Agisoft PhotoScan User Manual - Standard Edition, Version 0.8.5
Contents
1. 1 e coordinates of lens optical axis interception with sensor plane skew Skew transformation coefficient k1 k2 k3 Radial distortion coefficients pl p2 Tangential distortion coefficients How to set up appropriate values In the case of missing EXIF data or too large lens distortions manual setting of lens calibration may be required This may happen if you used an old manual lens which cannot be detected by the camera or if the photos were taken by the film camera and digitized later Some hints on selecting suitable calibration parameters are listed below e Itis very unlikely that skew transformation coefficient will have non zero value So its almost always safe to set skew parameter equal to zero e Unless you have used lenses with high distortion such as fish eyes the same applies to distortion coefficients Leave it zero and PhotoScan will estimate them on its own If ultra wide or fisheye lens were used to capture the photos you may try to estimate them using Agisoft Lens utility for lens calibration http www agisoft ru products lens e fx and fy coefficients are normally equal They may be different if the pixels of recording media were not square which is a very rare case If you know the focal length of the lens physical dimension of the optical sensor and size of picture in pixels you may easily calculate these coefficients as fx focal length in mm x dimension of sensor in pixels x dimension of sensor in2. Opens PhotoScan project file Appends existing PhotoScan project file to the current one Saves PhotoScan project file Saves PhotoScan project file with a new name Saves 3D model Saves sparse dense point cloud Quits the application Prompts to save active project Undo the last editing operation Redo the previously undone editing operation Removes selected faces from the mesh or selected points from the point cloud Crops selected faces points Inverts current selection Grows current selection Shrinks current selection Selects isolated mesh components Displays sparse point cloud reconstructed during photo alignment Displays 3D model in the shaded mode Displays 3D model in the solid mode Displays 3D model in the wireframe mode Displays 3D model in the textured mode Shows or hides camera positions reconstructed during image alignment Shows or hides the trackball Shows or hides the mesh information on screen display Switches visualisation view between Perspective and Orthographic Graphical User Interface View Menu Workspace Pe Ground Control Photos Console Workflow Menu fre Add Photos Align Photos Build Geometry Build Texture Align Chunks Merge Chunks Batch Process Tools Menu Decimate Mesh Close Holes Import Mesh Import Texture Export Texture View Mesh Statistics Import Cameras Export Cameras Undistort Photos View
3. Repeat the previos steps to add other processing steps as required 5 Click OK button to start processing Camera calibration While carrying out photos alignment PhotoScan estimates both internal and external camera orientation parameters including nonlinear radial distortions For a successful estimation the information on approximate focal length values is required Normally this data is extracted automatically from the EXIF metadata or the 50mm focal length 35mm film equivalent is assumed if the EXIF information is insufficient In some cases the 50mm guess can differ too much from the actual focal length This can lead to the failure of the alignment process In such cases it is required to specify the initial camera calibration manually To specify camera calibration manually 1 Select Camera Calibration command from the Tools menu 16 Advanced use 2 In the Camera Calibration dialog box select Initial viewing mode Select the photos to be calibrated 3 Modify the calibration parameters displayed in the corresponding edit boxes Normally only fx fy cx and cy values should be corrected 4 Click Apply button to set the calibration You will notice that the selected photos will be marked with the C flag Custom calibration 5 Click Close button to close the camera calibration dialog Calibration parameters fx fy Focal length in x and y dimensions measured in pixels CX Cy Principal point coordinates
4. bit e Intel Core 17 processor e 12GB of RAM The number of photos that can be processed by PhotoScan depends on the available RAM and reconstruction parameters used Assuming that a single photo resolution is of the order of 10 MPx 2GB RAM is sufficient to make a model based on 20 to 30 photos 12GB RAM will allow to process up to 200 300 photographs OpenCL acceleration PhotoScan supports accelerated geometry reconstruction due to exploiting of the graphics hardware GPU NVidia GeForce 8xxx series and later ATI Radeon HD 5xxx series and later PhotoScan allegedly is able to utilize processing power of any OpenCL enabled device provided that OpenCL drivers for the device are properly installed However due to the large number of various combinations of video chips driver versions and operating systems we are unable to test and guarantee PhotoScan s compatibility with every device and on every platform The table below lists currently supported devices on Windows platform only We will pay particular attention to possible problems with PhotoScan running on these devices Table 1 1 Supported Desktop GPUs on Windows platform NVIDIA AMD GeForce GTX 580 Radeon HD 7970 GeForce GTX 570 Radeon HD 6970 GeForce GTX 560 Radeon HD 6950 GeForce GTX 480 Radeon HD 6870 Installation NVIDIA AMD GeForce GTX 470 Radeon HD 5870 GeForce GTX 465 Radeon HD 5850 GeForce GTX 285 Radeon HD 5830 GeForce GTX 280 Although PhotoScan is
5. for texturing Masking areas on the photos that are occluded by outliers or obstacles helps to prevent the ghosting effect on the resulting texture atlas Mask editing Masks in PhotoScan are represented by contours bounding the certain image regions Modification of the current mask is performed by adding or subtracting selections A selection is created by using one of the supported selection tools and is not incorporated in the current mask until it is merged with a mask using Add Selection or Subtract Selection operations To edit the mask l Open the photo to be masked by double clicking on its name on the Workspace Photo Ground Control pane The photo will be opened in the main window The existing mask will be displayed as a shaded region on the photo Select the desired selection tool and generate a selection mann Click on al Add Selection toolbar button to add current selection to the mask or Subtract Selection to subtract the selection from the mask Invert Selection button allows to invert current selection prior to adding or subtracting it from the mask To import export a mask 1 Open the photo to be masked by double clicking on its name in the Workspace Photo Ground Control pane The photo will be opened in the main window Select Import Export mask command from Photo menu In the Import Export Mask dialog box select whether the mask should be imported to exported from the cu
6. set so that the object spans background objects occupies only a small significant area on the photos If masking of the area on the photos background areas is used make sure that Constrain features by mask option in the Align Photos dialog box is checked Reconstructed geometry appears to be cut and some important parts are missing Usually this indicates that a wrong reconstruction volume was selected By default PhotoScan uses an automatic reconstruction volume selection algorithm which can produce undesirable selections in some cases All object parts outside of the selected reconstruction volume are cropped and are not included in 32 Troubleshooting the final model Too large reconstruction volume selections are also undesirable as they result in longer processing time and greater memory consumption To overcome this problem a manual reconstruction volume selection tool should be used 4 Note e The photos must be aligned before the reconstruction volume can be defined To select the reconstruction volume manually 1 l Select the manual reconstruction volume selection tool using Select Volume toolbar button 2 Modify the displayed bounding box by dragging the corners of the box to the desired locations 3 Before starting to reconstruct geometry make sure that Selection option is chosen from the Reconstruction volume drop down list The photos included in the project file can t be opened and operations from the
