Home

CloudCompare - Danielgm.net

1. Classic rendering using normals left and PCV rendering right The two principal parameters of the algorithm modifiable via the associated dialogue box figure 2 35 are the number of light rays For each direction of lighting ray the algorithm projects a graphical map of the entities as seen from that direction and calculates the visibility of the points or vertices of a mesh This information is accumulated for each direction to calculate the global lighting The more rays that are used the more differences between points will be apparent and the more dynamic it will appear On the other hand the calculation time is proportional to the number of rays the resolution of the rendering buffer The projection of entities in a viewing direction is done in a video buffer of which the resolution contributes to the ability to distinguish between points The higher the resolution the more distinctly the points will appear with better calculations of the individual lighting effects and a more the points will appear separate enabling better lighting calculations and a higher quality of result On the other hand if the resolution is too high calculation time and memory consumption will be unweildly depending on the performance of the graphics card Also be aware that the cloud may become porous and the light may shine through see remark below In the case of a mesh this does not pose any problem Current graphics cards provide g
2. LE Cloud Mesh distance cf section 2 3 6 H Statistical test cf section 2 3 9 SS Label connected components cf section 2 3 10 op Segment cf section 1 5 1 Translate Rotate cf section 1 5 2 1 8 Keyboard shortcuts 21 Scalar Field Tools Toolbar Show histogram display the histogram of the scalar field Compute statistical parameters cf section 2 3 8 Filter by value cf section 2 2 20 Gradient cf section 2 2 18 o Gaussian filter cf section 2 2 19 Delete current scalar field removes the active scalar field of the selected uy cloud Difference cf section 2 2 21 diff E M Hidden Points Removal cf section 2 5 2 HPR me Global enlightening cf section 2 5 1 Viewing Tools Toolbar 0 za OCI IDE A Oo 0 This toolbar is explained in section 1 3 4 The toolbars are displayed or hidden according to whether they are checked or not in the Display Toolbars sub menu To change the visibility of a toolbar click on its name in Display Toolbars 1 8 Keyboard shortcuts The following table lists the keyboard shortcuts available in CloudCompare Keys Command Peer F1 Help Display the help documentation for CloudCompare e e 22 Chapter 1 Interface Keys Toggle centered Position the center of rotation of the cf section 1 3 4 perspective point of view on the center of the scene activate deactivate the projection perspective CTRL F3 New 3D View Open a new 3D Viewer Toggle viewer
3. Position the center of rotation of the cf section 1 3 4 based point of view on the camera re perspective establish the default point of view CTRL F4 Close 3D View Close the current 3D view window before applying this shortcut at least one object in the window to close must be selected FS Refresh Refresh the display H ES S l Toggle sunlight Turn on off the global light source cf section 1 3 4 mE Toggle custom Turn on off the secondary light source cf section 1 3 4 light ME return to the default display ae eee 3D view CTRL O 0 Open Open an object from a file an object from a file cf section 2 1 1 section 2 1 1 nec a list Remove the selected list EE ae the application 23 Chapter 2 Functions In this section we will describe the functions accessible through the main menu They are arranged by sub menu 2 1 File Menu 2 1 1 Open file Fis ker 24011 KO Lezenec F e Ke 24 11 erie IC spe Fr Fr File ke 210 52 18 m Testata File dde 06112 37 rre 3850 bur TBMBbnFRe 10111 L Opera ZI1M bin 24 MB bnFie Lol 50 41 Fdo pee erer 2M pev 256 640 bn Pisa of twee DIN Clsdcompare binae bin Camel Figure 2 1 File selection interface Loads a 3D entity into memory via the standard interface figure 2 1 Notes Keyboard shortcut CTRL O df opening the file is successful the entity will be automatically displayed in the acti
4. and point cloud filtered by HPR bottom To calculate the occlusions using this function the 3d viewer must be using the projection perspective cf section 2 4 4 Otherwise an error message will warn the user and ask them to activate projection perspective The user must then choose the octree level to be used by the function figure 2 37 The octree level speeds up the calculation of the convex envelope a rather large structure by reducing the number of points used via subsampling The higher the level the finer the calculation but it will take longer Once the filtering is done it is only valid for the current camera position give or take a small amount The tool must be restarted to update the filtering for each new point of view Warning the points hidden by this method cannot be re displayed ad hoc for the moment In the meantime one must use a workaround activate the tool to manually segment the cloud the scissors icon see section 1 5 1 which re calculates the visibility information for each point and then quit this mode 2 6 3D Views Menu 57 HPR Level Figure 2 37 Interface to choose the octree level 2 6 3D Views Menu 2 6 1 3D Views gt New Opens a new 3D view window Remarks Keyboard shortcut CTRL F3 The name of the new 3D view is named as a function of the number of views opened since starting CloudCompare If n 3D views have been opened in the current session then
5. by hand notably with the assistance of the and buttons minimize this window and reduce this window respectively available at the top right of each window It is also possible to navigate between the different open views using the commands Next and Previous on the 3D Views menu or accessing a context directly by clicking on its name in the 3D Views menu To close a 3D view window click Close in the 3D Views menu after selecting the context or directly via the X button on the window Finally to change the window in which an object appears select the object and then in the properties window choose the destination window in the Current Display dropdown list Properties B Name Cloud 0 Entities Visible Colors Scalar Field Current Display Cloud Scalar Field 3D View 2 Number 3D View 3 Current 3D View 4 Color ramp SF Scale 1 3 Object Display 9 An object will not displayed in any view if None is chosen as the destination display 1 3 3 Interactivity You can change the 3D point of view in a display by using the mouse Figure 1 2 Mouse commands L select rotate M zoom R pan L left click SELECTION click on an object in a 3D display to select it L held down ROTATION hold the left button down and move the mouse to perform a rotation around the current center of the display cf sections 2 4 4 and 2 4 5 M scrolling ZOOM scroll the mouse will forward
6. in the cloud average height the empty squares are filled with the mean Z altitude of all the points in the cloud maximum height the empty squares are filled with the maximum Z altitude of all the points in the cloud This function creates two files located by default in the binary folder of CloudCompare height grid image tiff the 2D raster image coded in 256 shades of grey corresponding to the Z altitudes projected in the squares of the grid height grid text file txt the height data of the grid in ASCII format easily read by a program or script Height grid generation xj grid step 5 00 type of projection maximum height sl Fill empty cells with average height 2 Files are gerenated by this Function height arid image tiff height grid Text Pie Eet Figure 2 18 Interface for configuring the height grid generation tool See figure 2 19 for an example of the results produced by this function 38 Chapter 2 Functions Figure 2 20 Configuration interface for the coarse registration between two entities 2 3 3 Tools gt Registration gt Align This function coarsely aligns two point clouds using the 4 points Congruent Sets For Robust Registration algorithm by Aiger Mitra and Cohen Or Siggraph 2008 A first input box top left lets the user specify the two clouds and their respective roles Model and Data The Model is the cloud of reference which does not move
7. see section 1 1 The method graphically shows the adequacy between the calculated law white line and the histogram of the scalar field in a window that appears at the end of the calculation see figure 2 26 The values of the parameters of the law are displayed at the top of this window CloudCompare returns lastly in the console the x2 distance between the estimated distribution and the values of the scalar field MN Cloudi ampare Histogram Peer ii al mean IE SIU sion ZE Figure 2 26 Example of the automatic estimation of the parameters of a normal fit for a scalar field 46 Chapter 2 Functions Remark the parameters of the law thus estimated can typically be used in the local statistical test function see section 2 3 9 which permits a point cloud to be filtered based on the distances to a reference cloud or mesh 2 3 9 Tools gt Statistics gt Statistical test Chi Test Gauss Xx mu 1 500000 sigma 40 000000 ER pi Chit 0 000500 E Neighbors 16 Figure 2 27 Example of an automatic estimation of the parameters of a normal law for a scalar field This function central to CloudCompare lets you apply a local x2 test on a point cloud that has a scalar field The x2 test is applied to each point from the histogram of scalar values of its n neighbors n being one of the parameters of the algorithm The test compares this histogram with a theoretical distribution with two parameters u and o
8. toward the top to zoom in get closer Conversely scroll down backwards to zoom out move farther away R held down PAN hold the right mouse button down while moving the mouse to perform a translation of the point of view in the plane of the screen By default the center of the rotation of the point of view is located in the center of the scene and the visualization is done according to an orthographic projection without perspective While changing the point of view rotating the viewing vector large point clouds are temporarily sampled in a manner to permit interactive rendering of the movements The graphical displays 3D views are independent from each other and changes in the point of view in one viewer don t necessarily have any repercussions in the others All the same it is possible to synchronize the camera movements so they will be applied to all the windows at once To do this simply check the Camera link option under the navigation window To re establish the normal functioning of the graphical displays un check this box DE Tree E MA CiiDonneesiPa4lPa4 ca Camera Link Properties E 10 Chapter 1 Interface 1 3 4 Display Options The following commands allow the user to modify the display properties Some of them are notably visible via the Viewing tools toolbar 903 a f7ge t d m Change the lighting parameters in the scene details below ei Change the display size of the poi
9. via the interface in figure 2 6 and CloudCompare will determine the level of octree having a number of cells closest to this value The re sampled point cloud will be formed by replacing each cell by a representative point actually the center of gravity of the points present inside that cell Resample with octree Points approx f 000000 Figure 2 6 Interface to select the number of points in the re sampling Note this method is different than the Subsample method cf section 2 2 28 because it creates points unlike Subsample which selects points that already exist in the source cloud 2 2 9 Mesh gt Compute This function calculates a mesh in 2 5D on top of a point cloud via Delaunay triangulation Because the method produces a 2 5D mesh from a 3D cloud we must project it onto a plane before calculating the triangulation So the user has two choices the function is available in two versions Axis aligned plane by default CloudCompare estimates that the altitudes are carried by the principal Z axis and the points are then projected onto the XY plane Best LS plane a more generic approach which projects the points on the best plane by interpolating the cloud in the least squares sense it is suited to clouds that are rather flat with elevations not necessarily along the Z axis 28 Chapter 2 Functions 2 2 10 Mesh gt Sample Points This function randomly samples points on a surface described by a
10. 7 Tools gt Distances gt Closest Point Get 45 2 3 8 Tools gt Statistics gt Compute stat paramS n0nnnnannnnnnnennnnnnnnennnernennrrnnnrennrreernene 45 2 3 9 Tools gt Statistics gt Statistical Test 46 2 3 10 Tools gt Segmentation gt Label Connected Components 48 2 3 11 Tools gt Segmentation gt K Means sees nennen nnne nnns 48 2 3 12 Tools gt Segmentation gt Front propagation ansanneeannoennnnnnnnennnnnrnnnrrsnnrennrnnenenne 48 ZA DISPlay MED iccinideduzsdenveude ev ie a aaa abu b eR eno Re hene oh E baie ddadad 49 24 14 Display o o E 49 242 Display REMOS 49 Li Display TestFrame EE 49 2 4 4 Display gt Toggle Centered Perspective ccoooccccconnccconnncconnnccnonconacononanononanononanononas 50 2 4 5 Display gt Toggle Viewer Based Perspective cccoonccconccconcnccoccnconcnconcnonacoconnnonnncnnnnnnns 50 240 Display gt Render to Ple ita 50 2 4 7 Display gt Light and Materials gt Set Light and Materials 91 2 4 8 Display gt Light and Materials gt Toggle sun jobt 52 2 4 9 Display gt Light and Materials gt Toggle custom jobt 52 2 410 Display gt CONSO cd 53 Zas OS En c SE Zola Grein EE 53 2 5 2 A abetted aati tis ne sane nnno ditt dnas 55 2 6 3D Views Menu of 2 0 1 IES NOW ed of 2627 9D VIGWS Gl OSC sacra sms Od
