Support Website Contact Support Blog

Huge discrepancy between 2 different scans

I was hoping to patch in an an area to a previous scan so I flew a new portion of the site several weeks later. I tried to create a new project and associated Orthoimage, DSM, and DTM with the two flights and the results are very poor. The two scans do not line up at all in X,Y, and Z. One surface is floating above the other. See screenshot. Is there something I can do to salvage this? Thanks

Hi there

Use manual tie points and the blocks should align correctly. I have successfully merged flights 2 years apart.

2 Likes

Hello,

You can find more information about MTPs from here.

Kind regards,

1 Like

Thanks. This worked surprisingly well. One issue I did have was that the two flights were so far off (without correction), that I did not get a lot of images with the same tie point feature displayed in the panel.

This tutorial was pretty helpful. The related issue I am dealing with now is adding GCPs. When I import GCPs in the GCP/MTP Manager, they overwrite the MTPs (in the GCP/MTP Table) that I established to bring the flights together. I even tried creating a new series of labels for the MTPs

Do I need to start over and establish GCPS first, then MTPs?

Thanks!

1 Like

Could you please give us more information about what this mean? Could you share a screenshot highlighting the problem?

Hi. Yes. I cannot take a screenshot now because I am doing some processing so I will try to describe it. Basically, I have a table with 10 rows for MTPs (see discussion above) that I used to merge two flights together. When I then go to add GCPs, and import the ENZ text file, instead of adding 7 news rows to the table, as I would expect, it overwrites the first 7.

Maybe this isn’t a problem? The MTPs have already served their purpose and are no longer necessary? Anyway, I have just deleted all of the MTPs and added the GCPs and am re-running now. I will report back. I just want to make sure I am following the correct sequence and “best practices” for this workflow. Thanks!

Hi,
Please keep us updated with the outcome and if something similar happens again provides us with a video if possible.

So I followed the process described above and my two flights seem to be aligned horizontally based on the GCPs but one is “floating” above the other.
Screenshot_34

I guess now I need to go back and establish the MTPs to bring the surfaces together? I am wondering if this will work. See an example below where I am trying to use a fence corner post to bring the flights together. The bottom row of images is from my second flight and they show pink circles. Does that mean these outliers are out of tolerance and will be ignored or can I “force” them to be merged with the points in the first row (as they are obviously a common point)?
Screenshot_35

Let me know if I am not describing this correctly. Thanks

So I added 5 MTPS with common features in my two flights and Reoptimized and the second flight is still “floating” over the first. I must be doing something incorrectly. Any suggestions?
Screenshot_36

Here is summary of my Tie Points
Screenshot_37

Hi Christopher,

We would need more details to provide assistance in the most efficient way. Could you please send us:
▸ The quality report (.pdf format): …\project_name\1_initial\report\project_name_report.pdf

The 5 MTPs you are using are probably not enough so you will have to add more especially to the areas where the issue occurs. After adding the MTPs, make sure you Menu Process > Rematch and Optimize to allow Pix4Dmapper to compute more matches thanks to the MTPs you added.

I checked your images and I saw your project includes fields and snow which are not ideal image contents for photogrammetric projects. This doesn’t mean it is impossible but changing some processing options of step 1 (and 2) might help for the current project. Learn more in How to improve the outputs of dense vegetation areas using Pix4Dmapper?
The article also mentions best practices to capture images of areas that are difficult to reconstruct with photogrammetry so that you acquire the best dataset possible, especially regarding image overlap and GSD (influenced by flight height).

Let us know of the progress.

Thanks. Here is the report: BSR_HQ2.3_report.pdf (2.6 MB)

I tried the Rematch and Optimize and it looks like it may have helped a little. However, there was still a distinct different between the scans, i.e. one was hovering above the other.

Also after I re-ran Step 1 to generate the report, the scans reverted back to the original location with the big discrepancy. Annoying… Now I feel like I need to start over yet again.

Could you share the dataset? I could take a crack at it.

Hello,
May I ask if you edited the camera model of L1D-20c? In case you did, I would suggest not change the camera model.
Use also All Prior in internal parameters optimization


and process with a 1/2 Image Scale as well

First, try with all images together.

If this doesn`t work you could try to process the 2 flights (grid and free flight) separately and merge them , using for every flight the correct Matching Image Pairs option (Aerial Grid or Free Flight).


Take a look at these 2 articles:

Hi @Jaakko_Laihola. I really appreciate this. Would a folder with the images and text file with GCPs be all you would need? Let me know. I can share a dropbox link. Thanks!

That’s all I need + coordinate system of the GCP’s

Thank you so much! I had to break it into tow folders because of Dropbox 2 GB limit. Let me know if you need anything else or have any questions. I look forward to seeing you results.

@Jaakko_Laihola - Just an fyi, I’ve uploaded the data set. No rush. Thanks again for your help.

@Jaakko_Laihola I hope you are doing well. I was just curious to see if you had a chance to look at the scan data. Thanks again for your help and have a great weekend!