Grid vs Crosshatch overlap

Just a bit curious and thought i’d ask some other people!

I think I’ve always been a bit OTT with my crosshatch overlap. Typically i use at least 75/75, usually, or 80/80 with lots of trees.

Obviously overlap is situationally dependant and there is no one size fits all answer, but considering 75/75 overlap with a grid mission would be fine for terrain data, what would you say the alternative for crosshatch would be? The default of 60/60 seems low, but I know we’re getting nearly twice as much coverage, admittedly from a slightly different angle… so is 60/60 crosshatch a good enough alternative to 75/75 grid?

I usually go out and test these things but time and weather are an issue at the minute!

Thanks

Good morning @Vantage! You are on the right track. I assume by grid you mean what others like to call the lawnmower or snake pattern… I think the most common overlaps for a normal pattern terrain survey would be about 80/70, some like a little higher or lower. I think it is safe to say that if you are doing a nadir crosshatch then you can probably bring that down to 70/60 and if using an oblique crosshatch then 60/60 should be fine. You could actually probably run less than that, but you would have to play with where the threshold is on different scenarios. In most cases it should actually be a little better because of the fact that you are capturing from all sides.

1 Like

I would rather have more data than less data. If it is crucial that you do it right the first time you cant go wrong with more images. The law of large numbers comes into play at that point. Plus you can always re-process using less images.

1 Like

Always true, but you can look at it this way as well. What is the most efficient way to ensure that you do capture enough good data every time? I can run a standard plan like this that I know will work, but what if I want to make it better? Just increasing overlaps is not the answer. Capturing in multiple directions is. This is especially the case when you have obstructions like trees or in this case giant crushing and screening equipment and very steep slopes cut out at the edges. The default plan is 225ft AGL and you can see it has limited the speed to 16mph.

Here is a nadir crosshatch using the older style Auto Flight Mode app. I use this because it allows you to determine the gimbal pitch, which I set at 90. Doesn’t make sense to me why we can’t do that in the new Enhanced 3D, but that is another discussion that I am sure will be had. As you can see the logic automatically reduced the overlaps to compensate so per their recommendation it is doing what I described for you. Yet we still are getting 53 more images… and from different perspectives at the cost of 6 minutes. Also note that the speed has increased.

To take it one step further, since it pushed us just into 3 batteries I raised the AGL 5ft brought the side overlap down 1% and slight adjusted the direction to optimize the changeover between directions and got back down to 2 batteries while still maintaining the original image count. It says 3, but I know from experience it will be two. Sneaky though, think about how many more images I am going to get around the perimeter. Cross-hatch missions also help reduced warping/bowling.

1 Like

That’s good to know, thanks! I’m finally going to be able to test this tomorrow. Bowing/warping is one thing I’m keen to avoid without overdoing the crosshatch :grinning:

1 Like

How did the test go with this project?

Allen

2 Likes

Although the wind forced us to stop after 2 batteries worth of mapping, I’m confident in 60/60 crosshatch now as a minimum.
It was an abandoned quarry site with loads of variety in heights and plenty of rocks and sides to test the benefit of the default 65 degree gimbal in crosshatch…and we only flew at 45m

Also due to us stopping because of the wind, half the mapped area wasn’t even crosshatched, it was only covered in a lawnmower pattern and the detail/stitching was still ok with 60/60 and an angled gimbal.

I can’t comment on accuracy though, because that’s not what I was going for.

2 Likes

Sorry to hear that the conditions got the best of you. Would it be possible for you to share the map processing report? You can save the PDF or just snip a screenshot of the Orthomosaic Coverage. I’d like to see what the machine thought of the flight. In particular how many images/pixel it found on the half complete crosshatch.

The good thing about obliques is that they not only catch details that could not be seen with nadir imagery, but they capture ALLOT more area per image. The downside is that the detail at the top of the frame is much further away from the focal point and will get mass details instead of fine. Being that you were able to fly a little lower will help this, but it remains. This behaviour is also why I will not use GCP’s with obliques.

Outstanding! That’s what I was hoping to hear.

I’m going to play around with “CrossHatch” and see if we can have similar improvements as noted in this thread.

Thank you for the update.

Allen

Just a disclaimer as I said I don’t use GCP’s with obliques. This was because when we tested with GCP’s the terrain accuracy was not as accurate (in some cases much worse) than standard nadir flight. It seemed to be smooth subjects that processing had the most trouble with. Paving, flat-work and etc… My preference is to just use crosshatch when I need a good model of vertical subjects. I always process the map twice, Terrain with nadir-only and all images with Structure mode for projects with vertical structures. I wish there was a way to merge the two in DroneDeploy, but I know that would not be an easy process. We currently do it in CAD and/or Carlson Precision 3D.