7. supposed to be able to utilize other GPU models and on other operating systems than Windows we do not guarantee that it will work correctly Note e OpenCL acceleration can be enabled using OpenCL tab in the Preferences dialog box e Using OpenCL acceleration with mobile video chips is not recommended because of the low performance of mobile GPUs Installation procedure Installing PhotoScan on Microsoft Windows To install PhotoScan on Microsoft Windows simply run the downloaded msi file and follow the instructions Installing PhotoScan on Mac OS X Open the downloaded dmg image and drag PhotoScan application to the desired location on your hard drive Installing PhotoScan on Debian Ubuntu Unpack the downloaded archive with a program distribution kit to the desired location on your hard drive Start PhotoScan by running photoscan sh script from the program folder Restrictions of the trial mode Once PhotoScan is downloaded and installed on your computer you can run it either in the trial mode or in the full function mode At every run until you enter a serial number it will show a registration box offering two options 1 use PhotoScan in the trial mode or 2 enter the serial number to confirm the purchase The first choice is set by default so if you are still exploring PhotoScan click the Continue button and PhotoScan will start in the trial mode The employment of PhotScan in the trial mode is not time limited Several fu
8. to get through these operations and descriptions of the parameters controlling each step are explained in the corresponding sections of the Chapter 3 General workflow In some cases however additional actions may be required to get the desired results For instance pictures taken using uncommon lenses such as fish eyes may require preliminary calibration of optical system parameters In some capturing scenarios masking of certain regions of the photos may be required to exclude them from the calculations All these advanced functions are described in the Chapter 4 Advanced use 1V Overview It can take up quite a long time to reconstruct a 3D model PhotoScan allows to export obtained results and save intermediate data in a form of project files at any stage of the process If you are not familiar with the concept of projects its brief description is given at the end of the Chapter 3 General workflow In the manual you can also find instructions on the PhotoScan installation procedure and basic rules for taking good photographs 1 e pictures that provide most necessary information for 3D reconstruction Chapter 1 Installation System requirements Minimal configuration e Windows XP or later 32 or 64 bit Mac OS X Snow Leopard or later Debian Ubuntu 64 bit e Intel Core 2 Duo processor or equivalent e 2GB of RAM Recommended configuration e Windows XP or later 64 bit Mac OS X Snow Leopard or later Debian Ubuntu 64
9. use Delete Selection command from the Edit menu or click Delete Selection toolbar button or simply press Del button on the keyboard Manual points removal Points with high reprojection error can be also removed manually To remove points from the sparse cloud manually l 2 Switch to Point Cloud View mode using Point Cloud button from the toolbar Indicate points to be deleted using 1 Rectangle Selection or Circle Selection toolbar buttons To add new points to the current selection hold the Ctrl key during selection of additional points To remove some points from the current selection hold the Shift key during selection of points to be removed To delete selected points click the X Delete Selection toolbar button use Delete Selection command from the Edit menu To crop selection to the selected points click the Y Crop Selection toolbar button use Crop Selection command from the Edit menu Editing model geometry The following mesh editing tools are available in PhotoScan e Decimation tool e Connected component filtering e Manual face removal More complex editing can be done in the external 3D editing tools PhotoScan allows to export mesh and then import it back for this purpose Note e For face removal operations such as manual face removal and connected component filtering it is possible to undo the last mesh editing operation There are Undo Redo command in the Edit menu Please note that und
10. Agisoft PhotoScan User Manual Standard Edition Version 0 8 5 Agisoft PhotoScan User Manual Standard Edition Version 0 8 5 Publication date 2012 Copyright 2012 AgiSoft LLC Table of Contents OV CIVICW rra dass iv HOWE WOT etilo iv ADOUE h MAMMA n lt N S NNA ideas iv AN EDS 2 EE CO A A edaeeep ates l Ss AAA II O Ir esse l OpenGL ACC lera Oni etie a eine ans lun ete aes l Installation procedure sust a a a ii datada TEA 2 Restrictions OF he Tal MON esere AA a 2 Zo Captura POLOS dt aa tc a 3 BASIC A ARE no or An nn En a ane 3 CADRE SCCHAMOS Re A SN er binomette a ile ed RD MER Ce 3 FRCS MAICUIONS je Sn ne T E pie de de T ee nn die 4 General Work ON stat dd patada 6 Load PAS a ae iles tie a 6 AMOO POOS LISE didas ricota aaa ii NN alt tete 7 Bulidino model Ecom iy 0 A A A Po nue 8 Bulidino model text att a tdo 9 Saville mermediate eso MS da doi 11 Exportino Tesults orraa E its da io ds ae edad oa ads e ieee 11 PRIN ANC US A A ES 14 SPDS PLOJCCE lcd Tata a IS dia ae tiga dette de iest 14 Camera CADA ON adi lis 16 A A Time te CU 18 dit Ss parse POD Cloud E nt never awe 20 Edite model eco meda a a at Rite 21 A Graphical User Intertace td aaa 24 ADDHGCAUONR W INOW SR tae sa ticker A in ne 24 Meni Command SES ATA A E AA is 26 Toolbar BUMONS sisas VD A een etc ete niet 29 Bs TOUS clica 32 Photo alignment succeeds but the resulting camera positions appear to be wrong o 32 Reconstructed geometry a
11. EXIF Data View Matches Camera Calibration Preferences 4 Run Script 28 Shows or hides Workspace pane Shows or hides Grount Control pane Shows or hides Photos pane Shows or hides Console pane Loads additional photos to be processed by PhotoScan Generates camera positions and sparse point cloud Generates 3D model geometry Generates 3D model texture Aligns multiple chunks Merges multiple chunks into the single model Opens Batch Process dialog box Decimates mesh to the target face count Closes holes on the model surface Imports edited mesh from the external program Imports edited texture from the external program Exports model texture Collects and displays mesh statistics Imports camera positions and orientation data Exports camera positions and orientation data Removes nonlinear distortions by warping source photos Displayes EXIF data for the photos in the active chunk Displayes matches for the photos in the active chunk Shows camera calibration dialog box Shows preferences dialog box Opens Python Run Script dialog box Graphical User Interface Photo Menu Navigation pi _1 Rectangle Selection a Intelligent Scissors A Intelligent Paint 5 Magic Wand oo oo View Points Add Selection Subtract Selection 7 Invert Selection Import Mask Export Mask K Reset Mask Turn Shading On Off Help Menu Contents C
