Agisoft Metashape - Insert markers

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
so if you if you're running a structure from motion processing with a data set of images and you'd like to achieve really solid and reliable results in terms of positions of points of the point cloud you need to tell the software that you are using and in this video I'd like to show you how to do that with edges of meta shape that there are some points in the images that he is it is processing that this point are very special point because you know the coordinates of that point because you you measure that in the field and you'd like to tell the software that that points have a precise position in the the images of the data set and then and I'll go through in the next video that points have that coordinates that you have measured in the field these points are called ground control points in the strata promotion or photogrammetric processing inside edges of metal shape they are called markers and in this video I'd like to show you how to place these markers in the images that are part that may make the data set that you are processing inside edges of metal shape and I will also go through a couple of shortcuts in order to speed up the process because you need to go through each of the image of the data set and if you have a lot of images to process and a lot of markers in terms of points that you know the coordinates it may take quite a long time to start and finish but it's necessary to go through all the pictures and fix the position of these points maybe some shortcuts will will be useful for doing that in a much quicker way so let's jump inside edges of metal shape this is a really sure really the data set that it's made by twelve pictures so it's really really tiny data set let's switch off the position of the cameras this is the sparse point cloud after the image alignment processing these points have information about the position in the 3d space and they have coordinates too because the images that I processed have coordinates longitude latitude and altitude that there's an issue about altitude because there's there's been some there's there's been some issues in during the flight these pictures has been shot by a drone and the IMU of the drone of course did something wrong in terms of fixing the altitude and so this these numbers are of course wrong but that's no problem because we're gonna fix that with the coordinates of points that have been taken in the field with precise instrument precise topographic instrument so in the data set the images that compose the data that make the data set there are some relevant points here I placed in the field some target high contrast to our target this is our yellow m black and white and black in order to fix the position in the image and being able to measure the coordinates in the field with a GNSS antenna receiver now I need to tell the software that there are at this point in these images and next step I will tell the software which are the coordinates of these points first step is telling the software that there are some relevant points so let's double click on the first image of the data set or another image of the data set it is not necessary that you click on the first one let's zoom in in order to see the center of this target right click and you have two options add marker or place marker these are similar but the read options let's start with place marker place marker new marker the software will place a flag a green flag here and there's a new label here point number one there is a point that has been added here in this merc marker window marker space area here and that if you notice something happen in these other images of the data set there's one green flag here and there are some more blue flag in some other images so let's go through let's go through this marker point number one I do not have any information about the coordinates information about the accuracy in meters and errors in meters but I do have information about the projections and error in pixels this is the projection error the projections this number informs me on the number of images that contain this point number one this marker and these are nine projection if I count the flag that I've got here I will reach number nine one two three four five six seven eight nine that's nice because I got nine pictures out of 10 out of 12 sorry that have this point number one inside this is the projection this is the rejection error it's kind of a tricky a tricky tricky stuff the reproduction error but it's really really important in structure from motion and in photogrammetric processing let me try to explain that mom but maybe I think I will go through that in another videos more specific about that the reproduction error is as the name is saying itself is the rejection of this point from the 3d model and here we can see this marker placed in the 3d model back to the nine images in this case the reproductions the projections is number nine to the nine images that contain that point so the software project back the 3d position to the two-dimensional position and then what it does it calculate an error between the reprojection that it has calculated and the position that you assign to that marker in each of these images the number the resulting number is a pixel number because the coordinate system of each images it's a two-dimensional XY coordinate system that is based on the pixel so the unit measurement is the pixel and that's the projection error a big number here means that there's something wrong with the reprojection a lower a lower number means that things are fine so the software was able to represent back all the the marker from the 3d to the two-dimensional space of each image blue flag and grief flag so if I switch if I double click on this images image where there's the blue flag i zoom in and I can see that there's a point number one at the same point that I placed a couple of minutes ago but there's a blue flag what does that mean this means that the software was able to find what it thought it was the right position of that marker for that image and it placed that in that position blue flag stands for the fact that you didn't check the position and you let the software go with that position you have the chance to clicking on that point and refining the position considering what you're just looking what you just seen for each of this image at this very high detail zone once you've done that the software change changes the color of that flag for that image from blue to green as you can see here there's a flag that it was blue now it's green that's really important because in this way you you check all the images or the data set you check out the marker of the that you are that you have and that you know the coordinates and you fix the position for each images of the data set you may you may be tempted to skip this step and let the software go with the position that it was able to calculate I think that this is not the right way to process if you'd like to have a solid output in terms of accuracy of points of the point cloud because it it may happen that the position of these points is not so close to the center of the target as it is in this in this situation it may be farther and it may lead to mistakes to error so it's much better to check all the pictures where the markers are and refine and fix the position in order to be accurate and be safer in terms of position of these points once you've done you need to go through all the markers that you have all the relevant points that you can see in these pictures in the data set I got nine points here of course I'm not going through all that but I'd like to show you a couple of things that could be interesting number one thing is choosing let me choose this one is choosing right-click ad market instead of place markers so if I chose ad marker what will happen in terms of imaging it's the exist it is basically the same thing that happened with the place marker but if I enlarge this window I can see here that if I chose add marker for each point that have been added instead of placed the software will list the coordinates the coordinates that have been calculated from the image coordinates here longitude latitude altitude of course there's a huge mistake here in the altitude then it it calculates kind of accuracy no it's not calculated it's it's taken from the settings and will go through the settings in the next video it calculates an error in terms of meters that matches the accuracy and matches the coordinates but please do not refer to this error as the reference for the output error that you may wanna want that you may have in the point cloud that you may want to process or you may want to go to other person other people to work