11. Cloud FO Piara fee phan Plugin name IECH Plugn global enkghingl Cost ompa stated Ctif itor iLoad 10009 points 13700 Faces Disi em Load POS metidos keane Open 4 1 Application menu bar This bar provides access to the functions available in CloudCompare Sometimes as a function of the selected objects certain menus may not be available For example the Edit and Tools menu are inactive if no object is selected visible in the screenshot above 2 and 3 Toolbars The different icons visible on these toolbars provide quick access to selected functions These functions are also available through the appropriate menus in the menu bar 1 The toolbar 2 provides functions related to processing of objects while the toolbar 3 is purely for changing the visualization parameters 4 Navigation Tree This area provides a tree visualization of the set of objects imported or created by the program Objects can be selected by clicking on them in this area Subsequently we will use the term list for the set of elements grouped under the same tree structure on the screenshot above we can see two lists 5 Current object properties window Ifa single object is selected this area lets you view and edit some properties of this object 6and 7 3D views These areas let you visualize the objects available in memory in 3D In CloudCompare objects can be visualized together in a share
12. The octree is a structure designed to speed up the processing of spatial data It is a recursive and hierarchical slicing of space decomposing the space into cubes In most cases CloudCompare constructs one of these structures over a point cloud in order to accelerate its functions This type of octree is especially fast to construct From a general point of view the octree is defined by its subdivision level The first level level 0 is the smallest cube that entirely contains the point cloud At level N 1 the octree is built by dividing each of the cubes in level N into 8 sub cubes of the same size in practice we only store sub cubes that contain at least one cloud point It is important that the user understands the principle of this structure because it is a central part of CloudCompare The higher the octree level is raised the higher the number of cubes to process potentially 8 2 cubes at level N For instance though it is very improbable we could have up to 2 2 097 152 cubes at the 7th level of an octree At level 10 we have 2 cubes which would be a little over a billion again this is very unlikely especially since it would require at least as many points In practice many cubes are empty and are not stored in memory resulting in a structure that is generally relatively compact Finally CloudCompare uses a special coding for the octree which ensures that its size is always proportional to the
13. data sets accounting for the visibility of the scanner for each data set graphically mapping the PCV Portion of Visible Sky onto the point cloud which improves the display of point clouds on the screen Even though CloudCompare is able to manage triangular meshes or sets of 3D polylines these entities are primarily to CloudCompare point clouds sets of vertices that have particular structures in addition to numerous other structures octree kd tree colors normals scalar fields calibrated photos etc The user is therefore advised to keep this aspect in mind while using CloudCompare and should in particular pay attention to the role of each type of 3D entity in the processing tools offered by this software 0 1 License The CloudCompare software is comprised of the two following programs the CClib library the qCC executable which uses the CClib library Installing and using the two component programs of the CloudCompare software signifies that you accept the terms and conditions of their respective licenses Version 2 dated March 2007 and previous versions of the two component programs are the property of EDF and TELECOM Paris Tech License for the CClib library The CClib library is distributed under the GNU LGPL GNU Lesser General Public License as published by the FSF Free Software Foundation at http www gnu org licenses gpl html License for the qCC executable The qCC program is distributed
14. first will not appear to do anything but the next ones will add line segments connecting the current location to that from the previous click A right click will close the 1 6 Progress bars 19 contour necessary to perform the selection Also any right click performed after closing the contour will reset the contour The next icon corresponds to a selection of points inside the contour The outside points are hidden and the segmentation is paused To do this the contour must be closed The next icon corresponds to selecting points outside of the contour The points on the interior are hidden and the segmentation is paused To do this the contour must be closed The next icon resets the selection completely all the points become visible again The second to last icon validates the trim and quits The last icon cancels the cut and quits 1 5 2 Manual rotation translation This tool is accessible via the ch icon It applies a rotation or a translation manually to the selected entities When it is active 3 icons appear at the top right The first resets the transformation applied to the selected entities The second confirms the transformation and quits The last cancels the transformation and quits The transformation is done interactively with the mouse using the same conventions as for transforming the point of view cf section 1 3 3 The left button allows you to turn the entities around
15. mesh see figure 2 7 Figure 2 7 Illustration of sampling points on a mesh A new point cloud is created The user can specify via the interface the total number of points desired ora density per unit on the surface Be aware that the surface is expressed in the same units squared that is used for the coordinates of the mesh points To get the total surface area of the mesh call the function Mesh Measure surface section 2 2 11 E Points Number C Density 1000000 22 10 0000 A ome Figure 2 8 Interface to choose the parameters for sampling the points of a mesh 2 2 11 Mesh gt Measure Surface Calculate the surface area of a mesh This surface is expressed in the same unit squared that is used for the coordinates of the vertices of the mesh 2 2 12 Mesh gt Scalar Field gt Smooth Apply a smoothing of a scalar field that is associated with a mesh by using the topology of the mesh The scalar value assigned to a vertex is replaced by a mean of the values at the neighboring vertices 2 2 Edit Menu 29 smooth x2 r dl enhance x2 E f Vs N t A P Figure 2 9 Example of results obtained with the smooth and enhance options on a scalar field over a mesh Remark this function is much faster than the Gaussian filter section 2 2 19 for an approximate result but on one hand it requires a mesh associated with the cloud of points which correspond
16. of the current context In theory the more triangles or points to display the slower the frame rate will be Between 10 and 20 images per second a human perceives the display moving in discrete steps above 24 fps the display is perceived as moving fluidly If the refresh rate is too slow it is advised to limit the number of objects to render by only displaying the objects of interest 1 4 Properties of Objects 11 Toggle viewer based perspective switch to a perspective projection and position the center of rotation of the point of view on the camera itself To re establish the center of rotation to its original position in the center of the frame click once more on this command Toggle centered perspective switch to a perspective projection and position the center of the rotation of the point of view on the center of the observed scene To reset default projection parallel orthographic click once more on this command Light amp Materials Toggle sun light turn on off the global light source Light amp Materials Toggle custom light turn on off the personalized light source which can be moved by holding down the CTRL key while panning with the cursor over the screen Note shadow effects are not visible unless at least one light source is active It is also possible to change the individual display color of a point cloud The color is only visible if no scalar fields are actively displayed cf section 1 4 2 2 To change the co
17. on which the program will align if possible the Data cloud which will be moved if necessary As the alignment calculated is a rigid alignment only translations and rotations can be applied to the Data cloud Several other parameters must be entered for optimal use of this method Sampling this area relates to the prerequisite subsampling of the point clouds This can give a perceptible improvement in the efficiency of the algorithm Indeed tens of thousands of points are generally sufficient to obtain a good registration while the complexity of the algorithm increases rapidly with the number of points The user should always seek to minimize the number of points taken into account restarting the algorithm with more points if needed Here are the parameters of the sub sampling 2 3 Tools Menu 39 Method subsampling method see section 2 2 28 Select None to not sub sample Model a slider and or a variable field allows the user to choose the number of points kept for the reference cloud Data same as above a slider and or a variable field allows the user to choose the number of points kept for the adjusted cloud Registration parameters this zone corresponds to parameters of the registration algorithm itself We will explain these parameters in detail Number of trials The algorithm proceeds by successive iterations and only retains the trial that produces the best result This field lets you choose the number of trials to perform T
18. see the option closed mesh which is active by default In the case of a point cloud one must pay attention that the resolution is not too large otherwise holes could appear between the points during the internal rendering this is simply due to the fact that the density of a cloud is limited and that for a sufficient level of zoom we always observe empty zones between the points x dosed mesh 360 mode Light rays 256 Resolution 1024 Figure 2 35 Configuration interface for the PCV rendering 2 5 2 Plugins gt HPR The Hidden Points Removal function as its name indicates filters the selected point cloud to only keep the visible points corresponding to the visible surface from the current point of view The points considered to be thus masked are therefore hidden The result strongly depends on the point of view The notion of visibility for points in a cloud is relatively complex to estimate In reality it is very unlikely that a point will be actually hidden by other points in a cloud because that would require perfect alignment between pairs of points or a density so high that the points are practically in contact This function approximates the notion of visibility using a calculation of the convex envelope It is based on the article Direct Visibility of Point Sets by Katz Tal and Basri SIGGRAPH 2007 56 Chapter 2 Functions Figure 2 36 Complete point cloud top
19. that the two clouds be more or less aligned This function cannot align clouds that are positioned and oriented haphazardly Its role is primarily to refine the registration of two clouds that are roughly aligned The Align function described in section 2 3 3 can ensure that the clouds are approximately aligned and the Register function can be used on the resulting clouds In the top zone Model and Data of the configuration window the user may interactively assign roles to each entity The Model is the reference cloud which does not move and the Data cloud is the one that will be registered and will move if necessary To help the user CloudCompare colors the entities in their display Model in red and Data in yellow in the same way as in the interface for choosing roles when calculating distances see paragraph below The Swap button switches the roles if needed The lower part Registration parameters corresponds to the parameters of the algorithm itself In detail Stop criterion the user chooses either a fixed number of iterations this prevents an excessively long calculation times but does not guarantee the quality of registration or a minimal decrease in the error between two iterations to justify more iterations which guarantees a better quality but potentially may take a long time Enable furthest point removal heuristics adapted to align entities that differ only slightly while CloudCompare is made to compare p