12. Workflow menu fail Probably the locations of photos have changed in respect to the location of the project file The references to the original photos are stored in the project files in the form of relative paths These references become invalid when the project file alone is moved to another location or when the photos are moved without changing the project file placement To identify the expected photo location 1 Open the context menu of the photo by right clicking on its title in the Workspace pane 2 Select Show Info command from the context menu 3 A dialog box with information on the selected photo including the path to the image data file will be displayed To fix the problem move the photos to the original locations indicated by the data displayed in the information dialog box If the relative location of the project file in respect to the photos location must be changed the project file should be resaved using the Save As command from the File menu To move the project file in respect to the source photo locations 1 Open the project to be moved using Open command from the File menu 2 Select Save As command from the File menu Browse to the folder where the project file should be placed and specify the destination file name for the project Click Save button to save the project Once the project file was saved to a new location the original project file can be removed if necessary 33
13. al mesh 0 1f no decimation is required Filter threshold Specifies the maximum face count of small connected components to be removed after surface reconstruction in percent of the total face count The 0 value disables connected component filtering Hole threshold Height field methods only Specifies the maximum size of holes to be filled after surface reconstruction in percent of the total surface area The 0 value disables automatic hole filling i Note e PhotoScan tends to produce 3D models with excessive geometry resolution so it is recommended to perform mesh decimation after geometry computation More information on mesh decimation and other 3D model geometry editing tools is given in the Editing model geometry section Building model texture To generate 3D model texture 1 Select Build Texture command from the Workflow menu 2 Select the desired texture generation parameters in the Build Texture dialog box Click OK button when done 3 The progress dialog box will appear displaying the current processing status To cancel processing click Cancel button General workflow Texture mapping modes The texture mapping mode determines how the object texture will be packed in the texture atlas Proper texture mapping mode selection helps to obtain optimal texture packing and consequently better visual quality of the final model Generic The default mode is the Generic mapping mode it allows to parameterize tex
14. and oo 00 View Points Add Selection Subtract Selection Invert Selection e Undo 30 Circle selection tool Volume selection tool Volume rotation tool Removes selected faces points Crops selected faces points Undo the last editing operation Redo the previously undone editing operation Shows hides sparse point cloud reconstructed during image alignment Displays 3D model in the shaded mode Displays 3D model in the solid mode Displays 3D model in the wireframe mode Displays 3D model in the textured mode Shows hides camera positions reconstructed during image alignment Shows hides aligned chunks Resets model view Switches to the navigation mode Rectangle selection tool Intelligent scissors tool Intelligent paint tool Magic wand tool Shows hides feature points used for alignment of the photo Adds current selection to the mask Subtracts current selection from the mask Inverts current selection Undo the last mask editing operation Photo view commands qu Redo e IL Rotate Right Ja E Rotate Left He Zoom In Zoom Out q Turn Shading On Off Graphical User Interface 31 Redo the previously undone mask editing operation Rotates the photo clockwise Rotates the photo counterclockwise Increases magnification Decreases magnification Turns mask shading on or off Appendix B Troubleshooting Photo alignment succeeds but th
15. assumption If the correct focal length value differs significantly from 50 mm the alignment can give incorrect results or even fail In such cases it is required to specify initial camera calibration manually The details of necessary EXIF tags and instructions for manual setting of the calibration parameters are given in the Camera calibration section Capturing photos Lens distortion The distortion of the lens being used to capture the photos should be well modeled using the Brown s distortion model Otherwise it is almost impossible to build a precise 3D model Fish eyes and ultra wide angle lenses are poorly modeled by the distortion model implemented which leads to inaccurate reconstructions Chapter 3 General workflow The processing of images by PhotoScan includes the following main steps e loading photos into PhotoScan e inspecting loaded images removing unnecessary images e aligning photos e building 3D model e editing 3D model e exporting results If you are using PhotoScan in the full function not the trial mode intermediate results of the image processing can be saved at any stage in the form of project files and can be used later The concept of projects and project files 1s briefly explained in the Saving intermediate results section The list above represents all the necessary steps involved in the construction of a textured 3D model from your photos Some additional tools which you may find to be usefu
16. cancel processing click the Cancel button 4 Note e Chunks alignment can be performed if at least two chunks contain aligned photos Chunks alignment parameters The following parameters control the chunks alignment procedure and can be modified in the Align Chunks dialog box Method Defines the chunks alignment method Point based method alignes chunks by matching photos across the different chunks Accuracy Point based alignment only Higher accuracy setting helps to obtain more accurate chunk alignment results Lower accuracy setting can be used to get the rough chunk alignment in the shorter time Preselect image pairs Point based alignment only The alignment process of many chunks may take a long time A signficant portion of this time is spent for matching of detected features across the photos Image pair preselection option can speed up this process by selection of a subset of image pairs to be matched Constrain features by mask Point based alignment only When this option is enabled features detected in the masked image regions are discarded For additional information on the usage of masks refer to the Using masks section Merging models from several chunks After alignment is complete the separate models can be merged into a single model To merge models 1 Select Merge Chunks command from the Workflow menu 2 Inthe Merge Chunks dialog box select chunks to be merged and the desired merging options Click OK b
17. e Editing model geometry section of the manual Point Cloud export To export sparse or dense point cloud 1 Select Export Points command from the File menu 2 Browse the destination folder choose the file type and print in the file name Click Save button 3 In the Export Points dialog box set the Point Cloud value for the Quality parameter if you need to export sparse point cloud To export a dense point cloud choose an approprite quality among the rest of the values 4 Specify the coordinate system and indicate export parameters applicable to the selected file type 11 General workflow 5 Click OK button to start export 6 The progress dialog box will appear displaying the current processing status To cancel processing click Cancel button In some cases it may be reasonable to edit sparse point cloud before exporting it To read about point cloud edititing refer to the Editing sparse point cloud section of the manual PhotoScan supports point cloud export in the following formats e Wavefront OBJ e Stanford PLY e XYZ text file format only for sparse point cloud export e ASPRS LAS Note e Saving color information of the point cloud is supported by the PLY TXT and LAS file formats e Saving point normals information is supported by the OBJ PLY and TXT file formats Camera calibration and positions data export To export camera calibration and positions data select Export Cameras command from the Tools me
