Greetings
I tried processing 3800 Pictures taken by a Mavic PRO, i have processed many project with the Pix4d but this is the bigger one which cause the Program to crash on steps two.
My PC specs are:
i7-7700
32GB Ram
120GB SSD
1TB SATA
GTX750ti
Anyways…Having Initial processing complete including many GCP, i decided to divide the project into 3 sub-projects of 1800 images (yes number of images are taken for Overlap).
Once the 3 sub-projects were created i processed all of the 3 sub-projects completely (steps 1,2 and 3), once done merge the projects like is shown in the guide
https://support.pix4d.com/hc/en-us/articles/202558529-Merging-Projects#gsc.tab=0
After that i check the report and the the whole block of images all together, but i noticed that Processes 2 and 3 are RED, like if i have to do them again, so here are my questions:
Will the data of the sub-projects be merged? Point Cloud, Mesh, DSM, Ortho, Index, etc.
Because this i have to generate steps 2 and 3 with all the images, it will cause the program to freeze once again.
Is there a way to merge all sub-projects’s processes into a solo project?
Still i have more questions, but will stop here for now so we can take this a step at a time.
Thanks In Advance
Only Step 1 is merged so Steps 2 and 3 will have to be repeated…and even Step 1 must be re-optimized after merging.
Sub-projects really only help when doing all the pictures together results in multiple blocks, each block needs ran in Step 1 separately then after using MTPs all the blocks can be merged.
In your case you need a better computer or deal with the 3 projects worth of output in a third party program to merge. Personally I am not convinced HUGE datasets can be processed via sub-projects because of exactly what you see, Step 2 still does the whole thing together and can crash Pix4D.
My solution was to get a Core i9 with GTX 1080Ti and 128GB of RAM…I have done triple your gigapixel size without crashing and my computer is available for hire 
Adam summarized the situation well.
Only a small thing I’d like to add. The reoptimize should be done in the subprojects before merging. During the merging procedure there is an additional adjustment that is done between the projects, so that it is not necessary to reoptimize again afterwards. The exception would be if you add new MTPs or GCPs after the project was merged.
Unless something has changed recently, I get a noticeable difference if I don’t re-optimize after merging…and yes I re-optimize each sub-project. I think it was version 3.3 the last time I tested this.
I am planning to use Pix4D Cloud to rerun a megaproject after merging subprojects. What is the maximum number of images that I would be able to merge on Pix4D Desktop that Pix4D Cloud could also handle? Our site is 1000 Ha and our 20 MP photos are spaced every 27 m…
I had issues with over 200 gigapixels in a single project but I don’t think there is a specific limit. If you have over 10,000 pics at 20MP then I would say it is likely to be too much.
Thanks for the guidelines, Adam!
Do you or anyone else know the implications of merging within Pix4D versus simply mosaicking in ArcGIS or QGIS or R. Would the Pix4D merge result in a unified cross-calibrated output vs. maybe mosaicking would result in not being able to safely compare the output values from one tile to the next (i.e. use both tiles in same statistical analysis), like comparing apples to oranges?
On the last job I had 7000 pics at 20MP.
When I processed it in one job (it took few days but it didn’t crash) I noticed some areas had multiple blocks, and in some areas I didn’t have the accuracy I expected (about 30 cm vertically). So, I decided to split the project in smaller areas (about 1500-2000 pictures) and I didn’t have issues with multiple blocks and accuracy was satisfactory (merging afterwards went fine). So I guess either I don’t have enough RAM or the PIX4D software gives much better results on smaller data.
Core i9 with GTX 1080Ti and 64GB of RAM
Hi Adriano,
When having multiple flights merging it is expected to have differences in optimization than if you process them together. From one flight to the other due to the scene, flight conditions, landing, taking off, etc., the camera internals can slightly change.
When merging you will have a camera optimization per each flight while when processing together depending on the options the internals will be optimized once.
Best regards,
Ina
Thank you Ina, it makes sense, because flight conditions where quite different for some area.
1 Like
It’s also worth noting that Pix won’t split the projects such that you have control where you need it for each of the sub-projects. It would be pretty amazing to have the option to manually select the sub-processing areas by creating a series of polygons, thus allowing the user to choose areas with the appropriate control. This would make the merging process easier for smaller projects, and would allow for better alignment when opening the data in a different program (Arc, CAD, Global Mapper, etc.).
Hi Chris,
Thank you very much for your suggestion. It would indeed great to have such tool.
I have forwarded your feedback to our developers.
Best