20. the cells at different levels To call this function you must have selected a single 3D entity 2 3 11 Tools gt Segmentation gt K Means This function is no longer part of version 2 1 of CloudCompare 2 3 12 Tools gt Segmentation gt Front propagation This function is no longer part of version 2 1 of CloudCompare 2 4 Display Menu 49 2 4 Display Menu 2 4 1 Display gt Full Screen This function displays the main window of CloudCompare in full screen mode In this mode the entire screen is taken up by the application For the sake of visual comfort the title bar of CloudCompare and the Windows task bar are no longer accessible though this also hides the window manipulation functions minimizing changing the active window etc To return to normal view click again on the Full screen command or use the keyboard shortcut F11 Note it is still possible to change the active window even when in full screen mode by holding down the ALT key then tapping TAB until you have highlighted the desired application cf figure 2 30 When ALT is released Windows will activate the selected application window This command works in all applications CloudCompare 3D View 1 Figure 2 30 Windows application selector Keyboard shortcut F 11 2 4 2 Display Refresh The Refresh command refreshes the displays in the 3D viewers Keyboard shortcut F5 2 4 3 Display gt Test Frame Rate This fun
21. to the vertices of the mesh and on the other hand the size of the smoothing kernel cannot be changed 2 2 13 Mesh gt Scalar Field gt Enhance Applies a contrast enhancement on a scalar field associated with a mesh by using the topology of the mesh The scalar value at a vertex is modified to augment the contrast between it and the values at neighboring vertices This function is the inverse of Mesh Scalar Field Smooth section 2 2 12 2 2 14 Sensor Ground Based Lidar Show depth buffer This function is no longer part of version 2 1 of CloudCompare Display the depth map associated with a sensor 2 2 15 Sensor Ground Based Lidar Export depth buffer This function exports the depth buffer associated with a sensor as a text file It is applied to a cloud that contains the selected sensor it is necessary to first select a sensor entity The user is invited to specify a filename under which all the information relating to the depth map will be saved see section A 1 3 2 2 2 16 Sensor Create This function creates a sensor a virtual representation of a laser scanner associated with a cloud 30 Chapter 2 Functions In general the user wants to re create the sensor that acquired the point cloud after the fact In particular this allows applying smart filtering when calculating the distance between two point clouds cf section 2 3 5 r ae order theta phi GS Position center Orien
22. un UE 2f 229 AMES COMPU Eaa a 27 2210 Mesh gt sample FONS os 28 2 2 11 Mesh gt Measure Gurtace 28 Table of Contents 22412 Mesh Scala Field MOON ss 28 2 2 19 Mesh Scala Feld REA eege 29 2 2 14 Sensor gt Ground Based Lidar gt Show depth butter 29 2 2 15 Sensor gt Ground Based Lidar gt Export depth buffer 29 22 106 EE 29 22 ENEE MOYA O ce C E RUM ee 31 2210 Calar FICS Ia 31 2 2 19 Scalar Fields gt Gaussian kiter 32 2 2 20 Scalar Fields gt Filter by Value n nnnnnannnnnennnenennnsnnnnnsnrnnrnrensrnrresrrrrnsnrersrnrrrserrrennne 33 2 2 21 Scalar Fields gt Dtterence 33 2 2 22 Scalar Fields gt Multiply ooocccccoccocccccococccoconcnnocacononononocononononnnononnnononnnnnanononos 33 22 29 oGalar Fields gt Conven tO RGB EE 34 PAM EEG Lu 34 VA CMM UTE rrr 34 2 2 25 he EE A ee eee d e ete bububu b onobau dS osobot 34 22 A OO AAA A PA Pr OQ A 34 LADO SUIS ina 35 E Ber le EE 35 297 TOO E e TT 36 2 9 1 TOOS Pro Je cion E le dude AI AAA 36 2 3 2 Tools gt Projection gt Height Grid Generaton 37 2005 Tools gt Registration AlN cuisson ivo qun a Na an dad dac ddsb dic 38 2 3 4 Tools gt Registration gt Register csse nennen 40 2 9 0 Tools Distances Cloud Cloud dista 42 2 3 6 Tools gt Distances gt Cloud Mesh dist cooocccccocncccocnnccocnnoconncononoconacononaroconacononos 44 2 3
23. under the GNU GPL GNU General Public License as published by the FSF Free Software Foundation at http www gnu org licenses Igpl html EDF and TELECOM ParisTech grant the user the rights to install and to use the CloudCompare software after having downloaded it from the site http rd edf com The software is provided as is with no explicit or implicit warranties The authors take no responsibility for any harm caused directly or indirectly The user assumes all risks and responsibilities with regard to the quality of the CloudCompare software and its use 0 2 Installing the binary 3 0 2 Installing the binary CloudCompare works under the Windows 2000 and Windows XP operating systems It is no longer being ported to Linux The binary distribution of CloudCompare does not have an installer Simply decompress the zip archive containing the executable and the DLLs Listed below are the files exe or dll that you will find after expanding the archive For nc version 2 0 IHM based on FItk Flu CloudCompare exe primary executable CC DIl dll CClib library cv100 dll DLL of the openCV library used for image processing cxcore100 dll DLL of the openCV library used for image processing Devil d DLL of the Developers Image Library which permits importing numerous image formats glew32 dll DLL of the OpenGL Extension Wrangler Library highgui100 dll DLL of the openCV used for image processing I
24. via the interface shown in figure 2 32 We see three sections The first frame Light changes the lighting parameters The user can define a color for each of the three components of light ambient specular and diffuse The Ambient component is the constant lighting that bathes the scene at night for example the ambient component is black no ambient light The Diffuse component defines the color reflected by objects independent of the position of the camera Finally the Specular component defines the color re emitted by objects toward the camera the brighter this component the glossier objects appear the darker the more they appear matte The second frame Default materials configures the colors applied by default to meshes or point clouds In the case of meshes the user can define a different color for each surface front back The Points button defines the color of points in point clouds the same as the command in section 2 2 1 The final frame Others modifies some general display options Background sets the color of the background of the 3D viewer The background always appears in the form of a gradient from the 52 Chapter 2 Functions chosen color to black Bounding box changes the color of the box encompassing the selected objects Finally Text changes the color of text in the 3D viewers Each of the buttons just described opens the color selection interface described in section 2 2 1 At the bottom of th
25. 08 Raphael Marc research engineer at EDF R amp D along with elements graciously provided by Florent Duguet then a PhD student at Telecom Paris CloudCompare has been mainly used to deal with interiors of power plants mapped by EDF surveyors using laser scanners and it has occasionally been used to deal with readings from dams or cooling towers This software can among other things calculate the local distances between two dense point clouds figure below left filter out the measurement noise of the laser scanner to identify true differences figure below center identify the individual objects or parts of objects that differ between two data sets figure below right The originality of CloudCompare comes from many aspects the data structures used an octree that allows large point clouds many millions of points in 3D to be stored in memory and displayed as well as allowing the differences between two 2 Table of ContentsIntroduction large data sets to be calculated rapidly that is to say in a few seconds as well as a Kd tree which is used to quickly align two point clouds two novel distance measurements between point clouds a precise measure based on a Hausdorff distance distance to the nearest neighbour and a very fast though less precise measure based on the chamfer distance measurement noise filtering accounting for the sampling differences between the compared
26. 3 4 a first approximate distance calculation between the clouds chamfer distance calculated via the octree is automatically performed When this is done the upper part of the display in figure 2 24 Approx results shows various distance statistics that can help configure the refined distance calculations This information includes Min dist approximate minimal distance Max dist max approximate distance Mean dist approximate mean distance Sigma standard deviation Max relative error maximum relative error of the approximation expressed as a function of d the distances because this error is dependent on the actual distance between the points and generally decreases rapidly when d grows which means that the approximation of the minimal distance is generally very bad but that of the maximal distance can be reliable enough The user can lastly display the histogram of approximate distances by clicking the 8 icon but these are usually not very detailed given that they are chamfer distances calculated via the octree 2 3 Tools Menu 43 28745 7 1553 54 3154 15 Max relative error 3 77496 57116 4 d amp d Figure 2 24 Configuration interface to calculate the distances between two point clouds The lower part Precise results configures the precise distance calculation The user can choose the following values Local model indicates that a local model will be applied to the ref
27. CloudCompare Se CloudCompare User s Manual for version 2 1 Authors DGM AB RM Translator Rosemary Le Falve Funding for Translation Provided by Vehiclemetrics Inc www cloudcompare net Table of Contents INTFOQUCHION arras 1 S MENU I IR ROM m P daA TT 2 0 2 stalling NS DNI Yara aran attain en n deme mte 3 Chapter 7T Jet EE 4 a D Me WR ale Te EE 4 12 Ee OD eS o dd de ae do im idis 5 ho Navigalom Me licuado 2 122 EE 6 19 ee TE a T ns IA T 1 9 2 Multiple GD VIEWS aa li 8 A DS 9 159 MOS NOUNS sist 10 jr EM o e a 11 ES E elei ei Beie E TD E a 11 AZ OMe CIMAS aar AAA ARA 12 1 4 2 1 Common PNODCMICS dia te Lois de 12 t422 SC OS a Meets ss 12 1 4 3 Meshes groups of meshes 15 VAL VCS mt 16 WAR DES CPO MOM Meet e bdo de ab dede 16 A BT EE 17 1 5 Interactive modification of enttes 18 1 5 1 Manual segmentatlon EE 18 1 5 2 Manual rotatonitranslaton 19 MO PONES S DAS AAA Na 19 der KO seite 20 19 gt FRO VIOO AG SIOFICUIS an osos 21 Chapterz RUNCUONS iaa 23 2e I A a mee 23 XE EO Burr 23 ZN UAI rH aaseudes 24 22 elt a 24 EE e ee Rn e UE 24 2 2 2 COOLS gt KEE gedeien EES 25 22 E ee HEIO TRAIN erene aaa aa aa 26 224 ANERE a ce Ee dee 26 2 229 A IIS 26 2 2 6 Normals gt Resolve drechon 26 221 AA 2f LO Octree gt
28. During the interactive phases mouse movement in the 3D viewer cf section 1 3 3 the camera will turn around the objects in the scene If clicked again the command re establishes the display mode to orthographic projection Keyboard shortcut F3 2 4 5 Display gt Toggle Viewer Based Perspective This command toggles the 3D views between having a point of view centered on the camera and a point of view centered on the scene The first time it is called it positions the center of rotation on the camera itself This mode is associated with a projection perspective During the interactive phases movement of the mouse in the 3D viewer cf section 1 3 3 the camera turns around itself maintaining its position If the command is activated again it will re establish the default display centered on the scene and with an orthogonal projection Keyboard shortcut F4 2 4 6 Display gt Render to File Creates a screen capture of the active 3D viewer in a BMP file This function lets you specify a zoom factor via the window presented in figure 2 31 2 4 Display Menu 51 Snapshot Zoom 3 Zoom Factor 1 Figure 2 31 Zoom window for capturing screenshots of the 3D display 2 4 7 Display gt Light and Materials gt Set Light and Materials Lights amp Materials Apply Reset Cancel Figure 2 32 Lighting configuration window This function sets the parameters of the lighting in CloudCompare
29. LU dll DLL of the Developers Image Library which permits importing numerous image formats ILUT dll DLL of the Developer s Image Library which permits importing numerous image formats libguide4O dll allows linking to Intel s DLL Math Kernel Library xerces c 2 7 dll C library for parsing XML Remark it is possible that running CloudCompare 2 0 will require the presence of the msvcp7 1 dll and msvcr71 dll DLLs These DLLs are part of the Microsoft C Runtime Library available for download on the Microsoft web site For version 2 1 of CloudCompare IHM based on Qt 4 qCC exe primary executable A CC DIOU CClib library Freelmage dll a library from the open source project Freelmage which enables reading and writing of the latest image formats mingwm10 dll DLL for the minGW compiler which allows multi threading QtCore4 dll DLL required by Qt QtGui4 dll DLL required by Qt QtOpenGLA dll DLL required by Qt And the optional components in the plugins directory QHPRPlugin dll plugin for the Hidden Point Removal function qPCVPlugin dll plugin for the PCV portion of visible sky rendering function Chapter 1 Interface 1 1 Main Window The main window of CloudCompare has the following sections IO i di Toss QMplay Puge JO Yes Hebi Ke E X 19 e e BG Tres vi A CijDonneesPadiPas ca amp e M Mesh Group s v Bh CibonmeesiPaniPat ca Gel C23
