Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Eichhorn18

Pages: [1] 2
General / Re: Import marker coordinates for coded reference markers
« on: December 03, 2019, 12:08:31 PM »
Hello Eichhorn18,

I think you should export the measured (source) values for the markers to CSV file from the old project and then load them to the new project via Reference pane -> Import CSV command.

Alexey---this has worked. Thank you. can you please let me know why you recommend importing the estimated reference marker values rather than the true known values?


General / Re: Import marker coordinates for coded reference markers
« on: November 27, 2019, 12:43:40 PM »
Hello Eichhorn18,

I think you should export the measured (source) values for the markers to CSV file from the old project and then load them to the new project via Reference pane -> Import CSV command.

Will that allow me to automatically place the markers from the CSV onto the 12-bit coded targets?

General / Import marker coordinates for coded reference markers
« on: November 22, 2019, 12:56:09 PM »

I am trying to find out how I can assign local coordinates which are known (measured manually) to a set of coded 12-bit circular markers that metashape generates. My workflow is this:

1. Generate 12-bit coded circular markers in Metashape.
2. Print the markers, and attach to close-range photogrammetry model.
3. Measure the location of the coded-markers in the physical model.
4. Photograph model.
5. Align images.
6. Detect coded-markers.
7. Import marker locations (X, Y, Z). These were entered manually into Metashape in a previous model, however the coordinates remain the same and so to avoid re-typing them into the coded-markers, I import these from an XML markers file.

The issue is that when I import the markers, then appear as seperate markers from the identified coded targets. My question is how can I detect the coded markers (numbering 1 to 30), and then import the coordinates and assign them to those coded targets without the need to type in the coordinates?


I think we need the project where the referencing is properly performed via markers (valid projections specified and coordinate information input). Otherwise, if the coordinate system is not properly assigned, the model itself would be incorrectly oriented in space.
The accuracy of the marker's coordinates should be set to the real measurement accuracy. Even if the checkerboard pattern is printed with 600 dpi, the coordinates cannot be resolved with accuracy higher than 0.05 mm.

I understand better how the marker accuracy affects the dense cloud reconstruction now. Thank you.

Can you comment on whether the photo alignment for the sparse cloud uses a SIFT algorithm, and if so, does is do sub-pixel alignment?

A bit of an update on this post.

I've investiaged the problem of the sawtooth pattern in the very flat surface. I experimented with lighting sources with no effect. The orientation of my cameras relative to the flat surface of my model was parallel (the plane of the image sensor was paralell to the object of interest). It has been suggested that this orientation can result in systematic errors and poor depth construction in the direction perpendicular to the plane of the object.

I re-captured the photos in a ring around the flat surface with the camera orientied approximately 45 degrees to the surface. The result had much less of that sawtooth pattern.

I wonder if the systematic error is caused by the SIFT algorithm that metashape is using to identify tie-points between images. It would be nice if someone from Agisoft could comment on whether the SIFT algorithm for image matching is using sub-pixel interpolation--or pixel locking to the nearest pixel. If sub-pixel interpolation is not being used in the program, it could contribute to the systemtic error. Any comments from Agisoft?


General / Re: Help with source values
« on: June 05, 2019, 11:13:47 AM »
I've just done a test run using some steps on my university campus, and I assigned co-ords of 0,0,0 (x,y,z) to the bottom left target - I'm hoping that, with that co-ordinate, and the accurate scale bars, my DEM model will produce some accurate measurements.

It's likely that the raw reconstruction of your scene is pretty close to the correct scale if you're using a camera that retains the focal length, aperature, etc. in the EXIF data. The scale bars will obviously increase the accuracy, however your scale is likely correct un-adjusted, however--what is probably incorrect is the orientation of your model in real space. Unfortuantely setting one marker in your scene as 0,0,0 wont be enough to get the origitation (roll, pitch, yaw). Ideally you would have at least 3 points encompasing your space in 3D to get the orientation correct. I'm sure there are more scientificly rigerous statements on how many points you should have. If you are hoping to do any numerical calculations on volume or elevation models to get flowpaths on your gully, you'll need to orient your model correctly to gravity (I assume your application is scientific in nature).

General / Re: DEM accuracy
« on: June 05, 2019, 11:07:51 AM »
Hi again!

I'm doing the same as you in that at this stage I'm also only interested in the local scale of my model. How did you find the co-ordinates of your markers? Did you measure them in situ?

I'm thinking that the next stage, when I want a geodedic coordinate reference to my model (in WGS84), I'll do what you've suggested in that I place a marker somewhere in my plot and get as accurate a GPS location for that marker as possible.

So I'm doing much smaller scale objects. My scale markers were 3D printed in a grid and so I knew the design coordinates of my markers.

That being said, I have experimented with placing a geometric object of known size and dimensions in my photo capture area, with unknown marker locations. I then scale the reconstructed point cloud based on the know size, side lengths of the geometric object, to get the scale right. Then I can pick out the coordinates of the markers in another program by looking manually at each marker point in 3D space. There is likely systematic error involved in this and I cannot say what the accuracy of locating markers this way is. It was only possible owning to the small size of the area I'm photographing (300 mm2), and so likely is not an option for you.

I think your best option is measuring your marker placements insitu. You could simply set up a grid with wooden stakes and string and measure out some rectangles around your gully at different elevations, similar to how archeological digs are mapped out. You could then make the lowest corner the origin, and measure out with a tape the locations of the markers which could be placed beneath the string. It would be a bit of work, but in the abscence of a total station with GPS to the cm or better, you dont really have another option.

General / Re: Help with source values
« on: June 04, 2019, 10:49:46 AM »
I want to be able to optimise, and therefore reduce the error of marker placement for a more accurate model - does that make sense?

