if this is the case, one should enter the calibration data to the program to achieve good reconstruction results.

So this step is aimed to input the data into the software and to determine the structure of the project, thus having everything prepared for the processing procedure to start. PhotoScan does not require to perform interior orientation of the images manually, which is a significant time-saving feature in respect to the potential size of the projects. Concerning exterior orientations, these six parameters (x, y, z real world coordinates of the camera projection center and the roll, pitch and yaw angles of the camera) can be assigned to each image by the user at this very point of the workflow. The easiest way is to load the data from the EXIF metadata if it is included in the image file, otherwise, the data can be loaded as a file in character separated format of simple structure or entered manually. In case the airborne GPS/IMU data is not available, these orientation parameters will be determined by the program using ground control points (GCPs), tie points and aerial triangulation at the next step of the workflow.

Step 2: Ground Control. Aerial Triangulation

PhotoScan is capable of automatic tie points generation, based on interest points detection and matching. It means that generally triangulation can be performed without manual intervention at all. However, both to strengthen overall geometric robustness of the image block and to allow for more precise estimation of the exterior orientation parameters, which improves georeferensing accuracy of the final results, it is useful to add GCPs into the project.

To describe it as the most basic steps possible, one needs to distribute several (at least 9-10) targets evenly over the area to be shot and reconstructed, so that they are easily seen on the future images, and measure their positions. Alternatively, one can use some natural targets with known/measured positions. Then the targets need to be determined on the images by the user via PhotoScan interface. This is easily done within the marker placement concept, that includes time-saving guided approach option.

PhotoScan can process GCPs coordinates recorded in various coordinates systems, both geographical and projected ones, in particular supporting most of the systems from EPSG registry catalog. Moreover, coordinates can be measured in one coordinate system and then, just in case, converted with PhotoScan to a different one.

After GCPs have been added to the project, it is the time to run aerial triangulation procedure. (Within PhotoScan's names field the process is called Photo Alignment finished with maker based Optimization procedure). When tie points have already been determined, software takes the tie point information and exterior orientation data (if presented) and triangulates it to get the best fit of the data within all images in the block ("chunk", according to PhotoScan concept). In case project lacks GCPs information, this could be the end of the second step. Otherwise, one should run optimization procedure to improve the results. It relies on the accuracy of the input points, both in how well they were initially surveyed ("camera accuracy", "marker accuracy" optimization parameters) and in how well they were placed in the image ("reprojection accuracy" optimization parameter). All the parameters mentioned above could be tuned by the user according to the input data and project conditions.

The resulting output of this step, to say nothing of a 3D-model in Point Cloud quality available for making first approximation evaluation of the quality of the input data, is the numerical estimate of how well the all the data fits each other (estimated exterior orientation data to be compared with the initial one; GCPs estimated vs measured data error). This information can be used to check whether the result fulfills one's project requirements for the spatial and georeferencing accuracy.

Step 3: Dense Surface Reconstruction

DEM generation, one of the key tasks for any photogrammetry software, takes only minutes in PhotoSсan thanks to the sophisticated image matching and autocorrelation algorithms. The possibility of user intervention allows to set desirable effective resolution, automatically crop invalid regions as well as to define the region of the reconstructed area to be included in the resulting DEM. PhotoScan supports DEM export as GeoTIFF elevation data, Arc/Inpho ASCII grid, Band interleaved file format, XYZ file format that guarantees compatibility with major tools to be used to work with geoinformation products later.

Moreover, there is an option to export from PhotoScan a 3D model in popular formats like WaveFront OBJ, 3DS file format, VRML, COLLADA, Stanford PLY, Autodesk DXF, U3D, Adobe PDF.

Step 4: Orthophoto Generation

Having created a 3D model of the area with PhotoScan, one can move on with generation of planimetrically correct image map. The software uses the TIN surface to correct for terrain displacement and exterior orientations for georeferencing to create a very accurate orthophoto. The orthophoto generated is fully reliable and can be either used as a final-end product or as a basis for professional quality GIS data for specific purposes creation in external tools.

At this step the software interface allows user to select Blending mode and Projection plane as well as to opt for Fill holes feature to generate complete orthorectified map. Along with most commonly used GeoTIFF (or TIFF + World File) formats, PhotoScan support orthophoto export in JPEG and PNG formats. For the users who would like to immediately see the processing results in global context, PhotoScan supports Google KML mosaic export for the orthophoto to be smoothly opened with Google Earth/Maps applications.

