My project was divided into two blocks with a vertical gap of about 5 meters

Hello Everyone.

I am using Inspire 2 with X5S camera.

A total of 5 flight in 2 different days.

After processing I notice that there’s a gap on my data. I found out on the report that there’s 2 block generated on my project. I already tried adding 5 manual tie points on the intersection between the two blocks but the gap still exist. What can I do further. 

I will upload my raw data sets. about 254 images in total if you need it.

 

Thanks!

The gap is only in vertical. The horizontal of the two blocks is already overlapping but the vertical has 5m difference.

I also use 4 GCPs on the project area.

Well MTPs should bring it together when the data is poor but sometimes they just can’t fix horrible data overlap.  254 images is a snap to process so send me an email to setup a spot for me to get your dataset.

Adam.Jordan@nhiae.com

In this case, I would add MTPs on the border between the two blocks, making sure to mark them in images from both blocks. Then, click Process > Rematch and Optimize, which should remove the gap. Note that the Rematch and Optimize should work well here as there are only few images. 

If this does not work, I would try the merging procedure described here: https://support.pix4d.com/hc/en-us/articles/202558529

Just to follow up with those reading this thread.  I started processing Roy’s data and immediately saw part of the problem before even starting Step 1.

Pix4D identified 3 different camera setups so there is significant difference between flights.  This means that overlap between the flights needs to be high.  But as you can see, there is virtually no overlap between missions…now if this was all one flight then I think Pix4D would handle it fine.  Also looking at the pictures themselves there is either very poor focus or a very large GSD with this camera, which I am not familiar with.

I am still trying to configure Pix4D in a way to make this data work but the most obvious solution is to fly with overlap between missions.  Setting up the mission planner software at a certain overlap does not account for any overlap between flights/days.  I am working with Roy on this via email but this is a good example to show everybody the importance of multi-flight overlap, not just in-flight overlap.

 

Maybe some overlap in the blue circles but in general there is zero overlap between flights here:

1 Like

This is a very interesting problem because it does NOT make 2 blocks but does make a vertical offset 5-10 meters.  There is such a small overlap in side-to-side and such a large difference in picture quality between flights that the only solution I see is to re-fly the project.

I didn’t try dividing this up into 2-3 sub-projects and then merging since Step 1 put it all into one block from the start.

If there was enough overlapping between the flights, it would be interesting to have GCPs in the overlapping area. This would make it possible to mark them in both subprojects and then merge them. 

I ran the project with common tie points set as MTP and GCP with no difference as one project so doing a merge is about the last option before getting all new pictures.

You can see from the map view above that the side overlap is poor in areas and all the software features in the world can’t make up for lack of good data.

Good Day Everyone, I just came back finishing the whole project. The example above is just about 5 percent of the whole area. I was able to do some workaround on the focus Issue with Inspire 2-X5S. Please message me if you are also experiencing the same focus problem and I can share the work around.

 

By the way. I have no idea that not enough overlap between flights will result to this. I will try to process my data first and will update if I am going to encounter the same problem. Does anyone know why Pix4D does identify 3 camera model even though I used only one camera on the project?

 

 

We had a few cases with this camera where instead of one camera model several were identified (although it’s the same camera). There seems to be something going on with how we recognize this specific camera. The developers are aware and will check what they can do to improve the recognition of this camera. Note that this should not impact the calibration and processing. 

Another case where you can have more than one camera model for the same camera, is when you merge several projects. In this process one can select if the same camera was used in all subprojects or not, which will then create only one or several camera models accordingly. 

We have a similar problem too, we have one flight that lead to a 3 overlay of points, is there a way to select 3 mtp of the same thing and merge them directly in the project without divide into subprojects?

Hi nucleo,

I am not sure I understand your issue: was your project calibrated in several blocks (mentioned in the Quality Checks table in the quality report), or do you “just” see a shift in the rayCloud, or both?

You could share a picture of the Quality Checks table of the quality report and screenshots of the rayCloud for this project.

Introducing manual tie points (MTPs), then running Rematch and Optimize, can help tying the blocks together. When adding the MTPs, ensure you are marking them in images from all blocks, as precisely and accurately as possible. Aim for a reprojection error of maximum 1 pixel. See how to add / import and mark manual tie points (MTPs) in therayCloud or in the Basic Editor.