18. e in a separate file The texture file should be kept in the same directory as the main file describing the geometry If the texture atlas was not built only the model geometry 1s exported 13 Chapter 4 Advanced use Splitting project In some cases it is very hard or even impossible to generate 3D model of the whole object in one go This could happen for instance if the total amount of photographs is too large to be processed To overcome this PhotoScan offers the possibility of splitting the set of photos in several separate chunks within the project The alignment of photos building geometry and forming the texture atlas may be performed on each chunk separately and then the resulting 3D models may be combined together Working with chunks is no more difficult than using PhotoScan following the general workflow In fact in PhotoScan always exists at least one active chunk and all the 3D model processing workflow operations are applied to this chunk All you need to know to work with several chunks is how to create chunks and how to combine resulting 3D models from separate chunks into one model Creating a chunk To create new chunk click on the SA dd Chunk toolbar button on the Workspace pane or select Add Chunk command from the Workspace context menu available by right clicking on the root element in the Workspace pane After the chunk is created you may load photos in it align them generate mesh surface model build texture at
19. e point cloud and estimated camera positions can be exported for processing with another software if needed Incorrectly positioned photos can be realigned To realign a subset of photos 1 Reset alignment for incorrectly positioned photos using Reset Photo Alignment command from the photo context menu 2 Select photos to be realigned and use Align Selected Photos command from the photo context menu 3 The progress dialog box will appear displaying the current processing status To cancel processing click Cancel button General workflow Alignment parameters The following parameters control the photo alignment procedure and can be modified in the Align Photos dialog box Accuracy Higher accuracy setting helps to obtain more accurate camera position estimates Lower accuracy setting can be used to get the rough camera positions in a shorter period of time Pair preselection The alignment process of large photo sets can take a long time A significant portion of this time period is spent on matching of detected features across the photos Image pair preselection option may speed up this process due to selection of a subset of image pairs to be matched In the Generic preselection mode the overlapping pairs of photos are selected by matching photos using lower accuracy setting first Constrain features by mask When this option is enabled features detected in the masked image regions are discarded For additional information on the
20. e resulting camera positions appear to be wrong The main reasons for the wrong photo alignment are listed below e Small overlap between the photos e Object movement against the background scene e Insufficient number of object surface details captured by the camera e Strong blur or noise on the source photos To obtain more information about the reason for alignment failure for the specific dataset it may be helpful to inspect feature points used for alignment of the photos as seen by the PhotoScan algorithms To inspect feature points used for alignment 1 Open the photo to be inspected by double clicking on its name on the Workspace pane The photo will be displayed in the main window oo Switch to the point view mode using 48 View Points toolbar button The points used for alignment of the photo will be overlayed on the image Depending on the observed point placement the following recommendations may help to solve the problem Problem Possible reason Recommended solution Not enough points were Low image quality or Try making better photos with a better camera used too small photo overlap placement Pay attention to the camera settings like proper level of ISO Use a tripod if required Setting a proper lighting may be also helpful to capture higher amount of surface details The majority of points The object was not static Mask out background regions or consider used belong to the during capturing or shooting another photo
21. e when using a turn table to capture the photos Masking may be also useful when the object of interest occupies only a small part of the photo In this case a small number of useful matches can be filtered out mistakenly as a noise among a much greater number of matches between background objects Building 3D model While building the model masked areas are not used in the surface generation process Masking can be used to reduce the resulting model complexity by eliminating the areas on the photos that are not of interest Masked areas are always excluded from processing during surface reconstruction and texture generation Stages Let s take for instance a set of photos of some object Along with an object itself on each photo some background areas are present These areas may be useful for more precise camera positioning so it is better to use them while aligning the photos However impact of these areas at the building geometry stage is exactly opposite If they are used for building geometry the resulting model will contain object of interest and its background Background geometry will consume some part of mesh polygons that could be otherwise used for modeling the main object 18 Advanced use Setting the masks for such background areas allows to avoid this problem and increases the precision and quality of geometry reconstruction Building texture atlas During texture atlas generation masked areas on the photos are not used
22. ects of texture atlas generation Texture from Single photo mapping mode only Specifies the photo to be used for texturing Available only in the Single photo mapping mode Blending mode not used in Single photo mode Selects the way how pixel values from different photos will be combined in the final texture Average uses the average value of all pixels from individual photos Mosaic gives more quality for orthophoto and texture atlas than Average mode since it does not mix image details of overlapping photos but uses more appropriate photo Mosaic texture blending mode is especially useful for orthophoto generation based on approximate geometric model Max Intensity the photo which has maximum intensity of the corresponding pixel is selected Min Intensity the photo which has minimum intensity of the corresponding pixel is selected Atlas width Specifies the width of the texture atlas in pixels 10 General workflow Atlas height Specifies the height of the texture atlas in pixels Saving intermediate results Certain stages of 3D model reconstruction can take a long time The full chain of operations could easily last for 4 6 hours when building a model from hundreds of photos It is not always possible to finish all the operations in one run PhotoScan allows to save intermediate results in a project file PhotoScan project files may contain the following information e List of loaded photographs with reference paths