30. al and vice versa 2 2 25 Fuse Fuse two or more selected entities Note the fused lists are minimized when the function is executed All characteristics of the 3D entities are conserved Lists that do not possess a particular characteristic will be given default elements no normals white color etc 2 2 20 Translate Apply a translation to the selected entity or entities The user supplies the three components of the translation X Y and Z via an interface 2 2 27 Multiply Multiplies the coordinates of the points of the selected entities by constants The user chooses the 3 coefficients to multiply each component along its axis fy fy fz via a dialog box fx 11 000000 fy 1 000000 ES f 1 000000 _ Cancel Figure 2 14 Dialog box for multiplication of coordinates 2 2 Edit Menu 35 Cloud sub sampling Figure 2 15 Configuration interface for sub sampling of clouds 2 2 28 Subsample This function creates a subsample of a point cloud Different methods are now available The choice as well as the configuration is accessible via the dialog box shown in figure 2 15 Random random subsampling Space spatial subsampling the resulting local density is constant and configurable Octree rapid subsampling via the octree keeping one point per cell of the octree at the given level of subdivision Remarks The subsampling differs from re sampling cf section 2 2 8 in
31. ar field to the reference cloud The choice of a local model is not possible because the reference entity here is already a mesh 2 3 Tools Menu A5 2 3 7 Tools gt Distances gt Closest Point Set This function calculates the closest point in the compared cloud for each point in the reference cloud The set of these closest points forms a new cloud Closest Point Set or CPS Remarks To call this function you must have selected exactly two point clouds We see again the generic role choice interface cf section 2 3 4 which lets the user choose the cloud from which to extract the CPS points compared cloud and which is the reference cloud The result is a cloud which has exactly the same number of points as the reference cloud and in which each point is a member of the compared cloud by definition By construction there may be duplicate points This result is used for example by the alignment algorithm between two point clouds cf section 2 3 4 2 3 8 Tools gt Statistics gt Compute stat params This function calculates the parameters of the chosen statistical law Gauss or Weibull from the values of the active scalar field of the selected cloud The function typically returns the mean and standard deviation of the current scalar field if the Normal law is used or the parameters a b for a Weibull law in which case CloudCompare also displays estimates of the mean and standard deviation in the console
32. atically by a program such as RealWorks Mensi A 1 3 2 Depth Map Export file An export in ASCII of the depth map associated with a sensor lt may be generated by the function Sensor gt Ground based Lidar gt Export depth buffer section 2 2 15 Example file ground based laser scanner txt H CLOUDCOMPARE DEPTH MAP Associated cloud Cloud 0 dPhi 0 005000 0 383052 0 319331 dTheta 0 005000 1 626588 0 137948 pMax 78823 398438 Il L 353 ILH 141 UM 0 000000 0 000000 18132 496094 1 000000 0 000000 15145 963154 associated cloud name as disp in CC horizontal angular step vertical angular step max depth number of horiz pixels number of vert pixels 1st pixel coordinates i j and depth z 2nd pixel coordinates i j and depth z 352 000000 140 000000 132135 321542 L H pixel coordinates i j and depth z
33. c interface for choosing roles see section 2 3 4 The calculation is signed and the order for subtraction to take place their order must be defined field gis field fields The first cloud cloud A will receive a new scalar field corresponding to this difference Remarks The term identical is used loosely It suffices that all the points in cloud A are found in cloud B That is cloud A must be a subset of the points in cloud B The algorithm creates a new type of scalar field Diff which is signed 2 2 22 Scalar Fields gt Multiply This function is no longer part of version 2 1 of CloudCompare This function multiplies the scalar fields of two clouds for the selected points The two clouds must have the same number of points The scalar field of the first is updated with the result of the multiplication Remark to call this function one must select two and only two point clouds with scalar fields 34 Chapter 2 Functions 2 2 23 Scalar Fields gt Convert to RGB Assigns a color to each point from the current scalar field values and the current display parameters If the entity already has been assigned a color it is possible to remove the existing color or to mix the two values CloudCompare asks the user when needed 2 2 24 Clone Creates a new 3D entity that is point for point identical to but independent from the selected entity A modification to the cloned object has no impact on the origin
34. ction estimates the refresh rate of the displays in CloudCompare measured in frames per second fps The test will take about 10 seconds and is characterized by a rotation around the objects visible in the active 3D display Once the test is finished the result is displayed in the information zone of the 3D viewer Note this rate depends directly on the number of triangles and points to display For reasons of visual comfort it is preferable to have a refresh rate of about 25 fps or higher cf section 1 3 4 50 Chapter 2 Functions Keyboard shortcut F12 2 4 4 Display gt Toggle Centered Perspective In the display of objects the projection defines the manner in which 3D objects are drawn to the screen in a 2D visualization CloudCompare offers two types of projections parallel orthographic the points are projected orthogonally on the image plane The field of vision corresponds to a cylinder perspective the points are projected toward a single point that is not part of the image plane The field of vision corresponds to a cone The image plane can be considered to be the visualization screen In CloudCompare the Toggle Centered Perspective command lets you toggle between the orthographic projection view which is the default display view and the projection perspective display When this command is activated the center of rotation of the point of view is automatically placed in the center of the observable scene
35. d 3D view or in separate windows in the screenshot above two 3D viewers are open 1 2 Available Objects 5 8 Console This zone contains the log of information generated by the execution of CloudCompare typically supplementary non essential information produced by the algorithms lt can be displayed or hidden via the Console command on the Display menu 9 Status bar Quick help messages related to certain functions of CloudCompare are displayed here When you position the mouse cursor over a control button in the menus the related help message will appear in this bar However not all the controls necessarily have such a message All the entities available in CloudCompare can therefore be accessed via the navigation tree 4 and possibly in the 3D view s 6 7 It is possible to move or remove the different toolbars To do this simply grasp the manipulable part situated at the end of the bar by clicking on it then move it while holding down the mouse button UE ei ll The bar can be left floating or docked in a new position in which case the neighboring toolbars will rearrange themselves The toolbars can also be displayed or hidden via the Display Toolbar submenu the checked items correspond to the visible toolbars unchecked items are hidden Similarly it is possible to move the navigation window and the properties window by grasping their title bars and dragging them to the desired position In the rest of t
36. d the average number of points per cube as small as possible therefore not too low a level Finding the right level can require some experience 1 4 Properties of Objects 17 Figure 1 4 Cells at different levels of octree construction level 1 left 3 middle and 6 right 1 4 4 2 Display The octree is inseparable from the point cloud to which it is attached It is an abstract structure available in the CloudCompare interface only as a visual The display proposed by CloudCompare only allows visualization of one level at a time You can change the level of display of the octree by raising lowering the Display level field in the Octree section The display levels can be between 1 and 10 Level O can not be selected it is never used in practice not even for calculations because a level O octree is simply the entire cloud The display might slow down considerably at a certain level of subdivision a function of the capabilities of the computer on which the program is running as a result of the number of elements to display In this case choose lower display levels as much as possible The octree can be visualized in different forms Wireframe only the edges of the cubes are represented Points each cube is represented by the center of gravity of the points contained within it Plain cubes the surface of the cubes is fully displayed To change the display of the octree select a mode in the Display Type dr
37. e current window 2 6 7 3D Views gt Previous This command activates the previous 3d view in place of the current Note the previous window is the last one created before the current window 2 7 Help Menu 2 7 1 Help gt Help Display the user documentation for CloudCompare Keyboard shortcut F1 2 7 2 Help gt About Display the information window for the current version of CloudCompare cf figure 2 38 2 7 Help Menu 59 DIT ES CloudCompara V3 is an open source project under GPL license JE has originally been developed during a PhD study between 2003 and 2006 by Daniel Grardeau Montaut The PhD was financed by EDF and tutored by Telecom Paris TSITI Lab CloudCompare To be continued Figure 2 38 Information window 2 7 3 Help About plugins Display the plugin information window figure 2 39 This window displays the available plugins in the form of a tree The folders in which CloudCompare searches for plugins are displayed at the top of the window t Plugin Information ajx Found plugins looked in CS EDF CloudCompare gC Cl plugins h 2 HgHPRPlugin dll EB E CloudCompare Plugin D name HPR Plugin Hidden Paint i version 1 gPCVPlugin dll E 9 CloudCompare Plugin i E name DCH Plugin global enlight n E version 1 Figure 2 39 Plugin information window 60 A Appendix A 1 File Formats A 1 1 Recognized 2D 3D file typ
38. e window are buttons that let you apply the selected parameters OK and Apply re set all the values to their defaults Reset or leave the interface without making modifications Cancel 2 4 8 Display Light and Materials Toggle sun light Turns the principal light source on and off You must have at least one light source active to see the lighting effects shadow color Keyboard shortcut F6 2 4 9 Display gt Light and Materials gt Toggle custom light Turns the personalized light source on and off Figure 2 33 Secondary light source custom light This light source unlike the principal light source representing sunlight cf section 2 4 8 is a point light source It appears as a small yellow star around the objects see notes below It has the same properties as the main light source It is possible to move the light source by holding down the CTRL key while making a PAN with the mouse holding down the right mouse button 2 5 Plugins Menu 53 Notes Keyboard shortcut F7 The star only appears in the orthographic projection mode see section 2 4 4 or 2 4 5 Sometimes the star may have an initial position within the object itself 2 4 10 Display gt Console This command displays or hides the console visible by default in the lower part of the main window of CloudCompare cf section 1 1 2 5 Plugins Menu Plugins are extensions that offer advanced functionalitie
39. elect one object while keeping the others already selected click once more on that object while holding down the CTRL key Alternately while holding the SHIFT key down then selecting two objects in the navigation tree only all objects visible between the two selected objects will also be selected A final possibility is to hold down the left mouse button while moving the cursor over the objects to select in the navigation tree The majority of functions in CloudCompare only apply to selected entities Their commands are therefore deactivated until the user has selected the appropriate number and type of objects 1 3 Object Display 7 1 3 Object Display 1 3 1 3D view 2 S C2M Distances O Li m L o L L RW LC LI E r1 0 Figure 1 1 Standard elements of a 3D view The 3D views or graphical contexts see figure 1 1 are the sub windows in which the entities can be visualized They are all arranged in the following manner 1 Scale Provides a reference to estimate distances The scale is displayed using the distance units in which the coordinates are expressed CloudCompare does not use explicit distance units 2 Orientation triad The three displayed axes represent the X axis red Y axis green and Z axis blue The triad indicates the current orientation of the viewer 3 Bounding box This is the smallest parallelepiped rectangle that contains the entirety of the s
