Messages

481

General / Re: Pany GM1 ... banned for aerial! seriously

« on: October 24, 2014, 12:24:38 PM »

olympus-pen-e-pl7. Wifi. timelapse, excelent sensor... and very easy to hack trigger by wire
GR is a very good option but is not as flexible as an interchangeable lens one (http://www.four-thirds.org/en/microft/lens_chart.html)
Pana GX7 is excelent but heavier and bigger than Oly...
External power needs to be hacked too for the Oly as there is no oficial external power adaptor.

micro zd 12mm f2.0 and 17mm f1.8 are better choices than than the Panasonic 14mm, Panasonic 20 is also much better than 14mm whose only advantage is lightness.
 

482

General / Re: Panys GM1 and LX100... banned for aerial! seriously

« on: October 24, 2014, 01:21:21 AM »

Could somebody explain what is the problem with electronic shutter for aerial photography ?
I've seen many critics related to GoPro's shutter system, but honestly, I don't understand how it can interfere with photogrammetry priciples.

Electronic shutter means the sensor reading is done line by line. So, even though at pixel level, exposure may be very short (1/1600), the whole photo takes a while to be read, as much as 1/25seconds... that's ages!
Lets say you fly at 10m/s (fast for a copter but rather slowish for a plane), a photo that takes 1/25 s starts to be exposed at a point P and capture is finished 40cm ahead. Each pixel is excited for short time so the photo may look sharp, but the geometry is compressed along flight path if sensor readout rolls upwards... but the case is actually worse because the flight path is not a perfect line nor the speed is constant, and aircraft vibrations cause geometry ripples... the result is that photogrammetry principles are basically destroyed.
I got pixel errors for reprojections bigger than 14 pix for some markers that look perfect! the ground control error is 16 centimeters while the GSD is 4 milimeters. That's a nonsense!
I did the very same flight project one month ago just before excavations, with the very same lens but with an Olympus EPM1. It gave 0.15 pixel, and 0.009 m for ground control error... pretty decent.
After the flight with the GM1, the archaeologycal findings were covered with geo-textile and filled with sand to prevent damage and unwanted "visitors" during this winter. This was done inmediately after the flight and having photos briefly checked at my order "OK, surveying done, start closing the site"... but actually the survey done is a crap and I can't heal it by no means.

483

General / Re: Pany GM1 ... banned for aerial! seriously

« on: October 24, 2014, 12:34:31 AM »

  It seems You are right. Ive changed the title.
I read the maximum was 16000 and asumed that the new lx100 shared GM1 shutter (sorry, Panasonic)

484

General / Re: Panys GM1 and LX100... banned for aerial! seriously

« on: October 21, 2014, 09:37:15 PM »

in a Lumix GM1 the total exposure takes 1/25s depite you set speed to something as high as 1/4000... it uses electronic shutter for any S numbert over 1/500.
Look at the first image, it seems all decent... but hides enormous random distortions due to camera movements (shot at s1600, f4)
the second shows what you get at the same speed (*)
Beside this, the camera is all ways lovable: built-in time-lapse, almost ridiculously small, reasonably light and very good sensor rendition... if you are not flying and things stay steady.
Kind regards,
GEOBIT

(*) source http://m43photo.blogspot.com.es/2013/03/gh3-electronic-shutter.html

485

General / Pany GM1 ... banned for aerial! seriously

« on: October 21, 2014, 02:00:14 AM »

Yes, believe me! wonderfull cameras most senses, they're great for terrestrial job. but DON'T EVER put on board of a UAV, just trust me. Don't waste your money. It took me a while to understand why because photos looked beautifully sharp, but even sharpness can hide the worst evil.

Kind regards

GEOBIT

486

General / Re: Coordinates of a point from the 2d photo on the 3d model

« on: October 19, 2014, 01:45:29 AM »

If a mesh model has been created, one point in one photo gives a ray that intersects this mesh, so a 3d solution is posible. If no mesh, then you need at least two rays (from two photos) to get an intersection in the 3d space.
The tool for that is the marker. Put a marker in one photo and you'll get the marker's 3d coordinates in the estimated tab. (applies for Pscan Professional)
regards
GEOBIT

487

General / Re: Agisoft PhotoScan 1.1.0 pre-release

« on: October 02, 2014, 02:34:10 PM »

Please, dear friends, could you describe how do the so-called "non coded cross markers" should look like?
Thanks!

