Messages

16

General / Re: Workflow for better textures on 3d models

« on: January 09, 2018, 11:12:25 AM »

that would be very helpfull
it works great on the ortho images with new polygons and i use it ALL the time there
so texture same fuction would be nice to have
frank

17

General / Re: Difficulties getting orthomosaic with long focal length

« on: January 09, 2018, 10:59:30 AM »

hi
i have not tired it from this height but i had sucess with images taken from an airliner in cruise just for fun...
however i think the problem is not the height but instead its the focal leghth
while big is good for nice orthos and traditional photogrammetry it is very bad for agisoft
agisoft does not like it it preferes light wide angles the largest i have ever used was 80mm that worked ok but given the nature of the geometry i can imagine 200mm will not realy work.
if its the same area i would recomend alligning the two sets together and see what happens then you can disable the 55mm ones
frank

18

General / Re: Get Coordinate of a marker/ any point in picture

« on: October 31, 2017, 10:04:41 PM »

you will have to change the view in the marker / contol panel to estimated
alternative you can create points (as a shape) in the ortho view and then richt click "measure"
frank

19

General / Re: DEM creation problems with to merged point clouds

« on: April 27, 2017, 11:21:35 AM »

hello alexey
thanks for your reply
here is the two cunks and the merge

PhotoScan.app.document.chunk.dense_cloud.meta
Out[2]: 2017-04-27 10:18:04 {'dense_cloud/density': '100', 'dense_cloud/depth_downscale': '2', 'dense_cloud/depth_filter_mode': '2', 'dense_cloud/duration': '15728.681', 'dense_cloud/resolution': '0.00200377260553607'}

PhotoScan.app.document.chunk.dense_cloud.meta
Out[3]: 2017-04-27 10:18:14 {'dense_cloud/density': '100', 'dense_cloud/depth_downscale': '2', 'dense_cloud/depth_filter_mode': '2', 'dense_cloud/duration': '1551.936', 'dense_cloud/resolution': '0.00200959661626811'}

PhotoScan.app.document.chunk.dense_cloud.meta
Out[4]: 2017-04-27 10:18:24 {'dense_cloud/depth_downscale': '2', 'dense_cloud/depth_filter_mode': '2', 'dense_cloud/resolution': '0.00231038693081674'}

both chunks were created in high but the altitude is different one was flown at around 100 to 120 m the other at 500m
however i have the idea it might have to do wit the points being classified
thanks a lot!
frank

20

General / Re: DEM creation problems with to merged point clouds

« on: April 19, 2017, 08:46:47 PM »

i tried again with different settings but got the same results
ps should not no this and i dont get it why it is doing this...
anyone?
frank

21

General / Re: Pixelpeeper: a free utility to visually estimate image sharpness

« on: April 13, 2017, 07:05:43 PM »

hi
since the topic is up again...
i have used this programm for quite a while and i realy liked it but now i have changed to DNG images and i can not use it anymore
any chance this could be implemented?
thanks
frank

22

General / DEM creation problems with to merged point clouds

« on: April 13, 2017, 06:55:18 PM »

Hi
i created a model from to chuks with a core model being in a higher resolution and a extended area model ...
now i have prepared clasefied and cut the point clouds leaving a hole in the overview and a excactly fitting core area to fill it
in the model view it looks very nice and there is no real overlap and noise in the cut area it looks realy nice
however when i create a DEM or mesh the different resolution of the clouds semmes to mess up the processing and it looks wrong
the center area is displayed correctly and everything is fine
the surounding is displayed just very corse mch worse then the actual resolution in the point cloud maybe 5% of the points seems to be used
this is the case also when i create with the option use all points insted of just the ground points
any idea how i can force ps to use this points as well..
frank
here is the point cloud view

and here the DEM

23

Feature Requests / Point cloud clasification by shape file

« on: April 12, 2017, 10:43:53 PM »

hi
iam currently working on a lot of clasifications while the technique has improved a lot here and iam usualy happy with the results i have to manualy adjust quite a lot.
so i thought maybe it is posible to clasify outside of agisoft using the ortho image or other availible sourves in GIS such as multispectral clasification or simple datasets such as house outlines and stuff like that... you know all the data that is already availible.
for excample in the scale i use i could use satelite data to idetyfy trees and big bushes automaticaly and would create a buffer in gis and export it into a shapefile
then i would do the same with my road and building data.
i could then idealy import that shapefile in Agisoft and have all the points classified by 2D location.
also it would be great if i have to re do the same area again and again... maybee even alowing to safe the clasification areas from agisoft for reuse and export them into a shape file ....
so is there a way to do this right now and i just missed it or could this be implemented in future releases
thanks a lot
frank

24

General / Re: JPG vs DNG... same amount of dense cloud points

« on: January 04, 2017, 03:15:40 PM »

the dense clud depends on the number of pixels in the image since this is not changing you can expect the same results in dense cloud
however the allignment works better with dng and it produces much less noise especialy in pmodels with a high dynamic range
i made tests with a monument half in shadow and half in light and the dng model was so much cleaner...
frank

25

General / Re: GPS Recommendation?

« on: December 30, 2016, 03:54:23 PM »

hi
especialy with buildings a total station is much better since with standing wals you get all kind of problems with gps..
also you can get a used total station for a good price everywhere.
preferably one which can survey without reflector this will give you nice accuraccy and a solid workflow.
as for gps i work with the leica 530 and iam hapy with it
frank

26

Feature Requests / Re: DNG format

« on: October 30, 2016, 09:26:14 PM »

this is good news
it would be good to keep the edit information only this makes use of dng realy practical for me since i have to do some edits before i get decent textures
frank

27

General / Re: georeferencing one cloud with another

« on: October 22, 2016, 08:55:27 PM »

create two chunks
then allign chunks using the points option will take a long time but should work
frank

28

General / Re: Perfect XYZ align without GPS

« on: September 22, 2016, 08:15:11 PM »

yes this sounds good your coordinates wil have x and y and all the z coordinates will be 0
to draw this and create the map in DS
start with one arbitrary point and measure the distance to the next draw the line in draftsight with the same legth
this is your baseline
now measure from point 1 and 2 the distance to point 3
draw a circle in draftsight with the radius of the measurements from point 1 and 2 to 3 the circles will intersect in 2 places select the corect one and place point 3 in the drawing then you can move forward to point 4 and so on
frank

29

General / Re: 3cm absolute accuracy without any ground control points using Agisoft

« on: September 21, 2016, 01:51:40 PM »

sound very good
i was using something similar home build for kite photography but since i never got the time calibration right i deserted the project.
so anyone know that the price tag might be on the UAV version minus the ground station since i can just use my leica GNSS for that
frank

30

General / Re: Perfect XYZ align without GPS

« on: September 21, 2016, 01:42:24 PM »

hi if your model is on even ground i would recomend use markers stiking to the ground as a preparation use some targets or the one photosan provides print them and stick them on the ground with sticky tape then do a systematic survey by whatever instrument you have
you can just use a tape meter since it is all flat you dont have to wory about the Z axis and if you are a litle bit up in your skils you can simply do a trilateration network in between the points
https://en.wikipedia.org/wiki/Trilateration
this is super easy and you dont have to calculate it just start measuring and then use circles (and thier intercepts) in draftsight to draw a map of your points finaly extract to coordinates for this points from draftsight into a csv file
then start photography and in agisoft import the csv file as a local coordinate system (dont use wgs it will do al kind of crazy projection stuff and think that your model is 8degrees long ;-) )
in the local coordiante system place the markers on the photos and optimize ---- DONE
as easy as 123
frank