40. elected objects The axes of the box are aligned on the X Y and Z axes as per the orientation triad 2 4 Name of active scalar field refer to section 1 4 2 2 for more information about scalar fields 5 Temporary information concerning the display lt may be the current dimension of the context in pixels after resizing the window the type of projection used etc 8 Chapter 1 Interface 6 Color gradient Identifies the values associated with the colors when a scalar field is active and displayed 1 3 2 Multiple 3D views The majority of entities loaded in memory and visible in the navigation tree can be displayed in 3D views Any number of these visualization windows can be created and the entities can each be assigned individually to a different window To create a new 3D view click on the command New in the 3D Views menu or simultaneously type CTRL and F3 A new window will then appear by default it will fill the entire area dedicated to the 3D views To divide the display area between the different open windows different methods exist Partitioning each window occupies a portion of the available space and the windows do not overlap To partition the display click on Tile in the 3D Views menu Cascading Display each window occupies the same portion of the allotted space and the windows all overlap each other To cascade the views click Cascade in the 3D Views menu Resize every context
41. erence cloud to improve the precision of the cloud to cloud distance calculation to a certain extent This technique improves the overall precision but not necessarily the local This improvement depends on the chosen model and has the cost of slowing the calculation the extent of which depends again on the chosen model NONE no local modeling default calculates distances from point to nearest point Least Square Plane local approximation of the cloud by a plane adjusted to least squares less precise but fast 2 5D triangulation local approximation of the cloud by a 2 5D Delaunay triangulation after projection of the points on a plane adjusted to least squares intermediate speed and precision Height Function local approximation of the cloud by a height function of the type z ax by cx dy exy here again after a projection of the points onto a least squares plane best precision but reduced speed Max dist define a distance above which it is not necessary to calculate a precise distance This may greatly improve the calculation s performance in particular on clouds having few zones in common by avoiding calculating the far distances the most costly while their precise knowledge is generally not useful The points in question retain their approximate distances The information displayed in the upper part may be of help in choosing this limiting value Octree level this parameter of t
42. es i id d i K ASCII point cloud point cloud s binary format specific to CloudCompare o 7 7 m Fa o i am ae c bin binary cloud C Hernandez MAROON roma normal pem pa ascii cloud image association Y mesh Stanford mesh Wavefront BL BUB z vp Ea J E AA KAN BE A Points M Meshes RGB Colors Red Green Blue LG Levels of Grey Normals Scalars 2 7 Help Menu 61 A 1 2 Opening and Saving asc txt neu xyz etc O ET D 5 A 1 3 Special Formats CD A 1 3 1 ICM files This type of file associates a point cloud with a VRML file the latter defines calibrated photos camera image file Example file toto icm ZCC ICM FILE Header FILENAME pa4 asc Points file FILETYPE ASC Type of points file IMAGES DESCRIPTOR photo_match wrl VRML file describing the calibrated photos 62 Chapter 2 Functions file photo_match wrl VRML v2 0 utf8 DEF photo1 jpg Viewpoint fieldOfView 0 621379 position 10 5418 15 6091 5 95961 description VANNE PETIT TUYEAU orientation 0 70724 0 37292 0 600618 3 74252 DEF photo2 jpg Viewpoint fieldOfView 0 621379 position 3 9782 21 276 5 95616 description PORTE orientation 0 572629 0 696275 0 432778 2 02135 Header photo 1 I FOV Optical Center Description View vector Header photo 2 letc E _ _ AAA AAA A VRML file like this can be generated quasi autom
43. hat is actually calculated for each point is the weighted average of the scalar values of the neighbors weighted by their distance according to a Gaussian law Given that 30 corresponds to an overwhelming 99 9 it is not useful to consider more distant points The larger the kernel the slower the calculation will be This function is very useful for smoothing the result of a gradient calculation section 2 2 18 but also for a Portion of Visible Sky calculation section 2 5 1 for example 2 2 Edit Menu 33 2 2 20 Scalar Fields gt Filter by Value 4 Max 0 030000 OK Cancel Min 0 000000 Figure 2 13 Configuration interface for filtering points by the value of their scalar field This function segments a cloud by defining an interval of scalar values figure 2 13 A new cloud will be created with all the points for which the scalar values for the active scalar field are included in this interval 2 2 21 Scalar Fields gt Difference This function is no longer part of version 2 1 of CloudCompare Tool for calculating the point to point differences between two scalar fields applied to identical clouds This function requires two spatially identical clouds each with an applied scalar field and calculates the differences between the values of homologous points This algorithm requires selecting two and only two clouds CloudCompare asks the user to define the role of the clouds via the generi
44. he higher the chosen value the longer the calculation will take but the better the probability of obtaining good results It may therefore be necessary to adapt this parameter as a function of the number of points in the clouds to obtain a good alignment in a reasonable time To give a general idea to align two clouds of 5000 points each about fifty tries will produce a decent result in a few minutes about 2 5 minutes depending on the speed of the computer Overlap This parameter which can be between 0 0 and 1 0 corresponds to an estimation of the overlap between two clouds when they are correctly aligned An overlap of 1 signifies that the two clouds overlap completely O signifies that they are disjoint in this case alignment doesn t make much sense A very approximate estimation of overlap is generally sufficient Delta This parameter is an estimation of the mean distance between points in the two clouds when they are correctly aligned This parameter functions like an error tolerance the closer it is to 0 0 the closer we constrain the clouds to be but the likelihood of finding a good solution is lowered In principle if the Delta is zero the program can never find an alignment between the two clouds As a general rule to obtain good results Delta should correspond to the resolution inverse of the density of the reference cloud The interface contains an Estimate button that automatically estimates the parameter based on the mean de
45. he algorithm is normally optimized automatically by CloudCompare but it is possible to override it in cases where the determining heuristic fails 44 Chapter 2 Functions Remarks 2 3 6 This function adds a scalar field of C2C cloud to cloud distances to the reference cloud To calculate the precise distances click the red Compute button Otherwise only the approximate distances are saved All the distances calculated by this function or entered as parameters are expressed in the same unites as the coordinates of the point cloud there is no explicit unit in CloudCompare 2 1 Tools gt Distances gt Cloud Mesh dist Figure 2 25 Example of the results of calculating the distances between a cloud and a mesh This function calculates the distance approximate or exact between a cloud of points and a mesh This function is largely equivalent to the distance calculation between clouds section 2 3 5 except for a few details if one of the two selected entities is a mesh the choice of roles section 2 3 4 is not necessary the mesh is assigned as the reference entity if the two selected entities are meshes the vertices of the compared entity will be used as the cloud lt may be of interest to use the sampling function for points on a mesh section 2 2 10 beforehand to have a better view of the differences between meshes if this is the expected result this function adds a C2M Distances Cloud to mesh scal
46. he point used will be the center of gravity of the cloud the radius of the cylinder or base of the cone and the opening angle of the cone in the applicable case Warning to optimize memory usage this function applies the transformation directly on the selected entity It may be necessary to use the clone tool first see section 2 2 24 Figure 2 17 Example of results a cylindrical point cloud left and its unrolled version right 2 3 Tools Menu 37 2 3 2 Tools gt Projection gt Height Grid Generation This function projects a point cloud onto a flat grid normal to the Z axis An interface figure 2 18 allows the user to choose various parameters grid step size of a grid section expressed in the units of the point cloud coordinates type of projection this parameter may take one of the following two values maximum height Let E be the subset of cloud points that is projected in the i j th square of the grid For each grid square i j we keep the Z altitude of the highest point in Ej average height for each grid square i j we keep the mean Z altitude of the points in Ej fill empty cells with some grid squares may remain empty after projection no cloud points were projected there This parameter indicates the value with which to fill these squares and can take one of the following three values minimum height the empty squares are filled with the minimum Z altitude of all the points
47. his document we will describe the basic functionalities of CloudCompare and describe all the available manipulations that affect the visualization of entities The user can refer to the next part of this document Chapter 2 to get detailed information on the advanced functions 1 2 Available Objects 1 2 1 Navigation tree DB Tree ei E A CAEDF DonneesiPa Pad_casemat a S Mesh group JA CHEDF Donnees PadiPat casemat 5 Cloud 0 Lei E Octree As previously seen CloudCompare displays the set of available objects loaded into or created by this application in the navigation tree window which is located by default on the left side of the main window We can find the following elements there 6 Chapter 1 Interface A group of entities If this corresponds to an open file it will group all the elements loaded from this file and will be named with the source file path A point cloud A simple mesh A group of meshes treated here as a standard mesh we can also find the vertices of these meshes as point clouds within this element type s SO e A structure is i ew Tee mm An object e A photo possibly calibrated Note groups of entities are not technically speaking manipulable entities Typically they cannot be used as entities for CloudCompare functions they serve only as a folder In the usual way the tree can be opened by clicking on the buttons and situated at the left
48. in precise results one must know or measure the distribution of measurement noise as a first approximation it can be modeled as the average noise including the measurement error due to the sensor due to the surface that was scanned to the lighting to the ambient temperature during the measurement to the creation of the mesh in the case of comparing clouds meshes etc The parameters of the statistical distribution can therefore be defined from a priori knowledge but can also be determined from a scalar field typically a portion of a cloud with the function that computes statistical parameters of a scalar field cf section 2 3 8 The algorithm creates a new scalar field named Chi2 Distances This field is assigned to the current point cloud Figure 2 28 Field of initial differences top left statistical filtering top right followed by extraction of points out of the theoretical distribution bottom left and finally the extraction of the connected components bottom right 48 Chapter 2 Functions 2 3 10 Tools gt Segmentation gt Label Connected Components Figure 2 29 Configuration interface for extracting connected components This function decomposes a point cloud into compact sub clouds lf the selected cloud is composed of multiple groups of points sufficiently disassociated distant from the others it is possible to subdivide it fairly simply via the octree This is done in CloudCompare tha
