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 - mskancke

Pages: [1]
General / Yaw, pitch, roll not loading as reference
« on: July 06, 2016, 01:15:03 PM »

I have a dataset collected with a DJI Phantom 3 Pro and Pix4Dcapture app.

I loaded the photos into PS and lat/long is automatically inserted since its embedded in the exif data.
I have cleaned up the log from Pix4D which contains yaw, pitch and roll.
Created a new txt file which contains camera name and the orientation data.

When I try to import the orientation, I mess around with the column numbers to match the data in in the txt file.

Click OK, and no data is imported.
The existing lat/long is still there, but nothing added to yaw, pitch or roll

Is it not possible to do this task, or am I doing something wrong?

Dataset and logfile can be sent if needed, it ain't big since this is a test run..

General / Re: Alligning photos takes ages
« on: July 03, 2015, 10:51:48 AM »
Hi Alexey

Ye, my gut feeling was saying ive done something wrong with the settings prior to starting the task.

I aborted the job, and set preselection to referenced.

Its been so long since I worked with photoscan, so I dont remember the ideal settings anymore, so correct me if im wrong.

Accuracy: High
Preselection: Referenced (I have a logfile containing GCPs for all photos)
Max points 40000
Tie points 1000

Should I change the points value?

General / Alligning photos takes ages
« on: July 03, 2015, 10:39:46 AM »

Hopefully not everyone has started their summer vacation yet.

I havent been working with photoscan for a while, but got a task to make a simple Georeferenced Orthophoto.

I fired up the drone and took around 800 photos of the area needed.

Ive started the allignment process, but it says its gonna take over 7 hours (its already been working for 24 hours)

Ive done this type of assignments before, with around the same amount of photos, but it never took more than 1 hour ish to allign the photos.

Im not working from my usuall station (its being repaired), so I took it into consideration when I gave the customer a timeframe, but I never thought it would take this long.

Im on a I7 gen 4 processor, so I really dont understand why its taking so long.
Taskmanager lists photoscan to be using 90-100% of processor capacity

Is it some bug in the new version?
Or have i missed something prior to starting the allignment?


Marius Skancke

General / Re: New Nvidia GTX cards
« on: September 22, 2014, 10:24:16 AM »
On a little sidenote;

The 900 series has destroyed Titan on several benchmarks regarding pure processing power.

General / Need help to classify points
« on: September 19, 2014, 10:02:21 AM »

First of all, let me appologize, and say thanks in advance for any helpful tip. I am still pretty new to Photoscan.

We have been out with our Drone, and have taken pictures of a small quarry.
Our task is to produce a high quality georeferenced orthophoto, a detailed map with contour lines and do a mass calculation of remaining mass in the quarry.

The Orthophoto is fine, but the contour lines and mass prosjection is what I'm having trouble with.

The reason for this is the trees that are on top of the hill behind the quarry. They are generating too much fuzz, and makes the contour line software super confused.

Its the same with the mass calculation software, it includes the trees, which makes the projection accuracy low.

I have tried to classify points, but the result becomes even worse regarding the trees.

So if anyone have done a similar job, please share some tips, and I will be super gratefull.

Attaced is a screenshot of my classifications. (Please remember, im still pretty new to Photoscan)

General / Re: Not enough memory
« on: September 17, 2014, 11:11:26 AM »
Thanks, where too upload it?

Its pretty big though, 41GB

General / Re: Not enough memory
« on: September 12, 2014, 02:09:25 PM »

General / Re: Not enough memory
« on: September 08, 2014, 09:45:15 AM »
Hi Alexey.

I have checked memory status while the job was ongoing. And its not even close.
I can also verify that I'm using Photoscan 64bit



General / Re: Not enough memory
« on: September 05, 2014, 01:43:58 PM »
This table suggests otherwise:

Photos                     100   200             500   1000   2000   5000   10000
Lowest quality            25 MB   50 MB   125 MB   250 MB   500 MB   1.25 GB   2.5 GB
Low quality           100 MB   200 MB   500 MB   1 GB            2 GB   5 GB           10 GB
Medium quality            400 MB   800 MB   2 GB           4 GB           8 GB            20 GB   40 GB
High quality             1.6 GB   3.2 GB   8 GB           16 GB   32 GB   80 GB   160 GB
Ultra high quality   6.4 GB   12.8 GB   32 GB   64 GB   128 GB   320 GB   640 GB

General / Not enough memory
« on: September 05, 2014, 11:40:27 AM »

Im hoping you can point out what I need to upgrade on my computer.

I have started the build dense cloud process on a job which contains 356 pictures/cameras of 4928x3264 uncompressed TIF's. All pictures is shot by our drone.
Dense cloud settings are High quality, and agressive depth filtering.

I've tried twich and both times the job has failed due to not enough memory, but at different intervals.

In console it says:
[GPU] estimating 652x1883x128 disparity using 652x942x32f tiles, offset 0
timings: rectify: 0.118 disparity: 0.258 borders: 0.023 filter: 0.039 fill: 0
[GPU] estimating 813x2251x160 disparity using 813x751x32f tiles, offset 0
timings: rectify: 0.651 disparity: 0.742 borders: 0.066 filter: 0.066 fill: 0
Error: bad allocation
finished depth reconstruction in 2140.63 seconds
Device 1 performance: 362.593 million samples/sec (GeForce GTX 770)
Device 2 performance: 363.933 million samples/sec (GeForce GTX 770)
Total performance: 726.526 million samples/sec
Error: bad allocation
Finished processing in 2304.95 sec (exit code -1)

Some key specs about my computer

Intel I7 4770K
16GB 1600Mhz DDR3 RAM
2GB Geforce GTX770
Windows 7 Pro

My questions are:
Is there any settings in Photoscan that can remove this problem?
I have done a similar job with success, although with fewer pictures (around 200), but according to the table found on your Tips and tricks section 16GB should be more than enough when using Hight field.

Is it the RAM that needs upgrading or the GPU?
Should I go for a GTX770 SLI config or just make my boss buy TITAN (or maybe SLI Titan)


Marius Skancke

General / Re: GPU processing during the modeling and settings
« on: August 27, 2014, 02:39:52 PM »
Thanks Alexey.

Could it be a possibility to have a pinned post where the new drivers are verified as working or not working?

General / Not using GPU when building dence point cloud
« on: August 27, 2014, 02:07:40 PM »
Hi, I've just started using Photoscan, so please bear with me.

When Im trying to build dence point cloud, I can see in the console that it says:

oc1_engine.cpp line231: clEnqueueWriteBuffer failed, CL_MEM-OBJECT_ALLOCATION_FAILURE
GPU processing failed, switching to CPU mode

My initial thoughts goes to Driver issue.

Is there anyone here who can recommend a Nvidia driver that works well with Photoscan?

I have Nvidia GTX770 2GB,

Best Regards

Marius Skancke

Pages: [1]