Necessity for Nadir Images With Crosshatch for 3D Model?

Subject matter is a group of buildings and surrounding landscape in about a 20 acre area. Objective is to get good modeling of buildings and a bit lesser so with the surrounding landscape. Also need to generate a DEM+contours for the surrounding undeveloped landscape. My plan is to do an Enhanced 3D Mission (Crosshatch + Perimeter) for the entire site and then a bit lower altitude in a smaller area just around the group of buildings. I’ve always noticed the “Enhanced 3D” mode just includes crosshatch and perimeter flights, no nadir flights. It also defaults to 60/60 overlap.

Does this mean if you run Enhanced 3D also doing a nadir mission doesn’t provide a significant enough benefit to make it worthwhile? If so, if only using Enhanced 3D for modeling buildings, should the overlap be increased? Or would a better quality model be achieved from combining the Enhanced 3D at 60/60 overlap with a nadir mission at say 70 sidelap, 80 frontlap?

1 Like

I think this feature needs to be rethought. The inability to run nadir with crosshatch hampers the ability to get space in between near vertical objects like cars and buildings that are close together and thus robs us of points to stitch to when generating a DTM. We end up stitching across a wide gap of where all the objects/cars were. The 90deg transverse can’t get in between each set of objects and even if it does it is on the edge of frame and unusable. @Andrew_Fraser, what would it take to be able to do both?

I would run two mission. 80/65 nadir and 60/60 crosshatch.

That’s what I normally would do, but I’m always looking to improve efficiency, realizing the law of diminishing returns. However you make a good point. I think I will stick to my original method of doing nadir+crosshatch+perimeter.


Personally I don’t use the perimeter function. That is good for standalone buildings where you can orbit one subject. I think that is a point of diminishing returns as you mentioned.

That said the perimeter flight path needs to be rethought as well. It is best to capture walls perpendicular to the wall. The current perimeter function is more of an orbit with single point of interest so as it strafes the yaw is changing as well and you capture skewed obliques which do not have good perspective. The recommendation would be that the ability to have multiple points of interest or a perpendicular heading would be more useful.

@MichaelL You can now run both Nadir and Crosshatch in a single flight by using the Mission function. Have you given that a try?

That is a good point, but I don’t see that option. Is it only on the RC device in the flight interface? It should be in planning as well.

In flight. You make both plans normally and just select both when you fly.

Well that is great and thanks for that, but for organizations I think it would be better to be able to preplan the daisy-chain and have the operator hit one button to fly all missions consecutively instead of having to have them select from specific missions that have been planned. What if I have four or five smaller missions for specific purposes that don’t need to be flown every time? I told the operator run Mission Alpha and we’re done. Of course the configuration you have is great as well.

1 Like

Yep, just tried it and it seems great if you are making all the flight plans, know exactly which ones are to be used and flying them yourself. It really needs a more structured approach for companies that have separate Drone Analyst and Pilots, not to mention the rest of VDC and Project Management.

Does DD always default to 65-degree gimbal angle on crosshatch and perimeter regardless of the altitude? If you are shooting a 75-feet tall structure, the angle for the obliques to get the sides of the structure would be far different when flying at 50ft versus 100ft. Do you just have to essentially guesstimate the best gimbal angle? Seems like DD really needs a couple additional setting like two focal point settings (a) horizontal: center of plan (the current default) or perpendicular to flight path, and (b) vertical: some altitude above ground and then if can automatically calculate the appropriate gimbal and yaw angles.

I believe the crosshatch is 60deg and the perimeter follows a focal point, but being that you cannot change the altitude between the two portions it may appear to be the same.

If I use Perimeter 3D and/or Crosshatch 3D, are all the images taken from an oblique angle and not from directly overhead? I used these 3D options on the assumption that there would still be a network of top-down images that were complemented by the 3D function of the Perimeter/crosshatch???
But all of the images are oblique, which means the results we are trying to achieve are very poor.
Primarily we are using imagery to create orthomosaic images, but we are also using the imagery to create point clouds and determine spot heights and volume changes over time. Hence, the thoughts were that the 3D imagery functions would assist in more accurate point cloud and 3D rendering, but the oblique images has meant a reduction in orthomosaic image quality.
Are my assumptions wrong? Should I stop using the 3D functions? The 3D functions didn’t seem to have this effect in previous years.

I always make a nadir run with crosshatch even if it is at lower overlaps like 70/60. This allows you to run a little less overlaps (70/70) on the crosshatch as well. What I have seen is that tight space in between objects on the ground are not captured well with obliques and tighten up when you add some nadirs. I very rarely use the perimeter unless I am focused on a small to medium size structure. You get some decent progress photos as well if the entire structure is in frame so there’s that.

Do you mean nadir using crosshatch? Or a separate one directional nadir mission?
I think I need to do a little more experimenting but when sites are time critical and several hours drive away from processing tech, it is a bit hit and miss.

I’m going to close this thread and follow here.