Messages

46

Python and Java API / Split in Chunks Script Fails on Empty Chunk

« on: July 02, 2019, 08:14:55 PM »

Hi,

I am using the split in chunks script to process 10,000 aerial images. All cameras were aligned and optimised in a single chunk and I selected a 5x5 split in the script dialog. Due to the geometry of the project the script created a chunk with no data and processing subsequently failed - is this expected? I have resized the original bounding box and changed the split to 3x6 which hopefully won't have any empty chunks. Is there any way to restart the script without losing the data that's already been processed? Also, how do you set the dense cloud filtering options?

Thanks,

Rory

47

General / Re: RGB DEM Palette

« on: June 14, 2019, 05:30:29 PM »

Hi Alexey,

Yes, thanks for the info. I set a single palette from 0 to 300m and applied it to all the DEMs and exported them again, looks great on the map now.

Cheers,

Rory

48

General / Re: RGB DEM Palette

« on: June 13, 2019, 05:43:10 PM »

Perfect, thank you Alexey. I hadn't noticed the "set palette" option, so when exporting the palette raster transform the corresponding heights on each export were different.

Thanks,

Rory

49

General / RGB DEM Palette

« on: June 13, 2019, 04:38:59 PM »

Hi,

We are working on a very large project of over 20,000 hectares and delivering height-field models, DEMs (32bit and RGB) and orthos. The site is broken down into sections by individual flights (approx 600 - 800 images) and we would like to export the RGB DEMs with the same height palette, i.e. the absolute height colours are the same on every individual DEM.

I can do this by colourising the 32bit DEMs in ArcGIS or QGIS then exporting them as RGB tiffs but it's a huge amount of extra work, is there a way to specify the height colours in Metashape Pro when exporting a palletised DEM?

Thanks,

Rory

50

General / Re: Best hardware for a PC for metashape

« on: June 13, 2019, 01:42:18 AM »

What sort of projects are you working on? We work with large aerial datasets and have invested in a i9 9900X 10 core (20 with hyperthreading) running at 4.26GHz, 64GB RAM and 2 x RTX2080 GPUs. This system flies through our projects, we’re currently working on a 20,000 hectare job flown at 4cm/px with an eBee RTK and processing around 1,500 - 2,000 hectares a day (dense cloud on high, DEM, ortho and height field model per chunk of 500 - 700 photos).

The X series processor allows for up to 128GB RAM should we need it, although the system never gets near the 64GB at the moment. The GPUs really speed up pair selection / alignment and depth map creation but it’s raw processor power that does the rest. Lots of fast cores speeds things up considerably.

Check out the Puget Systems website, they have a lot of tests and hardware comparisons specific to Metashape / Photoscan:
https://www.pugetsystems.com/labs/articles/Agisoft-PhotoScan-1-4-3-GeForce-RTX-2080-2080-Ti-Comparison-1263/

Cheers,

Rory

51

Bug Reports / Re: Incorrect Elevations in Report

« on: March 10, 2019, 11:28:42 PM »

Hi Alexey,

It’s in OSGB36 using the OSGM15 geoid tiff.

Thanks,

Rory

52

Bug Reports / Re: Incorrect Elevations in Report

« on: March 10, 2019, 10:59:15 PM »

Nobody else seeing this?

53

General / Re: Workflow to process the Photography of "PHANTOM 4 RTK"

« on: March 10, 2019, 04:38:53 PM »

This is how I use the dewarp data in the XMP tags to calibrate the camera in Metashape:

Code: [Select]

<drone-dji:DewarpData> 2018-10-23;3662.110000000000,3655.160000000000,5.850000000000,-31.610000000000,-0.268279000000,0.110783000000,0.000693505000,-0.000105054000,-0.031359200000</drone-dji:DewarpData>

Date: 2018-10-23
fx: 3662.110000000000
fy: 3655.160000000000
cx: 5.850000000000
cy: -31.610000000000
k1: -0.268279000000
k2: 0.110783000000
p1: 0.000693505000
p2: -0.000105054000
k3: -0.031359200000


Where:

• fx, fy - focal length
• cx, cy - principal point coordinates
• K1, K2, K3, P1, P2 - radial distortion coefficients

As the camera calibration dialog in Metashape only supports and single value for f, I use the calibrated focal length tag from the XMP data;

<drone-dji:CalibratedFocalLength>3666.666504</drone-dji:CalibratedFocalLength>

In terms of the conversion to OSGB36 with the OSGM15 geoid inside Metashape, I've found it to do an excellent job. We recently did a survey that had to tie in with an existing traditionally surveyed area. We did use several check points surveyed using post processed solutions from a Reach RS+ then converted to OSGB36 using the OS Net online conversion tool, the resultant errors were around 2cm RMSE XY and 3cm RMSE Z. which was well within the tolerance.

54

Bug Reports / Incorrect Elevations in Report

« on: February 26, 2019, 01:12:21 AM »

When exporting the processing report the elevation scale on the DEM is way off, although the scale is correct when viewing the DEM in Metashape. This is using Metashape 1.5.1 on MacOS Mojave.

I've only noticed this since 1.5.1. I can edit the report in Acrobat to change the heights but not ideal.

55

General / New Workstation for Metashape

« on: January 25, 2019, 02:01:37 PM »

Hi,

We are a Mac based organisation but looking to build a Windows workstation purely for Metashape. Just wanted to sanity check our spec before we go ahead and order. We work with aerial datasets with sites up to a maximum of 100ha and GSDs between 2 and 5cm/px. The bulk of our jobs are smaller, with 200 - 500 images. The PC spec we're looking at is:

Intel i9 9900X 10 core 3.5GHz/4.4GHz
ASUS TUF X299 motherboard
128GB Corsair Vengeance LPX 3000MHz
2 x 8GB EVGA  GeForce RTX2080 Black Edition
500GB NAND Samsung 970 EVO
4 x 4TB Seagate Firecuda (RAID 1+0)

We're looking at at least a 3 year ROI on this hardware. Do you think this is a good spec?

Thanks,

Rory