A really neat tool. My primary comment would be the distribution of control points. In your airbase example, I would guess that the accuracy of the georeferencing would decrease towards the image periphery where there appears to be less (or no) control points? If I were georeferencing this manually, I would first look for several good CPs towards the image edges (if available) and work in from there. This may be more important where the equipment type and aircraft position/orientation is sub-optimal? At this particularly airbase, there appears to be some good available OSM base mapping, but perhaps the source imagery lacks the required definition for the algorithm to make a good match in these surrounding areas?
Yes, the GCP distribution was pretty bad in that example, however we'll be able to fix distribution to be much more even. Corners/edges are the best GCPs. In this instance, even though I zoomed into an OSM base map, the actual georeferencing was done to satellite/aerial imagery. This, for us, is the ultimate limit to georeferencing - not sure how to get better than that.
@@asierpv6497 not yet, this will be available through the Bunting Labs AI Vectorizer though, so if you install that plugin you'll automatically get the update!
This is an amazing tool. Brilliant work lads.
Great! Can't wait.
A really neat tool. My primary comment would be the distribution of control points. In your airbase example, I would guess that the accuracy of the georeferencing would decrease towards the image periphery where there appears to be less (or no) control points? If I were georeferencing this manually, I would first look for several good CPs towards the image edges (if available) and work in from there. This may be more important where the equipment type and aircraft position/orientation is sub-optimal? At this particularly airbase, there appears to be some good available OSM base mapping, but perhaps the source imagery lacks the required definition for the algorithm to make a good match in these surrounding areas?
Yes, the GCP distribution was pretty bad in that example, however we'll be able to fix distribution to be much more even. Corners/edges are the best GCPs. In this instance, even though I zoomed into an OSM base map, the actual georeferencing was done to satellite/aerial imagery. This, for us, is the ultimate limit to georeferencing - not sure how to get better than that.
Good job, thanks a lot.
can we georeference with vector data (Base Layer)?
in the future, yes! although it doesn't currently do this
Amazing! where can we find it?
We'll be releasing it publicly in about two weeks!
Great. Thanks 🎉
@@buntinglabscan we download now? What is the name to find It?
@@asierpv6497 not yet, this will be available through the Bunting Labs AI Vectorizer though, so if you install that plugin you'll automatically get the update!
@@buntinglabs Can we use now?
When will the tool be deployed?
downloaded the plugin, but i keep getting the error 'georeferencing failed, please try again'
sorry, we're still hunting down some last minute bugs, I'm trying to release it asap!
@@buntinglabs i got the same error, Georeferencing failed. Please try again.
Can I get the plugin in my qgis 3.34.4?
Can you tell us how can we install AI georeferencer tool?
How would it work with very large orthomosaics? (6+ GB, some over 10 GB)
It currently doesn't, right now the max file upload size is about 16 MB, but we're figuring out ways to increase this.
Georeferencing failed. please try again
Good morning sir
Amazing, could be useful for aerial photographies? Congrats
I've watched full video, great job.
downloaded the plugin, but i keep getting the error 'georeferencing failed, please try again'