23. eir status The following flags can appear next to the photo name NC Not calibrated Notifies that the EXIF data available 1s not sufficient to estimate the camera focal length In this case PhotoScan assumes that the corresponding photo was taken using 50mm lens 35mm film equivalent If the actual focal length differs significantly from this value manual calibration may be required More details on manual camera calibration can be found in the Camera calibration section NA Not aligned Notifies that external camera orientation parameters were not estimated for the current photo yet Images loaded to PhotoScan will not be aligned until you perform the next step photos alignment Aligning photos Once photos are loaded into PhotoScan they need to be aligned At this stage PhotoScan finds the camera position for each photo and builds a point cloud model To align a set of photos l Select Align Photos command from the Workflow menu 2 Inthe Align Photos dialog box select the desired alignment options Click OK button when done 3 The progress dialog box will appear displaying the current processing status To cancel processing click Cancel button Alignment having been complete computed camera positions and a sparse point cloud will be displayed You can inspect alignment results and remove incorrectly positioned photos if any To see the matches between any two photos use View Matches command from the Tools menu Th
24. er stages e Remove items e Increase or decrease icons size e Filter photos by feature points or by markers Console pane Console pane is used for e Displaying auxiliary information e Displaying error messages e Python commands Buttons located on the pane toolbar allow e Save log e Clear log e Execute Python script Ground Control pane Ground Control pane is designed for e Displaying camera and or markers coordinates e Displaying camera orientation e Displaying estimation errors Buttons located on the pane toolbar allow e Import export reference coordinates e Convert reference coordinates from one system to the other e Indicate coordinate system and measurement accuracy to be assumed through Settings dialog e Optimize estimation and update data e Switch between source coordinates estimated coordinates and errors views Note e To open any pane select a corresponding command from the View menu Menu Commands File Menu E New Creates an empty PhotoScan project 26 Graphical User Interface File Menu E Open Append Save Save As Export Model Export Points Exit Edit Menu e Undo qu Redo a Delete Selection pal Crop Selection Invert Selection Grow Selection Shrink Selection Select Components View Menu oo oo Point Cloud Y Shaded Solid El Wireframe Textured IO Show Cameras Show Trackball Show Info Perspective Orthographic 24
25. g of any kind of object It should be selected for closed objects such as statues buildings etc It doesn t make any assumptions on the type of the object modeled which comes at a cost of higher memory consumption General workflow Height field The Height field object type is optimized for modeling of planar surfaces such as terrains or bas reliefs It should be selected for aerial photography processing as it requires lower amount of memory and allows for larger data sets processing Geometry type Sharp Sharp geometry type option leads to more accurate reconstruction results and does not introduce extra geometry like hole filling patches Manual hole filling is usually required at the post processing step Smooth Smooth geometry type option produces watertight reconstructions with no or little holes on resulting surface Large areas of extra geometry might be generated with this method but they could be easily removed later using selection and cropping tools Smooth setting is recommended for orthophoto generation Reconstruction parameters Quality Specifies the desired reconstruction quality Higher quality settings can be used to obtain more detailed and accurate geometry but require longer time for processing Point Cloud quality setting is used for fast 3D model generation based solely on the sparse point cloud The option is available for Smooth geometry type only Face count Specifies the maximum face count in the fin
26. heck for Updates Register Product ed About PhotoScan Toolbar Buttons General commands H New E Open Save 3D view commands Navigation r 1 L 4 Rectangle Selection 29 Selects navigation mode Rectangle selection tool Intelligent Scissors selection tool Intelligent Paint selection tool Magic Wand selection tool Shows detected features used for alignment of the current photo Adds current selection to the mask Subtracts current selection from the mask Inverts current selection Imports masks or creates mask from alpha channel Exports masks templates Resets mask for the current photo Turns mask shading on or off Displays help contents Checks if PhotoScan update is available for download Activates the product using the activation key Displays program information version number and copyright Creates a new PhotoScan project file Opens a PhotoScan project file Saves a PhotoScan project file Navigation tool Rectangle selection tool Graphical User Interface 3D view commands CA Circle Selection de Resize region Ce Rotate Region X Delete Selection pal Crop Selection E Undo qu Redo 3D view settings oo o0 Point Cloud Shaded E solid El Wireframe Textured D Show Cameras Show Aligned Chunks 04 Reset View Photo view commands Navigation r 7 _1 Rectangle Selection a Intelligent Scissors LA Intelligent Paint 5 Magic W
27. ion toolbar buttons 22 Advanced use Make the selection using the mouse To add new faces to the current selection hold the Ctrl key during selection of additional faces To remove some faces from the current selection hold the Shift key during selection of faces to be removed To delete selected faces click the X Delete Selection toolbar button or use Delete Selection command from the Edit menu To crop selection to the selected faces click the Crop Selection toolbar button or use Crop Selection command from the Edit menu To grow or shrink current selection 1 2 To grow current selection press PageUp key in the selection mode To grow selection by even a larger amount press PageUp while holding Shift key pressed To shrink current selection press PageDown key in the selection mode To shrink selection by even a larger amount press PageDown while holding Shift key pressed Editing mesh in the external program To export mesh for editing in the external program 1 2 Select Export Model command from the File menu In the Save As dialog box specify the desired mesh format in the Save as type combo box Select the file name to be used for the model and click Save button In the opened dialog box specify additional parameters specific to the selected file format Click OK button when done To import edited mesh 1 2 Select Import Mesh command from the Tools menu In the Open dialog box browse to
28. l are described in the Chapter 4 Advanced use Loading photos Before starting any operation it is necessary to point out what photos will be used as a source for 3D reconstruction In fact photographs themselves are not loaded into PhotoScan until they are needed So when you load photos you only indicate photographs that will be used for further processing To load a set of photos I Select Add Photos command from the Workflow menu or click A dd Photos toolbar button 2 In the Add Photos dialog box browse to the folder containing the images and select files to be processed Then click Open button 3 Selected photos will appear on the Workspace pane i Note e PhotoScan accepts the following image formats JPEG TIFF PNG BMP PPM and JPEG Multi Picture Format MPO Photos in any other format will not be shown in the Add Photos dialog box To work with such photos you will need to convert them in one of the supported formats If you have loaded some unwanted photos you can easily remove them at any moment To remove unwanted photos 1 On the Workspace pane select the photos to be removed General workflow 2 Right click on the selected photos and choose Remove Items command from the opened context menu or click Remo Items toolbar button The selected photos will be removed from the working set Inspecting the loaded photos Loaded photos are displayed on the Workspace pane along with flags reflecting th