49. in the case of a normal law for example Before specifying the parameters the user should choose the type of theoretical distribution currently the choice is between Gauss and Weibull The result is a new scalar field one value for each point the x2 value which gives an indication of the local concordance between the scalar value and the tested distribution The theory of the x2 test gives us a threshold calculated from the margin of error p x2 final parameter in the algorithm which classifies points as a function of their non adherence to the tested law This law will typically represent measurement noise and we obtain from it a set of points where the distance to the other mesh cloud is not accounted for by the measurement noise From this we have the points that have actually undergone a change and we can ignore points that have not actually changed but have non zero distance measurements due to noise Once the cloud is separated into these two classes one can keep the group of points outside the distribution see figure 2 28 red and segment for example as a function of the relative proximity of the points by extracting the various connected components cf section 2 3 10 Remarks To call this function select a single 3D entity with an active scalar field When adjusting the parameter p x2 it is important to understand that the x2 test only allows us to reject the hypothesis that the values of the scalar fie
50. ion interface as seen for the Set unique method cf section 2 2 1 In either case the user can choose along which of the three principal directions refer to the triad in the bottom right of the display to apply the gradient 2 2 4 Normals gt Compute This function is no longer included in version 2 1 of CloudCompare This function calculates the normals unsigned of a point cloud that lacks normals and can do this via different methods To get a signed field of normals use the Resolve direction function cf section 2 2 6 The two methods currently available are LS obtained by local adjustment according to the least squares method fast method but noisy HF obtained by interpolation of the points by a function of quadratic height more robust but slower 2 2 5 Normals gt Invert Invert the normals for the components corresponding to the tridimensional unit vectors we replace each by its vector opposite Note this usually fixes problems that have to do with the sense direct or indirect of the mesh triangles because the normal is not shared by all programs that process or generate meshes 2 2 6 Normals gt Resolve direction This function is no longer included in version 2 1 of CloudCompare This algorithm resolves the correct sense of the normal vectors associated with a point cloud The resolution is performed gradually by propagating one or more fronts on the cloud using the Fast Marching a
51. it o 57 200 SD VIEWS ee le 57 A Ne nee Ne eee ee ee eee eee 57 2 6 5 3D Views gt CGascade coccoocccccoccncccccncconcncnoncnnnnnnnonnnnnnnnnnnnnnnnnnnnrnnnnnnnnnnnnnonnnnnnnnaninnnninnns 58 don ER I NON m 58 ZO INES O US eniti ideale caet eee 58 CR EM Ee BCE E 58 PNA PE AP ETT 58 Dale AODA AJOU ra e ded dei dedo ded S 58 2 1 3 Help gt About piugms 59 WE cunda 60 AT FIE en A PP 60 A 1 1 Recognized 2D 3D file types 0 nnn0nnnnannnnnnnnannnnnnnnnnnnnennnnrnnrnrrsnrrnnnrrnnrrrenrrnrrresnrrene 60 A2 Opening and gt ads 61 ABL OCCA ON EE 61 s o OE O RE ene 61 5 1 9 2 Depth Map EX DOM INC eremi one duci tacuit dio 62 Introduction CloudCompare is an application for managing and comparing 3D point clouds and to some extent surface meshes Its development began in 2004 as part of a CIFRE thesis funded by EDF R amp D and housed at the cole Nationale Sup rieure des T l communications ENST Telecom Paris TSI Laboratory TII team Work has continued after the thesis was completed in 2006 It is a platform to demonstrate algorithms that were studied in the thesis and developed thereafter In this sense it is not designed for commercial use This program was primarily created to deal with point clouds It was largely developed by Daniel Giradeau Montaut with the participation of Salma Bougacha intern at EDF R amp D in 2004 Aur lien Bey PhD student at EDF R amp D since the start of December 20
52. l time so you cannot have 1 gt 4 or 2 gt 3 Any valid modification to one of these values is interactively reflected in the current display 1 4 Properties of Objects 15 1 4 3 Meshes groups of meshes These two types of entities share the same properties Normals available in the CCObject section this field activates the use of the normals in the object rendering The normals permit us to create a semi realistic visual rendering of objects as they serve to calculate the lighting of the faces If this option is deactivated the lighting calculation cannot be performed and the surface of the object is displayed in a uniform color silhouette effect losing the effect of depth Faces available in the Mesh section indicates the number of faces composing the mesh In the case of a group of meshes the number corresponds to the total number of faces in the sub meshes that are part of this group Wireframe available in the Mesh section activates if the box is checked the wireframe rendering of the object The wireframe displays only the edges of the facets that compose the mesh as line segments It is created without masking hidden edges all edges are displayed even those which are not visible from the point of view of the camera as a result of being hidden by faces see figure 1 3 Figure 1 3 Example of a wireframe of a mesh 16 Chapter 1 Interface 1 4 4 Octree 1 4 4 1 Description
53. ld in the neighborhood of each point follow the tested law but not the inverse So the smaller the margin of error the higher the threshold of x2 will be thus we reject the aforementioned hypothesis less often and we class fewer points as not following the tested law i e fewer points show actual displacement Inversely the larger p x2 the more points will be found outside the law colored in red Note that this parameter serves uniquely to pre position the sliders for adjustment of the colors cut offs and saturation values of the scalar field for displaying the result on the screen cf section 1 4 2 2 These sliders can then be moved by the user before actually extracting points by calling the function Scalar Fields gt Filter by Value This will create a new point cloud composed of only the points presently displayed on the screen which is to say 2 3 Tools Menu 47 the points that do not follow the theoretical distribution In practical terms these are the points that showed meaningful distance measurements when you take noise into account Furthermore the y2 distance is extremely divergent amplifies significant differences in the original and this gives a lot of flexibility to the algorithm Therefore a relatively large change to the cutoff value will only have a small effect on the classification At worst we risk missing a small number of points at the edges of the boundary zones On the other hand to obta
54. lgorithm The propagation follows a 3D grid here an octree and one must therefore choose the level of octree on which to apply the algorithm Choosing a good parameter may take some trial and error because too low a level will result in large cells which propagate quickly but will miss local convolutions while too high a level will result in the opposite Furthermore the more difficult the propagation i e piece by piece the higher the risk of having areas near each other take on 2 2 Edit Menu 21 opposite senses Try the algorithm at different levels of the octree starting typically at 5 or 6 and raise the level until you find a satisfactory result The algorithm is very rapid and it can be executed as many times as needed it doesn t modify the normals only their sense Note the resolution is very fine it may still be necessary to use the Invert function cf section 2 2 5 to obtain the final desired result 2 2 7 Octree gt Compute This function calculates an octree structure recursive spatial subdivision on top of a selected point cloud or vertices in the case of a mesh Once the octree is successfully calculated it is attached automatically on top of the selected entity cf section 1 4 6 Note the maximum level of subdivision of an octree is 10 in this version of CloudCompare 2 2 8 Octree gt Resample Function for coarse point cloud re sampling The user can specify an approximate number of points
55. lor of a cloud select it then open the Edit Colors sub menu 1 4 1 4 1 The Set unique command lets the user select a single color to apply to all the points in the cloud cf section 2 2 1 The Colorize command enables the user to combine a new color with the current color of the point cloud cf section 2 2 2 This operation consists of multiplying the colors component by component the colors being expressed in the form of a Red Green Blue vector For example if the cloud is yellow at any point L L 0p and we apply Colorize with a purple color A 0 J we obtain a red cloud L 1 0 0 1 L 0 ol The Height ramp command enables the user to apply a gradient of colors along one of the 3 principal axes cf section 2 2 3 The Clear command removes the current color of the cloud After applying this command the points will display in the default color Properties of Objects Common properties Certain fields visible in the properties window are common to all entities in CloudCompare We find them in the CCObject section Properties B Mesh Group Zap 3D View 1 Name Entities Visible Current Display Name name associated with an object It can be modified by double clicking on the name right side of the field or by pressing the F2 key Entities this field shows the number of sub entities attached to the current object In the screenshot above for example the selected object is a gr
56. nks to an approach for extracting connected components This is a common algorithm generally applied to 2D binary images and which as been extended here to a 3D binary grid This function outputs one entity per sub cloud of points grouped in a new entity group in CloudCompare s navigation tree Figure 2 28 bottom right is a good illustration of its use The user usually chooses the octree level at which the algorithm will be applied This will roughly define the distance threshold beyond which groups of points connected components will be considered disjoint The larger the octree level the smaller the distance threshold will be the more subgroups will be extracted which is not necessarily desirable A second important parameter is the minimal number of points per connected component Min points If a group is composed of fewer points than this number then it will not be extracted as a new entity This limits the number of clouds created by the algorithm Finally the Random Colors option tells CloudCompare to generate random colors for each new cloud Remarks The larger the octree level the more memory RAM is required The octree level is therefore a sensitive parameter which is difficult to adjust a priori without experience An approach by trial and error might therefore be necessary typically starting at level 7 We can also display the octree Wire or Cubes display c f section 1 4 6 to visually estimate the size of
57. nsity of the reference cloud Limit max number of candidates when this field is activated the checkbox is checked it is possible to constrain the maximum number of candidates that the program may calculate for each trial In the 4 points Congruent Sets algorithm the program chooses a set of 4 coplanar points in the reference cloud and seeks congruent sets under translation and rotation within the data cloud candidates These sets are calculated as a function of the previously mentioned parameters and the program may find an enormous number of candidates hundreds of thousands of sets This parameter allows the algorithm to select only those candidates which are considered to be the best and can therefore considerably shorten the calculation time for each trial As a result you potentially deprive yourself from finding the best alignment as a result of the heuristic used to retain the best candidates When this field is deactivated the maximum number of candidates is unlimited which can result in a very long calculation time The Delta and Overlap parameters require that the user have a prior idea of how the clouds will be after alignment Figure 2 21 presents the results obtained from aligning two scans of a toy taken from two substantially different angles In theory the Align function can be applied to clouds with much less overlap than is shown in the example 40 Chapter 2 Functions Figure 2 21 Alignment of tw
58. nts 1 1 Reset zoom and re center the camera on the whole scene in the current 3D view De Zoom and re center the camera on the selected objects This has no effect if no objects are im selected Use these 6 buttons to toggle between different predefined points of view in the order that the buttons appear top bottom front back left right The predefined points of view are defined according to the orientation triad The button opens the lighting and materials parameters window In it the user can modify the mood of the light source the default color of point clouds the default color of meshes distinguish between front and back faces and set the colors of various elements that make up the graphical display This functionality is described in more detail in section 2 4 7 The P button permits you to access the point size parameter via a dialogue box visible over the top right of the graphical display area Point Sce ok To change the displayed size of the points simply slide the tracker the farther to the right the larger the points will be The display will update interactively Click OK to confirm and close the dialogue box Other functions are accessible via the Display menu Full screen show the main window in full screen mode To return to the normal display mode press the F11 key Refresh update the display Test frame rate start testing the frame rate expressed in Frames Per Second
