3D oblique models

Thanks Zane, so I need to subscribe to Basic or Pro?

Does MME have a guide for this, ie recommended overlap, altitudes, patterns to follow? DJI MIssion app has a setup for Oblique models. I have the smart controller (Android) so can't use the DME Pilot app.

No. That won't help. You need to purchase enough points to process your imagery. Don't use the DJI Oblique setting. It is pretty bad. 

The Full 3D setting is really only intended for use when reconstructing a model of a building or structure. It should not be used when flying a large area. 

Just fly a normal mission (not full grid) and fill in the overhung areas as needed. Obliques everywhere just makes a mess. 

Hi Ralph and Zane,

I have been looking for the definitive answer to the MPP and MME's collection of oblique and processing of images for 3D output.

Looking at:

https://www.mapsmadeeasy.com/data_collection (Highlighted image is below."

it says:

"Do Not Attempt to Stitch Vertical Surfaces"

"We do not create spherical panoramas like some other companies do. We create maps."

Which is definitive but then says:

"The only exception to this is if you are planning on processing your data as Full 3D in which our system will make use of oblique images. Even in this case, no horizon view should be uploaded."

Which to me muddies the water a tad or am I missing something here?

I love the MME pay as you go pricing but I looks like I am going to have to go back to DroneDeploy. Yes/No?

Cheers, Guy.

Guy: First of all, thank you for reading, thinking and asking questions. `Tis a rare trait these days... 

You are correct. There are some inconsistencies in our instructions. It basically all stems from our opinion that nadir imaging is fine in almost all circumstances. We do not recommend highly oblique imaging. 

Basically, we are saying we don't want you to do it, BUT... if you must please do it well. 

Our Full 3D setting should only really be used when modeling a single structure or area. It should not be used when mapping a neighborhood.

If you are trying to create a high quality model of building it may take some manual flight to get all the views you need. Just doing a full grid with the gimbal tipped up will often create more problems than it solves. 

If it was up to me we would take the option out of Map Pilot since people think that us offering it in the app means we condone it. We don't. It is in there for the cowboys that like to do crazy stuff and/or process their data on their own.

I recognize this is a bit of an older thread, but I'd still like to query the topic a bit if I may. Imagine I want to model a very small building perched high up on the height of land, like a fire lookout on a mountain summit. I would likely plan a very small grid flight fairly low (30m-50m altitude), and process as Full 3D. Would  processing this type of specific, small area job get better results if in addition to the grid I manually fly a radius or square path around the building capturing a set of obliques with ample overlap at something less than 40* up from nadir (maybe 25-30*) and upload these obliques with the grid set?

If, under these constraints, these would indeed be beneficial additional images for full 3D modelling of smaller structures, wouldn't it then be a pretty awesome option to be able to plan such an autonomous flight with proper speed, overlap, altitude etc. rather than manually fiddling? 

So what I'm driving at here is, would planning a linear mission around the object, circular or square, with specified altitude, overlap, speed etc., and then manually turning the gimble 90* inward and 20-30* up from nadir, essentially allow one to capture obliques with an autonomous flight mission? Would the drone landing foot come into frame above 15*/20* or so up while turned all the way inward? Optionally then, is it possible to add the option to fly a linear flight sideways as apposed to forward so that there would be no need to turn the gimble inward?

I imagine I'd have to experiment a bit to find the right distance from, angle up and flight altitude at which to fly the circle/square linear mission (and I will get around to experimenting with this eventually), but assuming all this being worked out, is there any glaring problems with this little scheme? Could a MPP linear mission provide an autonomous oblique flight plan for full 3D modelling? 

Chris: The fastest way to ruin a dataset is to include images taken while moving sideways or backwards. For 3D modeling we recommend using an overhead grid flight with the gimbal tipped up 25-30 degrees. Then use manual flight to fill in areas that cannot reliably captured from above. 

Proper overlap and orbit flight are almost mutually exclusive. When people do orbit flights we generally get 100+ views of the same objects and it bogs things down greatly. Proper overlap is 20-25 views of what you need to render. With complex structures this takes some thinking so as not to throw too much overlap at it.

Images that include any sky (or super long horizon shots) should never be uploaded to the system. 

Hi Zane, thx for the feedback.

Ok, I think I'm picking up what you're laying down. So no need for an orbital flight, the grid flight with gimble up 25*-30* sort of excludes the need for an additional orbital flight, but some minimal infill of the vertical surface is helpful. I guess I understood the idea of an orbital flight was exactly that, to capture and infill the the vertical surfaces not visible from nadir, ... but I also get that the gimble upward 25*-30* on the grid flight would capture some of the vertical views negating the need for an additional full orbit flight with a ton of overlap. Orbit flight idea scrapped.

But for a little bit more clarification, in your comments you state:

"With complex structures this takes some thinking so as not to throw too much overlap at it"

So do I understand from this when modelling full 3D complex structures I don't necessarily want to shoot for 80-83% overlap (as with mapping) on the grid flight gimble up 25*-30*, or were you referring here only to the manual infill images only?

Re the manual infill images: so on a smallish structure with say 4m x 4m square walls, would I only want a couple images of each side, or even just a single image (ie. minimal extra overlap to process)?

The manual infill images would be to provide more detail for vertical/overhanging areas that were not going to be captured well from above. Do both. 80% overlap should get you roughly 25 views of everything. When we texture the model that is created it uses the most perpendicular image to paint each surface so the side images would hopefully be the ones that get used there.

OK great thx, Zane. I probably won't get back up to that lookout again this season, but, in the meantime, I'll scrape up some other practice structure to test drive this process on.

Is there a way to adjust the gimbal angle while planning the mission, or does it need to be done manually once in flight? Thanks!

https://support.dronesmadeeasy.com/hc/en-us/articles/4401958968468-Camera-and-Gimbal-Controls

My app doesn't look the same with the camera option available