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 - sakaic

Pages: [1]
General / Re: Phantom 4 Multispectral
« on: December 04, 2019, 08:57:32 PM »
Hi Alexey,

Is it possible to use the FlightRollDegree, FlightPitchDegree and FlightYawDegree data in the XMP data since the sunshine sensor is bolted directly on the RTK antenna?

General / Re: Phantom 4 Multispectral
« on: December 03, 2019, 06:11:40 PM »
Hi Guys,

You can download it here

It stitches fine but I guess at the moment reflectance calibration is not possible yet.

Let me if there are problems downloading. My responses will be slow as I am in the field with a mapping mission.


General / Phantom 4 Multispectral
« on: December 02, 2019, 04:52:32 PM »
Hi all,

Just got delivery of the Phantom 4 Multispectral. Would anyone here be interested in a sample dataset? I am uploading it to my Google Drive so if anyone wants it let me know and I will share a link here.

Does anyone have one yet? Any recommended workflow?

General / Re: GPU image alignment causes computer to crash
« on: February 13, 2017, 02:54:07 AM »
Ok. Basically it looks like my card is too old. "Upgraded" to Quadro 6000 and all is well

General / GPU image alignment causes computer to crash
« on: February 09, 2017, 10:44:39 AM »

I just updated to the latest version of Photoscan and wanted to try the GPU features.

I enabled both the GPUs and as long as the GPUs are used, the PC either crashes or the program completely crashes and I have to force restart the computer. The project files are locked and I cannot use them.

I am using 2 x ATI FirePRO V4900 on Windows 10 Pro with the latest drivers.

I unchecked the GPUs and the alignment process works fine.

Is it a driver problem?

I just obtained a phantom 4 as a cheap deployment over my S900+Nikon D5300 setup. The phantom 4 is a large improvement over the previous models especially the phantom 2 vision plus. If there is GPS lock, every photo is geotagged with long, lat, and alt data included in EXIF data. RAW is also available in DNG format which can now be directly processed but can only be taken at a 10 second interval. Jpegs are captured at a 5 second interval. The GPS is definitely not centimeter accurate but good enough for alignment assistance.

If anyone has any experience with the phantom 4, I would like to know if you are obtaining better results with fish-eye model or frame?

Bug Reports / Re: Linux version Las file error
« on: February 03, 2016, 05:10:11 PM »
Hi Alexey,

I leave the Coordinate system in WGS84

Source data  : Dense Cloud
Point Classes : All
Enable Point Colours

Bug Reports / Linux version Las file error
« on: January 22, 2016, 06:38:01 PM »
Hi all,

I am using version 1.2.2 in both a Windows machine and a Ubuntu machine. When using the Windows PC, the Las file exports fine but using the same project file on the Ubuntu machine, the exported Las file has no points viewable. The Las file is viewed using Lastools and Global Mapper.

Thank you

General / Re: xy reference points without a z-value
« on: September 16, 2015, 08:49:19 PM »
Hi Gary,

I do something similar and my process is similar to what stihl recommended. I only use scale bars to help with the precision of alignment and the entire model is free floating in the 3d space. To align the model to earth reference I only insert the longitude, latitude and altitude of 3 points in the model. I also then relax the error of the GCP. At that point, you can play with the altitude values of the points to ensure the "flatness" of the area.

General / Re: Camera altitude vs. GCP Elevation
« on: August 12, 2015, 08:01:27 AM »
From my experience with my Phantom 2 and S900 the typical altitude values stored with th e internal camera is relative. Recently been messing around with the older photos and I find that if you are processing with GCPs and they are reliable then it produces better results if I remove all location data from the photos.

Keeping the location data tends to generate weird results like sunken or floating point cloud sections.

General / Aligning Point Clouds
« on: August 06, 2015, 11:01:21 AM »
Hi all,

I have been generating point clouds with Photoscan. A colleague of mine has also been generating point clouds using a Kinect. Any ideas on how to align the 2?

I see that we can match Lidar data and Point Clouds in the software. Any guides on how to do this?

Thank you

General / Re: Failure to detect markers
« on: July 30, 2015, 10:19:44 AM »
Thank you very much Alexey.

Didn't go through that tutorial. Also, for a case of a large area to cover, the targets will definitely vary according to the distance. How can I make sure that as many of the targets as possible are detected?

Thank you

General / Failure to detect markers
« on: July 30, 2015, 08:44:43 AM »
Hi all,

I was trying to do automatic marker detect.

I printed out the markers 12-bit with a 20mm center and 1 target per page on pieces of A3 paper.

For my initial test, I simply placed them on the table and took a photo of them with my SLR. A downsized image is attached.

PS failed to detect any of the markers.

Any tips on  what I am doing wrong?


General / System Performance improvement for high density models
« on: June 15, 2015, 07:44:37 PM »
Hi all,

I have been slowly going through this forum and have basically given up trying to find an answer by search.

My question is this. Given that the typical orthophoto or DEM generated can cover a few square kilometers, the amount of points generated would be immense especially if the intended resolution of the model is to be high what parameters of a system is most important to handle these large models or is there a strategy to do this?

I have been generating rather dense models recently. A few billion points is typical. I would also like to do some editing ( remove some extra points to remove some tree overhang etc.) but the thing is clearly so immense that even rotating the model is a pain in the ass for my system.

A typical setup would be as such: I have a GPS locked D5300 flying and a typical photoset would be about 1200 images large. I then put them in the same chunk and align photos at 80k points, high accuracy, 0 tie points and no preselection.

Then for dense cloud, my typical setting would be medium with a moderate filter.

For mesh, arbitrary, dense cloud, 0 face count and the rest is default.

By this point, the output is so complicated it is even a problem to rotate much less edit it. My requires this kind of detail as I am doing modelling and calculations. My work is done on a Dell Precision T5610, dual E5-2650v2, 256GB RAM and SSDs. If I were to upgrade, what would help most? A GPU or a better CPU?

Sorry for the wall of text

Pages: [1]