59. number of points in the cloud and doesn t depend on the cloud s spatial distribution It is also for this reason that the maximum level of an octree is fixed at 10 in CloudCompare so the encoding per point is not too large the smaller the cubes are if ay is the length of the edge of a cube of level N therefore ao is the size of the initial cube an a 2 In effect each time we go down a level increase N we half the lengths of the cubes along each dimension For example the cells of an octree at level 5 are 32 times smaller in side length than the cloud s bounding cube level 0 the fewer number of cloud points per cube intuitively the smaller the cubes are the fewer points they contain the closer the surface of the octree the global surface formed by the collection of external surfaces of the cubes is to the point cloud The 3 screenshots in figure 1 4 give an outline of the division and the size of the cells of the octree at different levels of subdivision Many functions that apply to point clouds make use of an octree As much as possible the optimal level of an octree for the chosen calculation is determined automatically and transparently by the program All the same certain calculations ask the user to estimate a level to use In some situations the user will have to find a level that provides the best compromise between the number of cubes to deal with not too large therefore not too high a level an
60. o clouds with partial overlap On the left is the initial configuration and on the right is the result of alignment with an estimated overlap of 90 0 9 and 20 iterations The calculated alignment using this function depends largely on the configuration of the initial clouds Their geometry and degree of resemblance will make them more or less difficult to compare Therefore it is possible that in some cases the results will be fairly poor In these situations you can then use the fine alignment function described in section 2 3 4 It is recommended in the general case to resort to the fine alignment after using this function This function creates a copy of the Data cloud aligned with the Model cloud It is therefore not necessary to clone the clouds beforehand since they will not be modified directly 2 3 4 Tools gt Registration gt Register This function aligns two point clouds using the terative Closest Point algorithm by Besl and McKay IEEE Trans PAMI 1992 Clouds registration E Model and data gt E Model Cloud 0 El Data Cloud Fi swap Registration parameters Stop cribEerion Number of terations 20 ES C Error difference 0 01000000 v Enable furthest points removal Figure 2 22 Configuration interface for the registration tool 2 3 Tools Menu 41 Warning for this registration to work it is necessary
61. of the junctions of the tree that respectively show or hide the sub entities that are attached there The checkbox situated at the left of an element activates or deactivates the corresponding entity respectively when it is checked or unchecked The concept of deactivation is stronger than simply showing hiding an entity the active status is a generic property of any 3D object see further The inactive entities and all their attached sub entities are not displayed but further they are neither affected by the interactive operations like graphical segmentation and more generally they can no longer be selected and can therefore not be used as inputs for CloudCompare functions 1 2 2 Selecting Objects To select an entity the user can do one of two things click with the left mouse button either on the entity in the navigation tree or on its representation in the 3D view In both cases the selected object appears underlined in the navigation tree and highlighted by a surrounding box in the graphical context or object figure When an entity is selected the information corresponding to it appears in the Properties window Some properties displayed in this window can be modified others are presented for information only It is possible to select multiple entities by using the classic multi selection commands in Windows clicking multiple objects in the navigation tree or in a graphical context while holding down the CTRL key To de s
62. ood performance in absolute terms so do not hesitate to use large values for these parameters such as the default values 2 5 Plugins Menu 55 Remarks The algorithm creates a new scalar field of the type PCV and the color scale Grey is automatically activated The light simulated by the PCV algorithm is considered as coming from the Z positive hemisphere Z corresponds to the vertical direction so the point cloud must be oriented accordingly before the calculation If the 360 mode box is checked the light will be applied to all directions equally and direction will not play a part As the illumination calculated by this algorithm is a scalar field it is possible to adjust the sliders to control the saturation and contrast In the case of a mesh we can also use the smoothing and enhancing functions see sections 2 2 12 and 2 2 13 Once the correct parameters are set we can transform the scalar field into color using the Scalar fields gt Convert to RGB function section 2 2 23 The lighting coming from the sky is represented in a discrete manner by a limited number of light rays which are sampled uniformly from the hemisphere or the full sphere if 360 mode is active There is not much use of ray tracing in ShadeVis rather we should talk of viewing angles cf the article by Cignoni et al for more information Inthe case of meshes it is possible to accelerate the algorithm if the mesh is closed
63. op down list in the Octree section Octree Display type Display lewel Wire Points Plain cubes 18 Chapter 1 Interface 1 5 Interactive modification of entities 1 5 1 Manual segmentation Ce p B This tool is accessible via the icon It is possible to manually cut the screen containing the selected entities The user may define a contour by a successive series of clicks and decide whether to keep the points inside or outside of this contour The process is repeatable at will and the rejected points are hidden If the user confirms this trim the current cloud is replaced by the points corresponding to the selected points and a new cloud with the remaining points is created This can also be done to meshes but the triangles on the border are not technically cut triangles completely within the border are kept and the rest are considered to be outside of the selection ps Segmentation ON Right click to draw contour Left click bo close The tool activates an icon bar which appears at the top right of the screen the first icon pauses the selection allowing the 3D entity to be rotated for example between two cuts Caution if the segmentation is paused the currently drawn contour will disappear If the cut is performed see the next two points the segmentation is automatically paused If the tool is not paused the user can define the contour by successive left clicks on the screen The
64. otentially different point clouds this algorithm assumes the clouds represent the same objects This heuristic removes the farthest points at each iteration of the alignment to avoid that the differences between the clouds pull the final position from its proper alignment Therefore this option should not be checked if the two clouds represent the same object Choice of roles generic interface This generic interface figure 2 23 is used by all the distance calculation methods and a few other methods which use it in a similar manner It lets the user interactively assign a role to the two entities which are selected at the same time CloudCompare colors the entities according to their assigned roles In the distance case for example the reference cloud is represented in yellow and the cloud to compare which will be assigned a scalar field after calculation in red A Swap button will invert the roles and therefore the colors of the two entities 42 Chapter 2 Functions Wi F e Edt Tools D splay Plugins 20 ems ek R pe Tree e agas B compared Goud 0 d o Reference Cloud 1 s Cancel Figure 2 23 Standard interface for choosing the roles of entities 2 3 5 Tools gt Distances gt Cloud Cloud dist This function calculates the distances approximate or exact between two clouds of points When this function is called and after having chosen the role of each cloud see section 2
65. oup of meshes comprising 759 12 Chapter 1 Interface sub meshes and point clouds The sub entities can be accessed independently by expanding the current object in the navigation tree window cf section 1 2 1 Visible if this box is checked and if the object is not deactivated in the navigation tree cf section 1 2 1 then the object is displayed in the 3D view selected as Current Display Otherwise it stays hidden and only its bounding box can appear when it is selected Current display selects the 3D view in which the object is displayed cf section 1 3 2 The next section presents the properties specific to each type of entity There are certain fields or even whole sections that are available only after certain manipulations or for certain object types 1 4 2 Point Clouds 1 4 2 1 Common Properties Two fields are common to all point clouds Colors available in the CCObject section it activates when checked or deactivates when not the display of the color parameter for the point cloud cf section 1 3 4 Warning do not confuse the colors of points with scalar fields which though they are displayed as colors actually represent real scalar values associated with each point for which color is merely a mode of visualization cf section 1 4 2 2 It is therefore possible that the points remain colored on the screen even though the color of the cloud is deactivated Points available in the Clo
66. rding to the chosen color scale To return to the normal display for the point cloud uncheck this same box Name Cloud 0 Entities 0 Visible Colors CS Ee AE MO Pee ee Different color scales are available Each corresponds to a gradient of colors the first color being associated with the minimum value of the point cloud the last to the maximum value and the rest are associated with intermediate values according to a linear scale minimum value maximum value Blue to red BH n n 10 n m n D m Shades of grey INIT eee y d db d Red to yellow M IEN ENIUN NNUS i le d Red to white HEH leet dE d To change the color scale of the scalar field visualization select the desired gradient in the Color Ramp drop down list in the Scalar field s section Scalar Field You can display the color scale in the 3D view where the point cloud is displayed cf figure in section 1 3 1 by checking the Display Scale box in the SF scale section To hide the scale uncheck this box 14 Chapter 1 Interface 0 000000 4 0 000000 24858 3465 2488 3465 E WW Finally CloudCompare offers the user the possibility to more finely adjust the visualization parameters of the current scalar field by manipulating the four sliders available in the SF scale section or by modifying the associated text fields which are found to their right In case of doubt to know what a slider correspond
67. regardless of the number of currently open views the new element will automatically be named 3D View n 1 3D viewer windows created this way will be virgin they will not contain any objects and it is up to the user to display available entities as desired cf section 1 3 2 2 6 2 3D views gt Close Closes the active 3D view window Remarks Keyboard shortcut CTRL F4 The objects in the 3D view are not reassigned to any other window so are no longer visible The user can if desired re display the objects in any of the remaining windows cf section 1 3 2 2 6 3 3D Views gt Close all Close all the open 3D views 2 6 4 3D Views gt Tile This command partitions the display space between the different open 3D views The windows are arranged so that the entire display space is filled and there is no overlap between them they form a mosaic 58 Chapter 2 Functions Note this arrangement is useful to see all 3D views at once 2 6 5 3D Views gt Cascade Organizes the 3D views in a cascade they are superimposed according to the order in which they were created Note the cascade display is useful to navigate quickly between the current views 2 6 6 3D Views gt Next This command activates the next 3D view Note the order of the 3D views is set by their names and therefore by the order in which they were created The next view is therefore the open window that was created first after th