488

Feature Requests / Re: Roadmap & Upgrades 2014

« on: September 30, 2014, 12:33:29 AM »

Every software gets pirated, extra protection usually only means more hassle for the legit users

I'd welcome any extra hassle:
1 The lesser is the amount of non legit users, the more money for Agisoft team... this money can be very well invested
2 The price of our work will keep according to its value.
3 If your competitor plays the same rules, you will feel stimulated to do it better instead of feel the idiot for paying.
4 The pirate has no respect for the developers, he chooses to ignore the work they make, he surely ignores also the photogrammetrists, mathematicians, surveyors, etc. that have made possible the fact that we work with Photoscan today.
Why not to accept some hurdles if they make it harder the piracy?
You have to pay a ticket for the train, but some just jump on, and they many times manage to arrive before you and make your legit work just negligible... and they will shout  "hey man, look, the so called Photogrammetry is just a matter of three buttons, it's not actual work!, why do you ask for money!"

489

General / Re: Making measurements of a 3D Photoscan model

« on: September 24, 2014, 04:21:43 PM »

For such an easy task, it may be just enough to take measurements in:
Meshlab, Cloudcompare or even Adobe reader out of 3d-pdf like in the example.

490

General / Re: Making measurements of a 3D Photoscan model

« on: September 24, 2014, 12:25:36 AM »

CAD is not the way to go for large polygon count meshes and even worse for dense point clouds unless you have specific plug-ins. You should better try with software packages designed to handle laser scanner data if your field of application is mainly close range; or, just use GIS tools if your field of application is Geomatics.
Georeferenced data are usually rejected by programs that are not intended for handling large numbers coordinaes (UTM), so what you could try is to translate your model so to a place its center near the (0,0,0) by using a big integers vector during export from photoscan to Solidworks. I have never tried with that program, but it happens with others.
You will find much more usable photoscan models exported as DEM inside GIS tools, you can also make good use of the dense point cloud with programs developed for lidar data.
Regards
PD, if you specify your need I may suggest a finer approach

491

General / Re: Identifying misaligned photos

« on: September 21, 2014, 09:01:51 PM »

Select a few of the points you think are wrong, then right-click on the scene and filter photos "by points".

492

General / Re: When to use camera calibrate function?

« on: September 21, 2014, 08:36:36 PM »

One could need to calibrate camera in a process apart from the actual scan project when the poor object's texture is expected to deliver very few feature points so it may lead to a weak estimation of lens inner parameters. The same applies to the case an object that fills poorly the frame in most photos of the project, specially near the frame borders.
But it just makes sense if your camera lens setting can be fixed the same for both calibration photo set and photo-scan one. Therefore, precalibration of non-metric cameras is usually unpractical for several reasons: It's sometimes hard to have the calibration pattern with right dimensions to be shot at the same distance as the actual object. If the calibration board/screen is not available during the scan photo session and viceversa, its likely to happen changes in lens parameters in the meanwhile. Most p&s cameras and also many dslr lenses lack of the desirable mechanical stability, and lens parameters may change significantly in spite of any care when you shut down the camera, go to somewhere else and then turn on again.

493

General / Re: Making measurements of a 3D Photoscan model

« on: September 19, 2014, 02:01:08 PM »

Inside Photoscan: After the model has been georeferenced. Put markers at the ends of the line you want to measure (You can do it directly on the model and adjust precisely their position on each photo where the marker is back-projected). Select the two markers in the project tree and create a ruler object by right click.
Then you can read the measurement if you select view estimations in the ground control pane.

ISeeBees: If no geo-reference nor direct scale input has been done by rulers, you cannot trust any measurement from the model, not even if you know the focal lenght very well.

Outside Photoscan, you should be able to take measurements with most of CAD and GIS tools, but also with Adobe reader if you export 3dmodel as pdf. Try Meshlab too

Jose Martinez
GEOBIT

494

General / Re: inconsistent results

« on: September 19, 2014, 10:13:51 AM »

GPS on board of the cameras are usually crapy, and have refresh rates that sometimes do not match UAV frame interval. Can you export camera centers just after loading photos and overlay them onto your shp before alignment?. That could help to understand what happened.
Also, if that was not the reason for the so called inconsistency, could you show us your settings in the GCP pane?

495

General / Re: How to measure elevation on specific terrain points

« on: September 15, 2014, 11:09:33 AM »

The question is not whether Photoscan DEM is accurate enough or not, but if your photos, the project design and the ground control can deliver the resolution that you require in the model.
Lets say it does. The easyest way to extract spot elevations (coordinates) from the model inside Photoscan is to pin markers wherever you want in the model, and then export estimated values for such markers.