Step 5: Measurements

Being a photogrammetric suite, PhotoScan presents functionality to carry out area and volume measurements. Though the features are quite simple ones at the current version, they enable to solve most commonly faced tasks like mine and construction pits volume measurements.

Step 6: Processing Report Production

As a post-analyzing/post-discussion facilitating tool PhotoScan support report composition feature. Agisoft PhotoScan Processing Report presents most essential figures along with visual information to be used as results estimates and statistic data in checking requirements fulfillment, planning of project improvements and further work strategy development.

Customized automation potential: Python API

PhotoScan allows for fully automated processing through Batch Process function. Even greater potential to organize typical projects processing in a manner of "one-button solution" delivers Python scripting feature, that provides opportunity to automate your personal processing workflow to achieve most accurate results without user intervention.

Dealing with artifacts: Masks

PhotoScan enables to exclude artifacts from processing due to the masking feature. Applying masks to any odd objects on the images allows to force the tie-point searching and the rest processing algorithms to, so to say, concentrate their power on the reconstruction of targeted areas that results in adding up to the accuracy of the final model.

Masks can also be useful in covering monotonous background while reconstructing some object. It allows to avoid the background noise in the reconstructed model that could have badly influenced the quality of the targeted object geometry.

Fields of Application

PhotoScan is capable of modeling a fingerprint and a huge stockpile. The range of its capabilities is almost unlimited. Being profounded with georeferenced orthophoto and DEM generation features, PhotoScan Pro extends its applications to mapping and ground monitoring, fitting perfectly well in pre-GIS workflow.

Below is the list of most frequent applications of PhotoScan for you to form the general understanding of the rich potential of the program. The list is organized according to the type of input imagery.

Aerial photography

Land & Topo Surveys

Today PhotoScan is quite often seen as a perfect tool for aerial imagery processing. The program is constantly developing its functionality and quality according to the tasks set by rapidly emerging UAV industry. PhotoScan has proved to be an exceptional post-possessing tool capable to generate extra dense point clouds, high-resolution orthophotos and exceptionally accurate DEMs, not to mention precise polygonal models of large scale objects. It is indispensable part of GIS workflow stating with a UAV system.

The list of possible applications in relation with land survey could be exemplified with the following:

Mining

PhotoScan is an easy to use tool for simple volume and area measurements. Providing that at least two ground control points are presented to set the reference distance of the model, the volume of a stockpile or construction pit could be calculated automatically with two mouse clicks, to say nothing of the line distances measurements.

PhotoScan could be employed by professionals in:

Close-range photography

Animation & Cinematography

PhotoScan has a great potential as a tool for creating 3D models for animation and multimedia applications and can be used in animation, film and video, and web site design. PhotoScan allows to export automatically generated detailed 3D models with photo-textures that could be later edited in external 3D design tools. Face and body capture are also popular among PhotoScan enthusiasts, who prove that PhotoScan potential goes beyond ones imagination.

Medicine & Biology

PhotoScan finds it way in medicine and biology due to precise model reconstruction capability accompanied with easy to use measurement tools. Morphological measurements are important in diseases tracking (scoliosis, foot shape), prosthetics, plastic surgery, etc.

Business

Prototyping and marketing materials creation is one more niche for 3D model tool like PhotoScan. The program proves to be valuable and cost-effective tool for marketing people providing catchy promoting pieces at a speed an price incomparable with traditional approach. Moreover, PhotoScan finds it application in specific businesses like laser engraving owing to a variety of output results.

Forensics

Photogrammetry meets perfectly well the need of those working on accidents reconstructions an forensics. PhotoScan could be employed in vehicle crash measurements as well as in crime/accident scene reconstruction.

Mix

Architecture & Preservation

PhotoScan could be used to obtain orthorectified photos of facades or reconstructions of the whole buildings that could later be printed and used as illustrative models of large-scale cultural heritage within museum environment.

In this respect, PhotoScan is a quality tool to solve the tasks of

Archaeology

Today archaeology relies on photogrammetric approaches more and more often, be it a need to model some artifacts or a demand for a historic excavation mapping. PhotoScan comprises all functionality necessary to follow a photogrammetric workflow at professional level.