68. s that are not covered by the main program of CloudCompare They correspond to d files or dynamic digital libraries in Windows The application can perform perfectly well without these functions as their domain is typically outside of the core purpose of CloudCompare At the program s launch CloudCompare searches for available plugins in certain pre defined folders and only the d files on Windows are loaded and made available in the application s interface This manual presents the plugins available with the official version of CloudCompare It is nonetheless possible that for one reason or another you do not have these plugins in your installation Again this will not hinder the functioning of CloudCompare 2 5 1 Plugins PCV This tool rapidly calculates the illumination of points in a cloud or vertices in a mesh by using the Portion of Visible Sky P C V portion de ciel visible see figure 2 36 This lighting consists of calculating for each point the quantity of light it can see or in other words the quantity of light energy it would receive if the cloud had been lit uniformly This colors points as a function of their relative depth and provides good relief to the micro geometry In practice this calculation is done with an algorithm equivalent to ShadeVis originally proposed by Cignoni et al of the VCG 54 Chapter 2 Functions Figure 2 34
69. s to just position the cursor of the mouse over it and wait a few seconds until a description appears you can also read the help text visible immediately in the status bar at the bottom of the main window The role of each slider is defined in the following manner 1 Minimum displayed value All the scalar values below 1 are ignored in the display 2 Minimum saturation value All the points having an associated scalar value below 2 are displayed in the weakest color here in blue 3 Maximum saturation value All the points having an associated scalar value above 3 are displayed in the strongest color here in red 4 Maximum displayed value All scalar values above 4 are ignored in the display The display of the ignored values is governed by the NaN in grey property if it is checked points associated with ignored values are displayed in grey If not the points are not displayed The four values presented above define two ranges values to be displayed for which the points are effectively displayed in color The behaviour outside of this range depends on whether NaN in grey is selected the saturation range within which the gradient is used The lower bound of this interval is associated with the first color of the scale the upper bound to the last color and the rest of the colors are distributed linearly onto the values comprising the interval CloudCompare verifies the chosen values in rea
70. tation Angular steps Other daa foo z oo Sensor base 0 00 P Figure 2 10 Configuration interface for creating a sensor When creating a sensor numerous parameters can be configured via the different tabs in interface 2 10 rotation order order of rotations of the scanner motor mirror Here we use the angles O and as per the usual conventions for spherical coordinates 9 represents the horizontal angle the declination of the scanner represents the vertical angle There are now two choices 0 then Trimble GS type or then 0 Trimble SOISIC type Position center X Y Z XYZ position of the optical center of the scanner expressed in the reference frame of the point cloud Position center Sensor base gap between the laser emitter and the receptor used for triangulation captures such as typical SOISIC Orientation Reference point of the sensor expressed from the reference point of the cloud three vectors By default the matrix formed by these three vectors is left as the identity matrix which makes it aligned with the current reference point Angular steps dPhi angular step in degrees of the sensor in the direction Angular steps dTheta angular steps in degrees of the sensor in the direction Other Uncertainty Uncertainty in the laser measurement in percentage calculated automatically from the projection Other Max range The maximum scope ded
71. the center of gravity of the bounding box of the selected items and the right button translates the entities on the plane of the screen 1 6 Progress bars Some functions in CloudCompare require a relatively long processing time which can be between several seconds to several minutes depending on the computer on which the program is running as well as the parameters of the function and the complexity inherent in the particular function The progress bar is a window that appears during non immediate calculations giving a visualization of the advancement of the calculation when the bar is full the processing is done Clouds registration E Trial 14 50 best score 4 Nn 28 Cancel Among the commands on the progress bar some allow one to stop the calculation during its execution via the Cancel button the stopping of the program isn t necessarily immediate the 20 Chapter 1 Interface function will stop when it is able to do so When this button is not visible the calculation cannot be interrupted before it finishes 1 7 Toolbars Listed below are the buttons available in the various toolbars in CloudCompare ici O lt lt IN Main Tools Toolbar PB AMX ISEB cf section 2 1 1 Clone cf section 2 1 24 Fuse cf section 2 1 25 19 Register cf section 2 3 4 Be w cf section 2 3 3 Ska Subsample cf section 2 2 28 Sample points cf section 2 2 10 Cloud Cloud distance cf section 2 3 5
72. the sense that it doesn t create new points but only selects a sub set of points from the source cloud The quick method of subsampling via the octree cf section 2 2 8 chooses the point closest to the center of each cell Therefore the distance between the points is closer to constant if the initial cloud is sufficiently dense 2 2 29 Synchronize Aligning two selected entities this function simply applies a translation to make the centers of gravity of the two entities coincide Remarks To call this function it is necessary to have selected two and only two entities Again we use the generic interface for choosing the role of each entity see section 2 3 4 which allows the user to specify which is the entity to displace and which entity is the reference 36 Chapter 2 Functions 2 3 Tools Menu 2 3 1 Tools gt Projection gt Unroll This function will unroll a point cloud from a cylindrical or conical shape onto a plane See figure 2 17 KE Radius 100 000000 E As position Sa x 0 000000 o co0000 2 fo o00000 Iv Auto gravity center Figure 2 16 Configuration interface for the cloud unroll tool To do this supply the parameters that define the object of revolution the type cylinder or cone the dimension along which the axis is positioned must be X Y or Z axis for the moment a point through which the axis passes if the auto axis checkbox is checked t
73. tion figure 2 2 One can choose a basic color at the top left or a previously saved color custom colors at the bottom left or one can click on the colored zone at the top right and can vary the intensity using the sliding indicator at the right or one can manually enter the color parameters in the RGB or HSV fields at the bottom right 2 2 Edit Menu 25 Keyboard shortcut F8 2 2 2 Colors Colorize Uses the same interface as the Set Unique function 2 2 1 This function lets the user modify the current colors of the points by multiplying the components of the current color with those of the selected color If the entity does not have a color then this function is equivalent to Set Unique Gradient color E x Use default ramp Custom ramp First calor m Second color pe direction E Y Figure 2 3 interface to define a color gradient Figure 2 4 Default color gradient Figure 2 5 Example of a red to green gradient applied on the Z axis 26 Chapter 2 Functions 2 2 3 Colors gt Height Ramp The user can choose between applying the default color ramp figure 2 4 or a custom ramp for which he can define the two extreme colors figure 2 3 In the latter case the user must deselect the Use default ramp box The two colors at the extremes of the gradient can be chosen by clicking on the First color and Second color fields which load the standard color select
74. uced automatically from the projection Once the parameters are entered CloudCompare creates a sensor object that it will associate with the cloud This sensor object contains among other things a map of the depth of the cloud obtained by projecting the points along the reference angle of the sensor and equivalent to a Z buffer which permits smart filtering when calculating the distances as mentioned above but which may also be viewed section 2 2 14 or exported 2 2 15 Finally the sensor object can be displayed in place in the form of a small schematic of a 3D sensor see figure 2 11 2 2 Edit Menu 31 Figure 2 11 Sensor object displayed in situ 2 2 17 Sensor Modify Configure the parameters of the sensor The same dialog box as was shown in figure 2 10 for the creation of a sensor object appears for its modification but here the configuration values of the sensor correspond to those which were captured during its creation and not the default parameters 2 2 18 Scalar Fields Gradient This function calculates the magnitude of the gradient of the active scalar field When this function is called CloudCompare asks the user to specify if the scalar field corresponds to a Euclidean distance such as distances calculated between two clouds or between a cloud and a mesh If yes the algorithm will filter the outlying values which in this case are easily detectable in fact in this case the gradient in absol
75. ud section indicates the number of points composing the cloud Point clouds that have normals normal vectors have a Normals option for whether these vectors should be used in rendering the graphical display via a semi realistic shadow effect Properties a AAA Colors wl Clauc Paints 2975731 1 4 2 2 Scalar fields Some functions when applied to point clouds associate a real value a scalar with each point The set of these scalars is saved in a structure called a Scalar field and is linked with the point cloud It is possible to link multiple scalar fields to the same cloud but only one can be active at a given time The number of scalar fields associated with a point cloud can be seen via the Number entry in the Scalar field s section To choose the active scalar field you must select it in the Properties window of the cloud in the Current drop down list in the Scalar field s section We note in passing that the scalar fields are named automatically by CloudCompare as a function of the analysis from which they were generated One field can be displayed at a time 1 4 Properties of Objects 13 Scalar Fields Gradient norms Cel Distances d SLT A The user can visualize the values in the active scalar field by means of a color gradient applied to the points in the cloud To do this simply check the Scalar field box in the CCObject section Each point of the cloud will then be colored as a function of its value acco
76. ute value may never exceed 1 cff D Girardeau Montaut s thesis Remarks The algorithm creates a new type of scalar field Gradient norms Asin classic 2D image processing the gradient notably highlights zones of strong variation in the scalar field which may be taken as evidence of a zone boundary see figure 2 12 Asin 2D image processing it is often necessary to apply a Gaussian filter to the data before and or after calculating the gradient cf section 2 2 19 32 Chapter 2 Functions The fact that the absolute value of the gradient is never greater than 1 is true in reality for all scalar fields in which the values vary proportionally to the distance between the points it is thus for the case of a distance field Figure 2 12 Configuration interface for calculating the normals 2 2 19 Scalar Fields gt Gaussian Filter Apply a Gaussian filter to the active scalar field The only parameter to define is the size of the sigma kernel of the Gaussian filter To quickly estimate this parameter one can refer to the octree taking for the size of the kernel the size of a cell at level 8 for a soft filter 7 for a relatively strong filter etc The size of a cell is displayed at the level of the console when an octree rendering is displayed cf section 1 4 6 Remarks Sigma gives us the radius of the sphere in 3D that delimitates the neighborhood which will be considered around each point W
77. ve 3D view Use the Look in dropdown list See above to get access to the usual locations as well as recently used locations Use the Files of type dropdown list at the bottom to choose a filter to apply to files in the current folder if the filter is the format of the selected file is automatically detected by CloudCompare according to its extension 24 Chapter 2 Functions 2 1 2 Save file saves the selected entity to file or multiple entities if the chosen file format permits The file is saved using the standard interface figure 2 1 Notes Use the Look in dropdown list to access the usual locations and the recently visited locations Use the Files of type dropdown list at the bottom to filter the currently displayed files and to choose the file format in which to save An extension will be automatically added to the file name according to the chosen file format if the user does not specify an extension 2 2 Edit Menu 2 2 1 Colors gt Set Unique E Select color l xj Basic colors A AA E E AS EEEE II BEMENEENILrLI NENNEN A E Custom colors mI IT Sat UE Define Custom Colors gt gt Figure 2 2 Interface for selecting an individual color Define a color that will be applied to all the points in the selected 3D entity The choice is manual and is made using a classic interface allowing various modes of selec

CloudCompare - Danielgm.net

Contents

Download Pdf Manuals

Related Search

Related Contents