29. las export the models at any stage and so on The models in the chunks are not linked with each other To move photos from one chunk to another simply select them in the list of photos on the Workspace pane and then drag and drop in the desirable chunk Working with chunks All operations within the chunk are carried out following the common workflow loading photographs aligning them building geometry model building texture atlas exporting 3D model and so on Note that all these operations are applied to the active chunk When a new chunk is created it is activated automatically Save project operation saves the content of all chunks To set another chunk as active 1 Right click on the chunk title on the Workspace pane 2 Select Set Active command from the context menu To remove chunk 1 Right click on the chunk title on the Workspace pane 2 Select Remove Items command from the context menu Aligning models from several chunks After the partial 3D models are built in several chunks they can be merged together Before merging the models they need to be aligned 14 Advanced use To align models from separate chunks 1 Select Align Chunks command from the Workflow menu 2 Inthe Align Chunks dialog box select chunks to be aligned indicate reference chunk with a double click Set desired alignment options Click OK button when done 3 The progress dialog box will appear displaying the current processing status To
30. lly transform the images e More photos is better than not enough e Spending some time planning your shot might be very useful e Make sure to study the following schemes and read the list of restrictions before you get out for shooting photographs Capturing scenarios The following figures represent several typical capturing scenarios Facade Incorrect Facade Correct Capturing photos Interior Incorrect Interior Correct Isolated Object Incorrect Isolated Object Correct nn Restrictions In some cases it might be very difficult or even impossible to build a correct 3D model from a set of pictures A short list of typical reasons for photographs unsuitability is given below Modifications of photographs PhotoScan can process only unmodified photos as they were taken by a digital photo camera Processing the photos which were manually cropped or geometrically warped is likely to fail or produce highly inaccurate results Photometric modifications do not affect reconstruction results Lack of EXIF data To estimate the field of view for each photo PhotoScan uses the information saved in the EXIF part of each picture If EXIF data are available you can expect to get the best possible 3D reconstruction However 3D scene can also be reconstructed in the absence of EXIF data In this case PhotoScan assumes that the 35mm focal length equivalent equals to 50 mm and tries to align the photos in accordance with this
31. mm Similar formula applies to fy e For conventional non tilt and shift lenses the optical axis of the lens is likely to intercept the optical sensor plane in its geometric center So for cx and cy coefficient the coordinates of the image center would be the good guess Automatic refinement of camera calibration parameters By default PhotoScan considers specified camera calibration parameters as the initial guess and refines them later during the photo alignment That is generally a desirable behaviour However in the cases when 17 Advanced use the camera calibration is known precisely like in case of a metric camera it may be required to protect camera calibration parameters from optimization To fix the camera calibration parameters select the Fix calibration check box in the Camera Calibration dialog Using masks Overview Masks are used in PhotoScan to specify the areas on the photos which can otherwise be confusing to the program or lead to incorrect reconstruction results Masks can be applied at the following stages of processing e Alignment of the photos e Building 3D model geometry e Building 3D model texture Alignment of the photos Masked areas can be excluded during feature point detection Thus the objects on the masked parts of the photos are not taken into account while estimating camera positions This is important in the setups where the object of interest is not static with respect to the scene lik
32. nctions however are not available in the trial mode These functions are the following e saving the project e exporting reconstruction results you can only view a 3D model on the screen To use PhotoScan in the full function mode you have to purchase it On purchasing you will get the serial number to enter into the registration box on starting PhotoScan Once the serial number is entered the registration box will not appear again and you will get full access to all functions of the program Chapter 2 Capturing photos Before loading your photographs into PhotoScan you need to take them and select those suitable for 3D model reconstruction Photographs can be taken by any standard digital camera as long as you follow some specific capturing guidelines This section explains the general principles of taking and selecting pictures that provide the most appropriate data for 3D model generation Basic rules e Use a digital camera with reasonably high resolution 5 MPix or more e Wide angle lenses suit better for reconstructing spatial relations between objects than telephoto ones e Avoid not textured and flat objects or scenes e Avoid shiny and transparent objects e Avoid unwanted foregrounds and moving objects as much as possible e Shoot shiny objects under a cloudy sky e Shoot pictures of the scene with a lot of overlap e Capture most important scene content from multiple viewpoints 3 or more e Do not crop or geometrica
33. nu To export import only camera calibration data select Camera Calibration command from the Tools menu PhotoScan supports camera data export in the following formats e PhotoScan structure file format XML based e Bundler OUT file format e CHAN file format e Boujou TXT file format e Omega Phi Kappa text file format 4 Note e Camera data export in Bundler and Boujou file formats will save sparse point cloud data in the same file e Camera data export in Bundler file format would not save distortion coefficient k3 3D model export To export 3D model 1 Select Export Model command from the File menu 12 General workflow 2 Browse the destination folder choose the file type and print in the file name Click Save button 3 In the Export Model dialog box specify the coordinate system and indicate export parameters applicable to the selected file type if needed 4 Click OK button to start export 5 The progress dialog box will appear displaying the current processing status To cancel processing click Cancel button Note e If the model is referenced in local coordinates PhotoScan can write a KML file for the exported model to be correctly located on Google maps PhotoScan supports model export in the following formats e Wavefront OBJ e 3DS file format e VRML e COLLADA e Stanford PLY e Autodesk DXF e U3D e Adobe PDF Some file formats OBJ 3DS VRML COLLADA PLY save texture imag
34. o redo commands are not supported for mesh decimation and this operation can not be undone Decimation tool Decimation is a tool used to decrease the geometric resolution of the model by replacing high resolution mesh with a lower resolution one which is still capable of representing the object geometry with high accuracy PhotoScan tends to produce 3D models with excessive geometry resolution so mesh decimation is usually a desirable step after geometry computation 21 Advanced use The high detailed models may contain hundreds of thousands polygons While it is acceptable to work with such a complex models in 3D editor tools in most conventional tools like Adobe Reader or Google Earth high complexity of 3D models may noticeably decrease application performance High complexity also results in longer time required to build texture and to export model in pdf file format In some cases it is desirable to keep as much geometry details as possible like it is needed for scientific and archive purposes However if there are no special requirements it is recommended to decimate the model down to 100 000 200 000 polygons for exporting in PDF and to 100 000 or even less for displaying in Google Earth and alike tools To decimate 3D model 1 Select Decimate Mesh command from the Tools menu 2 Inthe Decimate Mesh dialog box specify the target number of faces which should remain in the final model Click on the OK button to start decimati