I’m sorry i haven’t saved quality report, but i think we have found out a solution with Basic Editor.

The more focused problem was:

I have  the same object in 3 different cloud on the same project (due to fact that we have images from different heights). Now i would like to join the 3 clouds together but when i select an ATP on cloud 1 or 2 or 3 i only see images related to this point or related to uncalibrated images. With basic editor i can see every images in project so i can select an ATP in cloud 1 OR 2 OR 3 and then mark on images that are related to cloud 1 AND 2 AND 3. I think it could be more comfortable to have an option to view all images (calibrated related to ATP and not related and uncalibrated) in raycloud too without using basic editor.

 

So i think this could be labelled as partially solved

Thanks

Hi nucleo,

It is good to know that being able to add and mark manual tie points (MTPs) with the Basic Editor helped solve your problem.

I agree that the Basic Editor does have the most user-friendly interface and have reported your suggestion to our product development team.

Thank you for your feedback :slight_smile:

Hi,
recently I used my DJI Phantom Pro + for a project. I shoot a family house with 388 pictures in 4 missions (3 circle mission at heights of 30, 20 and 15 meters and one manual mission. Also, i used 4 GCPs, and at the end, i have 2 blocks with huge vertical and small horizontal gap. I tried with MTPs, but with no success. What should I do?

Hi @ognjenkralj1,

I’m sorry to hear about the issues you’re experiencing with your project in PIX4Dmapper. Let’s see if we can resolve the multiple blocks and vertical inaccuracies you’ve encountered.

Based on the information provided in the PIX4Dmapper documentation, here are some steps you can take to address the problem:

  1. Flight Height Consistency : Ensure that the flight height is not too different between the flights, as different heights lead to different spatial resolutions. This can affect the matching process and lead to the creation of multiple blocks.
  1. Lighting Conditions : Try to fly with the same lighting conditions, same hour of the day, same sun direction, and same weather conditions. This consistency helps in matching images across different flights.

  2. Overlap Between Image Acquisition Plans : Check that each plan captures the images with enough overlap. Inadequate overlap can result in insufficient matches between images, leading to multiple blocks.

  1. Calibration Method : Verify that the calibration method used for step 1 is appropriate for your project. The calibration method can significantly impact the outcome of the image matching process.
  1. Manual Tie Points MTPs : Adding common MTPs between the blocks can help to merge them into a single block. Make sure that the MTPs are accurately marked in the images.

Since you’ve mentioned that you’ve already tried using MTPs without success, it might be worth revisiting the calibration method and ensuring that the overlap and flight conditions were consistent across all missions. Additionally, you may want to reprocess the images with a lower Keypoints Image Scale and enable Geometrically Verified Matching if you haven’t done so already.

For more detailed guidance, you can refer to the resources provided below each point above.

If you follow these steps and still encounter issues, please share your quality report with us. If you have any more questions or need further assistance, feel free to ask.

Cheers,
Daniele

Thank you for your answer.
Tried everything, now I have one block, but my 3D model looks poor.
I am sending you quality report.
Quality Report 1.pdf (921.4 KB)

Hi @ognjenkralj1 ,

Thank you for your reply and the quality report.

I see an important shift in the images between the initial and computed position.
Please ensure the vertical coordinate system of the images and the GCPs is set correctly.

Also, It seems you are merging drone imagery with terrestrial imagers.
I suggest you process the step 1 of these two flights separately and then merge the two projects together as described here:

I see also a warning in the georeferencing section in the quality check.
This can have several causes:

  • Accuracy values for GCPs are not properly set.
    • If the accuracy of the GCPs is known then it should be entered instead of the default values in the GCP/MTP Manager.
    • When using the default accuracy values in projects that are set to imperial (ft) then the GCP accuracy should be adjusted to 0.066.
  • Poor GCP marking: The quality of GCP marks can be estimated by reviewing the projection error for each GCP shown in the quality report. All projection error values should be less than 1 pixel.
  • GCPs are not marked on enough images.

Cheers,
Daniele