Various processing queries

I thought I’d post this as one thread as some bits might be relevant to others and some are existing threads…

I have this issue with 1443798370 and 1443794092.
There is something wrong with the 3D Model on 1443798370, but because I cannot see any other projections I’m not sure what it might be. I’ve processed this fine on other engines. I know it’s just a field of cabbages, so it’s a bit repetitive.

Land Spikes
On 1443794092 (Cliff collapse) I have spikes on land. Can you tell me what is causing these as I don’t see these with other engines. I’m guessing it is because I have oblique and nadir in there and something is wrong in the oblique set?

Timeline across UIs.
Where I’ve got two maps recorded through two different methods (standard mission and a Map Engine upload) for the same location I don’t get a timeline option to switch between them. 1443785413 and 1443794092, for example. (Cliff collapse 6 months apart for coastal erosion monitoring)

How can I get rid of the “clouds” in 1443798781 (hill-top monument) ? They look :sunglasses:, but they shouldn’t be there. This is another nadir / oblique mix. Is it shallow angles on some obliques, perhaps? I need to do a better pre-selection?

Thanks, guys!

The spikes are often caused by reflective surfaces or areas such as water, glass, etc. Do you not see this with other engines when including your oblique set?

When I look at the maps you mentioned I see the timeline view and can switch between the old and new maps.

@jeremy might have some more ideas about the clouds and spikes.

Timeline across UIs seems to be fine, now! :confounded:

I don’t get the funky clouds on other engines, although they are not perfect, either…

I fully appreciate the mathematical reason for spikes on highly reflective surfaces, but I don’t think that is the case here. Hence the question…

DroneDeploy vs Pix4D

That’s interesting. This is useful data. I’ll pass it along to the engineering team.

Hi Roger,

Firstly we’re loving the cliff-edge maps here, and as a fellow Brit, it’s great to see the sunny Cornish coastline again!

For the land-spike map, it looks like you’ve supplemented the nadir imagery with some shots taken horizontally of the cliff face; we’ve generally had much more success when the supplemental shots are oblique (say 45deg), as they tie in to the nadir set more effectively. I believe the spikes are a result of the misidentification of the horizon / cliff edge boundary. We’re working on making the Map Engine more robust to horizontal imagery, but for the near future I’d suggest taking shots at 45deg (more info on capturing oblique imagery in our docs here - hopefully it’s useful to you).

Let us know if you have any more questions / issues, otherwise we’re looking forward to seeing more maps!

Hi Jeremy,

Glad you like the Cornish cliffs!

Thanks for the input. I pulled anything shot at less than around 45° or showing significant horizon out of the image set and reprocessed. This got rid of the clifftop spikes:

…so that one is looking much better, now. Thanks.

Similarly with the clouds on the monument which are also now gone, but I think I need more shots at high altitude and angle around the top of the monument to fully capture the vertical detail. Now I don’t have those shallow angled shots covering the upper vertical surfaces I notice it has lost a fair amount of vertical detail on the monument and rocks in the foreground.

I will go and re-fly this when the weather is a bit better. I’ve had similar problems with masts and towers, so will try to run a mission on a mast as well.

I still have an issue with maps not showing once data processing has finished. 1443953401 is the latest one I cannot see. All others have now appeared.

Looking a lot better - as I mentioned in my reply on your other post (just now), for the 3D scenery such as the monument, I’d recommend using the “structure” option when uploading to Map Engine. Maybe you could try uploading the data set again using the terrain option so we can compare the quality of the monument?