35. ograms For instance the point cloud model can be used in a 3D editor as a reference On the contrary the set of camera positions is required for further 3D model construction by PhotoScan 2 The next stage is building geometry Based on the estimated camera positions and pictures themselves a 3D polygon mesh representing the object surface is build by PhotoScan Four algorithmic methods available in PhotoScan can be applied to 3D mesh generation Arbitrary Smooth Arbitrary Sharp Height field Smooth and Height field Sharp methods Additionally there is a Point Cloud based method for fast geometry generation based on the sparse point cloud alone Having built the mesh it may be necessary to edit it Some corrections such as mesh decimation removal of detached components closing of holes in the mesh etc can be performed by PhotoScan For more complex editing you have to engage external 3D editor tools PhotoScan allows to export the mesh edit it by another software and import it back 3 After the geometry i e the mesh is constructed it can be textured and or used for orthophoto generation Several texturing modes are available in PhotoScan they are described in the corresponding section of this manual About the manual Basically the sequence of axtions described above covers most of the model processing needs All these operations are carried out automatically according to the parameters set by user Instructions on how
36. on 3 The progress dialog box will appear displaying the current processing status To cancel processing click on the Cancel button i Note e Texture atlas is discarded during decimation process You will have to rebuild texture atlas after decimation is complete Connected component filtering In some cases reconstructed geometry may contain the cloud of small isolated mesh fragments surrounding the main model Connected component filtering helps to remove small isolated mesh fragments which are usually reconstructed incorrectly or are not needed To remove small isolated mesh fragments 1 Select Gradual Selection command from the Edit menu 2 In the Gradual Selection dialog box select the size of isolated components to be removed using the slider You can observe how the selection changes while dragging the slider Click OK button to finalize the selection To remove the selected components use Delete Selection command from the Edit menu or click K Delete Selection toolbar button or simply press Del button on the keyboard Note that PhotoScan always selects the fragments starting from the smallest ones If the model contains only one component the selection will be empty Manual face removal Unnecessary and excessive sections of model geometry can be also removed manually To remove part of the faces manually r 4 Pi I Select rectangle or circle selection tool using 1 Rectangle Selection or Circle Select
37. ppears to be cut and some important parts are missing 32 The photos included in the project file can t be opened and operations from the Workflow MEN aoe deena eapana eating see 33 111 Overview Agisoft PhotoScan is an advanced image based 3D modeling solution aimed at creating professional quality 3D content from still images Based on the latest multi view 3D reconstruction technology it operates with arbitrary images and is efficient in both controlled and uncontrolled conditions Photos can be taken from any position providing that the object to be reconstructed is visible on at least two photos Both image alignment and 3D model reconstruction are fully automated How it works Generally the final goal of photographs processing by PhotoScan is to build a textured 3D model The procedure of photographs processing and 3D model construction comprises three main stages 1 The first stage is photographs alignment At this stage PhotoScan searches for common points on photographs and matches them as well as it finds the position of the camera for each picture and refines camera calibration parameters As a result a sparse point cloud and a set of camera positions are formed The point cloud represents the results of photos alignment and will not be directly used in the further 3D model construction procedure except for the point cloud based reconstruction method However it can be exported for further usage in external pr
38. r navigation in the 3D view 24 Graphical User Interface Tool Keyboard modifier Rotation Tool Default Pan Tool Ctrl key pressed Zooming Tool Shift key pressed All navigation tools are accessible in the navigation mode only To enter the navigation mode click the ki Navigation toolbar button Note e Zooming into the model can be also controlled by the mouse wheel Photo view Photo view tab 1s used for displaying individual photos as well as masks and markers on them Photo view is visible only if any photo is opened To open the photo double click on its name on the Workspace Ground Control or Photos pane Workspace pane On the Workspace pane all elements comprising the current project are diplayed These elements can include e List of chunks in the project e List of photographs in each chunk e List of markers in each chunk e 3D models in separate chunks Buttons located on the pane toolbar allow e Add chunks e Add photographs e Add markers e Enable or disable certain photographs or chunks for processing at further stages e Remove items Each element in the list is linked with the context menu providing quick access to some common operations Photos pane Photos pane displays the list of photos in the active chunk in the form of thumbnails Buttons located on the pane toolbar allow e Add photos 25 Graphical User Interface e Enable or disable certain photos or chunks for processing at furth
39. rrent photo only all photos in the chunk or to from the entire Workspace Click OK button when done If photo has alpha channel 1t could be applied as a mask to one or more photos in Workspace through Import mask dialog Supported selection tools The following tools can be used for creating selections Rectangle selection tool Rectangle selection tool 1s used to select large areas or to clean up the mask after other selection tools were applied Note e To add new area to the current selection hold the Ctrl key during selection of additional arae 19 Advanced use Intelligent scissors tool Intelligent scissors is used to generate a selection by specifying its boundary The boundary is formed by selecting a sequence of vertices with a mouse which are automatically connected with segments The segments can be formed either by straight lines or by curved contours snapped to the object boundaries To enable snapping hold Ctrl key while selecting the next vertex To complete the selection the boundary should be closed by clicking on the first boundary vertex X Intelligent paint tool Intelligent paint tool is used to paint a selection by the mouse continuously adding small image regions bounded by object boundaries Magic Wand tool Magic Wand tool is used to select uniform areas of the image To make a selection with a Magic Wand tool click inside the region to be selected The range of pixel colors selec