It makes sense what you're saying in so far as I understand you want to measure the accuracy of your model compared to the real world. I don't think marker placement will increase your accuracy in a meaningful way compared to the quality of your images, number of images, orientation of the images relative to your object, lighting conditions (presence of shadows between images).

Your markers should be placed around the extent of your model such that you have markers that encompase the object of interest in all three dimensions if possible. You said in your other post that you're photographing erosion gullies. I presume that these are somewhat inclined on a hillside? This incline may be sufficient elevation change that you don't need markers above the gully. How exactly do you know that your markers are 1 m apart exactly?

I have not worked with coded markers but I would hope that when you identify the markers automatically in metashape that you could then import and assign the coordinates to some of them, then refresh your model, and measure the coordinates of the others. This will give you some measure of accuracy.

General / Re: Unexpected results
« on: June 03, 2019, 05:21:39 PM »
Just a further comment to say I am aware of the depth of focus issues. I rotate the specimen around in 20 degree steps and focus on either the forward rim or the back rim, I do this for 3,4,5 heights depending on the object geometry. I imagine there is lots of in focus areas between the adjacent images and between the heights.

Can you share one of the individual images used for the reconstruction?

Looking a bit more in depth at the dense point cloud within Metashape, if I look at the underside of the dense point cloud, a latticework of gaps in the pointcloud becomes apparent.

What are the gaps in the generation of the lattice work that is being shown in the screencapture?

The spacing between these in the referenced model scale is 2 mm and this matches closely with the sawtooth like surface structure that I'm seeing when the elevation data is contoured.

General / Re: Help with source values
« on: June 03, 2019, 08:51:04 AM »
I want to be able to check the markers in the reference plane so I can optimise them to reduce the error.

What exactly do you mean by this? What do you think can be optimized, and how would you optimize it?

General / Re: DEM accuracy
« on: June 03, 2019, 08:47:49 AM »
Hi there,

I recently had a similar problem getting my local coordinate system to make sense. I'm not doing exactly the same thing as you since I don't need to georeference my model to a non-local geodedic system (eg. WGS84). I am only intersted in the local scale of my model--however, I made my scale work by using markers, not scale bars. After photo alignment, filtering, optimization of the camera etc., and dense cloud construction, I then place markers in photos and type in the known coordinates. You have to then locate the markers on all photos (or allow metashape to autolocate them with some adjustment by you). You would then make sure the check mark box beside the marker in the references table is turned on, and then you must click the refresh button at the top of the references table.

Whether you type in the coordinates of your markers in local coordinates (i.e. the distance in metres between the markers) or input GPS coordiantes likely depends on the accuracy of your measuring system. As you said you do not have the ability to use a GPS enabled camera due to the accuracy, I assume you have some other way of knowing where your survey plot is located within the world?

I'm not sure how you would translate it from a local coordinate system to a geodedic system, but I think it would be a similar process of assigning some markers in your scene with GPS coordinates. If you know the location of the centre of your survey from some more accurate method (total station, better GPS device) then  you might have to manually determine the coordinates of your markers in your scene. But since you say you know these locations to exactly 1 metre, you should be able to figure out their UTM coordinates provided you have the location of your survey plot (and maybe a compass).

Are you using raw images?

The image format is .jpeg caputred with the minimal amount of compression. There is no image manipulation / processing prior to loading into chunks in Metashape. ISO = 100 and the aperature and shutter speed are locked between photographs.

Thanks for your comments Alexey. Please see some questions I have below.

I think that there could be several reasons of the effect that you are observing:
- low quality images - they seem noisy and the surface of interest looks to be out of focus (could be related to aperture settings or unstable camera
orientation), so I cannot fully agree that the features on the surface of interest are sharp,
I agree that the sharpness of the surface of interest in these imaging is not perfect. These images were taken with a Raspberry Pi camera module, and as such the pixel size is not retained in the EXIF data. I've had to manually enter this as 0.00112 mm / px based on what is posted here
Focusing this camera is done manually by moving the lens towards or away from the CCD. This makes focusing the camera subjective.

- inaccurate marker placement - projections do not seem to be adjusted on the source images, so there's quite a large placement error compared to accuracy settings used (also the image quality doesn't allow to get precise location of the marker),
Would using coded markers and auto locating these in images be much more precise than what I've done?

- accuracy of 0.001 mm doesn't seem to reasonable value selected, as the nodes of the printed checkerboard pattern cannot be measured with such precision (I would suggest to use 0.1 mm max),
Can you please explain what changing this value in metashape actually has an impact on in the generation of sparse / dense cloud process? Does this affect the SIFT algorithm somehow or the bundle adjustment? How does having a grosely wrong accuracy value come into play in the reconstrution?

- disabled filtering for the dense cloud would result in considerably higher noise compared to standard settings (mild, moderate, aggressive),
My understanding is that if small surface details are important to retain that depth filtering should be disabled. This was the advice in one of the tutorials or white papers listed on your site, for the example of drone captured aerial imagery. It's obviously several orders of magnitude of scale difference between drone footage and what I'm doing but I assume the mathematics of the depth filtering is equally applicable for both scales. If I'm trying to retain micron level detail, then should this not be disabled?

- the surface of interest may have some inclination compared to the base level defined by markers, in this case the contour lines would represent the certain levels of the "slope".
I agree, there is likely a differential slope between the checkerboard surface and the surface of interest. This accounts for the gradual elevation change across the surface of interest but not the sawtooth patten.

Hello Eichhorn18,

Can you please send the source image set and the processed project in PSZ format to, so that we could reproduce the result and investigate the case?

I'm happy to email these files over. The photos are all 6MB each and when compressed as a ZIP file, it's 198 MB. Can you recommend how I should go about transferring these?


Pages: [1] 2