Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - WSurvey

Pages: [1] 2
Feature Requests / Centre of weight detection for placing or moving markers
« on: February 16, 2019, 08:58:54 AM »
It sure would please me to be able to place or move targets to the calculated centre of weight of a white-on-black or black-on-white dot. Drawing a circle around the region then have the software add or move a marker to the centre would be a bonus for situations where automatic detection just won’t cut it.

Or even better a snap-type of mode could be enabled whereby if a marker is moved then it snaps to the greyscale centre of mass of the surrounding 4, 9, 16 or whatever amount of cells you specify as a tolerance.

Feature Requests / Re: Background Map
« on: November 12, 2018, 01:12:30 PM »

and ...

Web Map Services

General / GPS INS/Offset direction?
« on: October 03, 2018, 05:08:06 PM »

Pardon me if this question has an obvious answer, but is the GPS/INS offset from camera to GPS unit or GPS unit to camera?

I didn't find the answer in the documentation for whatever reason.

If my GPS unit is mounted 0.25m above he camera centre, is the offset I should enter +0.25 or -0.25?


General / nm
« on: September 19, 2018, 08:43:10 AM »

In search of a more accurate orthomosaic:

I'm using a phantom 4 pro to collect images to process using photoscan. I have the ability to tag some of the photos with very accurate GPS locations (within a few cm). If I have a project with 100 images using the standard 10 meter default accuracy, can I add 10 photos where i specify  much greater accuracy, and have photoscan use those as aerial GCPs?

For the curious: I've equipped by phantom with a Reach M+ GPS. I log raw GPS data during a mission, and post process the data using EZsurv. The problem is that the image timestamps aren't accurate enough to use the more accurate GPS data. My idea is to fly a normal mapping mission, then manually hover the drone at various locations and manually take the photo. Based on my testing, manually taking a photo while hovering ensures the timestamp is within the correct second, and drone will generally stay within 10 cm during that second if it isn't windy.

Hello sswallace,

If you ever log in again I’ve sent you a PM. I’m doing similar things to yourself. Hopefully we can swap some ideas.

General / Ideal shape for feature point detection?
« on: August 06, 2018, 05:22:56 PM »

What, specifically, is the ideal shape, size and composition of a feature point that Photoscan is looking for?

If I was wanting to print off some artificial textures from my laser printer for close-range photogrammetry that would be perfect for feature point detection, what would the texture look like? How big in pixels? Soft or sharp edges? Dark marks on light background or light marks on dark background?


General / Re: Dense Cloud - Where do the rings come from?
« on: July 24, 2018, 01:02:39 PM »

The circles are called "moiré" and they are a consequence of interference between the two spatial frequencies of the two overlapping images. This moiré will become less noticeable when other paterns superimpose in the same place as more photos are overlapping.


Hello JMR,

Thanks for the explanation.

General / Re: Dense Cloud - Where do the rings come from?
« on: July 23, 2018, 08:56:14 AM »
Hi WSurvey,

These 'rings' commonly appear in areas with insufficient or low image overlap (e.g., at the edges of surveys).


Hello SAV,

Thanks for replying. There is sufficient overlap between this pair of photos, but I think I understand what you meant. They generally appear towards the edges of jobs where only 2 images overlap. If you generate a dense cloud where a particular piece of ground is covered by 3 or more images you don't tend to see them. Or, at least, their presence is clouded (pun intended).
I was kind of interested to see if anyone was aware of the mathematics of their appearance; are they an unavoidable part of the depth extraction from 2D images?

To me, it seems that if they can't be avoided then there is an imprecision about the whole depth-from-2D enterprise. Otherwise something is lacking in my equipment /techniques.

General / Dense Cloud - Where do the rings come from?
« on: July 21, 2018, 01:00:18 PM »

Attached is an example of a dense cloud generated using 2 x 20 Mpx photos from a P4 Pro. An accurate camera calibration is in use and they are part of a larger set. Relative position and orientation should not be a problem.

Why do the rings appear? They change in number and amplitude depending on the settings but are always there.

General / Limitations of camera distortion model???
« on: July 01, 2018, 03:46:47 PM »

I have noticed that towards the edge of my dense clouds, where the cloud seems to have been created using a small number of photo-pairs (only 2 perhaps) there occurs a ripple type effect. I'm noticing the same effect in my camera calibration residuals. I'm processing aerial photography from a Phantom 4 Advanced, 20Mpx camera. There are a good number of survey-grade markers in my area of interest and this occurs no matter how many parameters are solved for in the camera distortion model.

Could this be an effect or limitation of the camera model, or in the generation of the depth maps, or something I'm missing in my own techniques?

Screen caps attached.

General / is CPU speed/core-count important for GPU processing?
« on: January 31, 2018, 05:18:51 AM »
The following question only pertains to GPU accelerated processes: Once you have the GPU handling the heavy lifting, does it really matter what kind of CPU is feeding the GPU?

for example, is an i3-2100 able to adequately feed a GTX 1050Ti all the instructions that it needs to be working at max capacity? would having an i7-3770 feeding the same card make much difference at all for the same computational problem?

Feature Requests / DEM Elevation heat map as Orthophoto
« on: March 19, 2017, 04:55:12 AM »
I realise you can export a geotiff for GIS programs like QGIS that can use them, but for simple CAD programs it would be nice to be able to export the DEM heat map as a JPEG with a world file.  Apologies if this is already possible but I can't seem to find the option.

General / More GPU's or more nodes?
« on: March 09, 2017, 01:35:31 AM »

Cost aside, if one wanted to decrease processing times for aligning photos and building dense clouds, or any time consuming operations, would one go about getting more licences to run network processing or start throwing more graphics cards at the problem? Assuming each extra computer has useless (integrated) graphics but good core i7 processor.

Also second question. I'm processing on an i7-3770 with 16Gb and no graphics card (just motherboard integrated GPU) and that's the best I've got in my office. Am I missing out on much speed by not having a good, single, stand-alone graphics card like GTX 970?


General / Re: Disabling all estimated marker projections
« on: March 03, 2017, 12:39:51 PM »
Hi Jon,

I think I just went thru your problem myself. Instead of Right Click > Create Marker, define the marker in the first photo by: Right Click > Place Marker > New Marker. In the second photo you'll see a guideline crossing your marker, but there won't be an instance of that marker created automatically. Right Click > Place Marker > 'Point x'.

If your photos are aligned well then the instance of this marker created by intersection in the remaining photos will be very close, hopefully only requiring small movements.

General / Too much wasted geometry
« on: March 03, 2017, 01:49:49 AM »

I am trying to make reasonably sized 3D PDF models of various sites for a client. The PDF has to be reasonably sized to be able to be displayed and rotated smoothly on ordinary computers. So I need the sites to be faithfully modelled with as few triangles as possible. Trees and building rooves are in the scene and no matter the final polygon count you specify they take up too high a percentage of the polygon count because of the roughness of their surfaces. Then the rest of the site is poorly modeled.

Rather than extensive editing by hand, are there any processing or filtering strategies that will serve my needs? I need a more uniform distribution of triangles. Not 1000's of triangles just to model tree tops and rooves.


Pages: [1] 2