40. ted by Magic Wand is controlled by the tolerance value At lower tolerance values the tool selects fewer colors similar to the pixel you click with the Magic Wand tool Higher value broadens the range of colors selected Note e The masks are generated individually for each image If some object should be masked out it should be masked out on all photos where that object appears Editing sparse point cloud The following point cloud editing tools are available in PhotoScan e Reprojection error filtering e Manual points removal 4 Note e For points removal operations such as manual points removal and reprojection error filtering it is possible to undo the last point cloud editing operation There are Undo Redo command in the Edit menu Reprojection error filtering In some cases you may need to find out where the points with high reprojection error are located within the cloud Reprojection error filtering helps to highlight such points which usually are supposed to be removed To remove points with high reprojection error 1 Switch to Point Cloud View mode using Point Cloud button from the toolbar 2 Select Gradual Selection command from the Edit menu 20 Advanced use In the Gradual Selection dialog box set the value of reprojection error of points to be selected using the slider You can observe how the selection changes while dragging the slider Click OK button to finalize the selection To remove selected points
41. the file with the edited model and click Open Note e PhotoScan suppots loading models in Wavefront OBJ and Stanford PLY file formats only Please make sure to select one of these file formats when exporting model from the external 3D editor 23 Appendix A Graphical User Interface Application Window 5 xX H saharovl_3 p5z Agisoft PhotoScan names Grm File Edit View Workflow Tools Photo Help Depor xy 0e ses 50 Gat oise ea 3 x 8 Workspace 1 chunk 4 Chunk 1 32 phot 4 Photos IMG 3023 JP IMG_3024 JP IMG_3025 JP IMG _3026 JP IMG _3027 JP IMG 3028 JP E IMG_3029 JP IMG _3030 JP IMG 3031 JP IMG _3032 JP IMG_3033 JP IMG _3034 JP IMG_3035 JP IMG_3036 JP IMG 3037 JP IMG _3038 JP IMG 2029 IP gt Photos SS BOOx AAR IMG_3023 JPG IMG _3024JPG IMG_3025JPG IMG_3026 JPG IMG_3027 JPG Photos General view of application window Model view Model view tab is used for displaying 3D data as well as for 3D model and point cloud editing The view of the model depends on the current processing stage and is also controlled by mode selection buttons on the PhotoScan toolbar Model can be shown in textured solid shaded or wireframe mode Along with the model the results of photo alignment can be displayed These include point cloud and camera positions for each photo PhotoScan supports the following tools fo
42. to the image files e Photo alignment data such as information on camera positions point cloud model and set of refined camera calibration parameters for each photo e Reconstructed 3D model with any changes made by user This includes geometry and texture if they were built e Structure of the project 1 e number of chunks in the project and their content You can save the project at the end of any processing stage and return to it later To restart work simply load the corresponding file into PhotoScan Project files can also serve as backup files or be used to save different versions of the same model Project files use relative paths to reference original photos Thus when moving or copying the project file to another location do not forget to move or copy photographs with all the folder structure involved as well Otherwise PhotoScan will fail to run any operation requiring source images although the project file including the reconstructed model will be loaded up correctly Exporting results PhotoScan supports export of processing results in various representations Point clouds and camera calibration data can be exported right after photo alignment is completed All other export options are available after the geometry is built In some cases editing model geometry in the external software may be required PhotoScan supports model export for editing in external software and then allows to import it back as it is described in th
43. ture atlas for arbitrary geometry No assumptions regarding the type of the scene to be processed are made program tries to create as uniform texture as possible Adaptive orthophoto In the Adaptive orthophoto mapping mode the object surface is split into the flat part and vertical regions The flat part of the surface is textured using the orthographic projection while vertical regions are textured separately to maintain accurate texture representation in such regions When in the Adaptive orthophoto mapping mode program tends to produce more compact texture representation for nearly planar scenes while maintaining good texture quality for vertical surfaces such as walls of the buildings Orthophoto In the Orthophoto mapping mode the whole object surface is textured in the orthographic projection The Orthophoto mapping mode produces even more compact texture representation than the Adaptive orthophoto mode at the expense of texture quality in vertical regions Single photo The Single photo mapping mode allows to generate texture from a single photo The photo to be used for texturing can be selected from Texture from list Keep uv The Keep uv mapping mode generates texture atlas using current texture parameterization It can be used to rebuild texture atlas using different resolution or to generate the atlas for the model parametrized in the external software Texture generation parameters The following parameters control various asp
44. usage of masks please refer to the Using masks section Building model geometry 3D model reconstruction is a computationally intensive operation and can take a long time depending on the quantity and resolution of loaded photos It is recommended to build a model with the lowest quality first to estimate the applicability of the chosen reconstruction method and then to recompute the results using a higher quality setting It is also recommended to save the project before building the geometry To build a 3D model Check the reconstruction volume bounding box To adjust the bounding box use the resize Region and Rotate Region toolbar buttons Rotate the bounding box and then drag corners of the box to the desired positions If the Height field reconstruction method is applied the red side of the bounding box will define the reconstruction plane In this case make sure that the bounding box is correctly oriented 2 Select the Build Geometry command from the Workflow menu 3 Inthe Build Geometry dialog box select the desired reconstruction parameters Click OK button when done 4 The progress dialog box will appear displaying the current processing status To cancel processing click Cancel button Reconstruction methods Agisoft PhotoScan supports several reconstruction methods and settings which help to produce optimal reconstructions for a given data set Object type Arbitrary Arbitrary object type can be used for modelin
45. utton when done 3 PhotoScan will merge the separate chunks into one The merged chunk will be displayed in the project content list on Workspace pane Chunks merging parameters The following parameters control the chunks merging procedure and can be modified in the Merge Chunks dialog box 15 Advanced use Merge models Defines if models from the selected chunks are merged Merge markers Defines if markers from the selected chunks are merged only markers with the same labels would be merged Chunks merging result i e photos points and geometry will be stored in the new chunk and it may be treated as common chunk e g textured and or exported Batch processing PhotoScan allows to perform general workflow operations with multiple chunks automatically It is useful when dealing with a large number of chunks to be processed Batch processing can be applied to all chunks in the Workspace to unprocessed chunks only or to the chunks selected by the user Batch processing can perform the following operations e Align photos e Build geometry e Build texture e Save project after every completed operation To start batch processing 1 Select Batch Process command from the Workflow menu 2 Click Add to add the desired processing stages 3 In the Add Job dialog select the kind of operation to be performed the list of chunks it should be applied to and desired processing parameters Click OK button when done 4
Download Pdf Manuals
Related Search