Hello everyone,
I’m having a big issue marking my gcps. My project is composed of 3 flights with repeated image/directory names, with 10 bands, for a total of 17840. I had no issues importing the different folders and I got no warning about same name images. However, Step 1 did yield only 95% of calibrated images plus 10 blocks, even after many tests ( Iteration1_reopt_J4_Bellevue_MS_all_report.pdf (2.4 MB)).
I decided to ago ahead with GCPs marking anyway. On my first try, I ended getting a pink circle marking on pretty much all my images, which I found problematic given that:
- After marking a GCP on 2 images, a pink circle appears when the point is wrongly marked on other images. Outliers are indicated with pink circle and they do not influence the calibration results. (How to mark GCPs in the rayCloud - PIX4Dmapper)
So I erased everything and started again. Instead of sorting images by distance, I went by names. However, that’s when the weird bit happened. I realized that some images were simply missing for the selection panel. Like here, I go from 504_3 to 504_7 to 505_11 to 505_3 and to 506_3.
I’m starting to wonder if the fact that I have same name images could be creating some sort of problem within my project. Otherwise, how come I can’t access ALL the images to do GCPs marking??
Thank you for your help in this matter,
Anne-Marie
Update: I have renamed all the same images and redone step 1. Marking is ok as long as I use same image/different bands but all my markings are still shifting to outliers whenever I shift to mark a new image .
Also I still have missing images within the image panel
well this is sad. Even after testing other suggestion (like the multispectral template) I still didn’t get any better results.
But Agisoft Metashape processed the same project/images just fine! I guess Pix4D has some limitations.
Hi there, thank you for posting about the situation including the outcome. I can confirm that there should not be any issue with the duplicated names of some of your images because PIX4Dmapper always considers the whole file path.
Also, it sounds like there is an image-sorting issue. A workaround can be to use the Basic Editor to mark your GCPs on more images because it does not filter the list of images.
Finally, you mentioned that the calibration results were below expectations. Could you share the quality report? We would also be interested in further investigating the issue if you could share the dataset.
Cheers
Hello @Rhea_Garratt,
Thank you for getting back. Looking into my first post, you will find a quality report attached.
I will try to add some manual tie points and see if it does the trick. But I noticed something else weird. All the cameras are not oriented in the same direction for the same image (see below). We had that problem some time ago before upgrading our license, but that fixed it back then. However, now it seems to happen still. Do you reckon this could explain some of the issues I experience?
@Rhea_Garratt and just to follow up on something you said: what do you mean that the basic editor “does not filter” ? What does the raycloud editor do and why can’t it display all the images?