Archaeologist use PhotoScan for:

Platform

The software can run on any of these operating systems: Microsoft Windows, MacOS or Linux.

Table

# Title Studio Genre Reference
1 Unreal Engine Epic Games Video game [10][11]
2 Metal Gear Solid V: The Phantom Pain Kojima Productions Video game [12]
3 The Vanishing of Ethan Carter The Astronauts Video game [13][14]
4 War Thunder Gaijin Entertainment Video game [15]
5 Halo 4 343 Industries Video game [16]
6 Cyberpunk 2077 CD Projekt RED Video game [17]
7 The Talos Principle Croteam Video game[18]
8 Ryse: Son of Rome Crytek Video game [19]
9 Rustclad Skull Theatre Video game [20]
10 Chappie Image Engine Film [21]
11 Europa Report Phosphene Film [22]
12 Dracula Untold Framestore Film [23]
13 Lincoln Framestore Film [24]
14 Edge of Tomorrow Sony Pictures Imageworks Film [25]
15 Mad Max: Fury Road Iloura Film [26]
16 San Andreas Cinesite Film [27]
17 The Strain Mr. X Inc TV show [28]
18 The Knick Phosphene TV show [29]
19 Merlin Vine TV show [30]
20 Falling Skies MastersFX TV show [31]
21 White Tiger Legend Animation [32]
22 Star Wars Battlefront EA DICE Video game [33]

External links

Look up photoscan in Wiktionary, the free dictionary.
Wikimedia Commons has media related to PhotoScan.

References

  1. ↑ "Agisoft PhotoScan Wiki".
  2. ↑ "Agisoft Wiki".
  3. ↑ "Guarding Machu Picchu". Faine Greenwood.
  4. ↑ "Archeologists Wield Very Big Stick in Yucatan Exploration". Bill Giduz.
  5. ↑ "Agisoft PhotoScan". Geospatial Modeling and Visualization.
  6. ↑ "Mapping Vertical Terrain". Greg Hosilyk.
  7. ↑ "3D reconstruction and photogrammetry software PhotoScan". geoscan.aero.
  8. ↑ "Agisoft PhotoScan Pro - UAS - Australia". UAS - Australia.
  9. ↑ "MAVinci - Unmanned Aerial Systems". mavinci.de.
  10. ↑ "EPIC’s Unreal Engine Open World: Behind the Scenes". Mike Seymour.
  11. ↑ "Creating the Open World Kite Real-Time Demo in Unreal Engine 4".
  12. ↑ "Hideo Kojima GDC 2013 Panel - MGS5 & Fox Engine".
  13. ↑ "VISUAL REVOLUTION OF THE VANISHING OF ETHAN CARTER". the astronauts studio.
  14. ↑ ""The Vanishing of Ethan Carter" Explains Photogrammetry".
  15. ↑ "War Thunder 3D Body and Head Scanning".
  16. ↑ "Halo 4 Spartan Ops 3D Scanning".
  17. ↑ "Cyberpunk 2077".
  18. ↑ "Gamescom 2014: Croteam on developing The Talos Principle". Kirill Tokarev.
  19. ↑ "The tech of Crytek’s Ryse: Son of Rome". Ian Failes.
  20. ↑ "Photogrammetry in Video Games: Frequently Asked Questions".
  21. ↑ "Creative Robot". Barbara Robertson.
  22. ↑ "The making of Europa Report".
  23. ↑ "Going full vamp: Dracula Untold".
  24. ↑ "By the people and for the people: the VFX of Lincoln".
  25. ↑ "Live. Die. Repeat the effects".
  26. ↑ "A graphic tale: the visual effects of Mad Max: Fury Road".
  27. ↑ "Surviving San Andreas".
  28. ↑ "The Strain – Visual Effects". Graham Edwards.
  29. ↑ "Phosphene Contributes VFX to 'The Knick'". Computer Graphics World.
  30. ↑ "Merlin: VFX Arnold TV pipeline".
  31. ↑ "Creature Q&A: inside the Cochise effects on Falling Skies".
  32. ↑ "Interview: White Tiger Legend". VFXrecruit.
  33. ↑ "To ship Star Wars Battlefront, developers had to build their own AT-AT ... foot".
This article is issued from Wikipedia - version of the Friday, April 29, 2016. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.