on this is not the overall accuracy of your output this is just an error that has been calculated in a similar way than the projection error was but it's internal it's just everything is is completely inside the information that the software has now and this information are really the aren't really accurate because this information has been have been taken from a GPS device connected to the camera the GPS as a low accuracy positioning and also the altitude as a huge mistake so the main difference is the the fact that for added marker the software will tell you the coordinate that it calculates it processed as you can see now there are some tools some white flags appear here so as I placed as I added the marker in these pictures grab some blue flags here so double click here and let's go to check this marker this is a blue flag it's quite in a good position so I can turn it green but let's go to one of this picture that has the white flag and let's go through the white flag what does the white flag mean white flag means that the software is telling you listen I think that the position of this marker in this picture is here and it's related to the position of this flag but I'm not sure about that there's something in this picture that I'm not sure about the there's something in this picture that is kind of freaking me out there I'm not sure about the position so instead of placing it with a blue flag and using it in next processing I'm placing it with a white flag and you have the options for keeping white and if you chose to keep the white flag white the software will not use the position of this marker in this picture or you have the chance for placing it in the right position and moving it from white not to blue but directly to green so reasons for a white flag could be many could be overexposed image underexposed image could be also an obstacle between the point of view and the position of the marker the real position of the marker could be several you need to check them out in order to see if you can fix them or leave them wide if you choose to leave them why the software will not take in account that pictures and the position of the marketing that pictures in next processing if you are somehow uncertain or the position of the market if you if you're not able to clearly see the marker in the picture please keep the flag white and so you will not introduce enemy Stacy you will not introduce more mistakes in next next next step of the processing so due to the fact that you may have a lot of pictures this processing could take a lot of time so I will suggest you to double click on an image where you've just added a marker right click on the marker filter photos by marker now the software deleted here just it's just a filter he removed the pictures that do not have the marker in this case I think that all the pictures contain this marker because quite in the central position but now if you confirm the position of the marker and you just go do you just step to the next picture double click here what happens is that the software will keep the zoom level the same and will Center the image to the position in the position of the marker or the point number two and easy and it is extremely faster because you do not have to zoom out and zoom in to find that marker and zooming in in order to fix it there's an even more quicker way that is instead of double-clicking Queeny so moving from the the two-dimensional the image space here double-click and then go back here with your your selector just click page down or page up and the software will automatically switch from this picture to the next one keeping the zoom level the same and allowing you to really go through all the markers not all the pictures that contain that marker in a really really fast way and that's it we've down with point number two there's another options that I'd like to show you so you can remove the filters by clicking here reset filter there's another option and that's the option that allows you to go through all the markers for each picture so I'm gonna place point number 3 I'm gonna go extremely fast here I'm not sure about this picture add marker here add marker and then let's go through that I'm gonna place marker here new marker new marker and I think we've done two more add marker I generally speak generally speaking I prefer place in the market instead of adding marker because I always will tell the software which will be the coordinates the precise coordinates here so that's no point for me in having coordinates calculated by the software but it's up to you now I placed I think I place all the markers for all the pictures now there's another note another way to go through these pictures very quickly if you'd like to process market by marker as I showed you to show you a couple of minutes ago I processed the pictures marker by marker instead of image by image market by marker means that I fix the position of the marker then I go through all the image images that contain the marker that had AZ that have that market inside them and fix the position of the marker now I can work image by week by image so I have all the marker placed in a briefly a roughly way now I go through each image and go through all the markers that are inside the image and I can zoom out and zoom in but it will take me a lot of time so I'm gonna go zoom in to the first marker I've already fixed the position and then hit the tab button and I will go through all the the other market that are inside this this picture so the software allows me to keep the zoom level the same and that's really nice because it's really quick to go through all the markers inside this window inside these these images and fix the position as you can see there's a lot of time that could be wasted in zooming in and zooming out trying to find the position of the market and it could really be fixed in a really quicker way just moving using the tab button marker buy marker for each of that inside of each images of the data set and that's it as you can see I'm not I will stop that because I do not want to go through all the images even if they are not so many if you move to the 3d space you can see that in the 3d space clicking in making visible or invisible that marker you have the options for having them visible or invisible so you can have a good view a clear view of that position in the in the 3d model in the sparse point cloud here you have information visual information about what happened in the images in terms of fixing the position of the markers so you got green flag that means that all the markers inside that window in inside the image have been fixed in terms of position if you have blue flag and green flag that means that in that image there are some marker let's zoom in that there are some marker that have been fixed and this is point number one number two but there are there are also some markers that haven't been fixed by you so the software defining the position you will take in account that position but the software is telling you you didn't check them or maybe you check them but you left the position as I have calculated before thanks to the processing thanks to the align tend to the fact that I've already calculated a sparse point cloud so that's it for this video I hope you enjoyed it the markers positioning the markers inside or the pictures of the data set assign the coordinates to all these markers it's really important really really important in order to in order to have reliable and solid output for your processing the next videos I think we'll go through some settings about markers about definition and fix fixing position of the marker and we'll go through the we go through the process that will lead to insert and to type in the coordinates of this marker and finally we go through the optimized alignments processing that is that process that will lead you to understand and to realize how good and how have how good was how good is your processing and how good will be your output the sparse point cloud will be refined the position of the cameras will be refined the camera calibration information will be adjusted and all the next steps will really count could really count on a solid base that is the spice point cloud that has been optimized with the coordinates of this marker thank you for watching this video you can reach me out at www.mathantics.com so in the next video ciao
Info
Channel: Metrical Talks
Views: 9,887
Rating: 5 out of 5
Keywords: agisoft, agisof metashape, agisoft metashape pro, metashape, metashape pro, marker, gcp, ground control point, check point, structure from motion, sfm, photogrammetry
Id: Q8edLfwNzxU
Channel Id: undefined
Length: 23min 57sec (1437 seconds)
Published: Thu Jun 11 2020
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.