/
SfM Lab

SfM Lab

Mar 31, 2016

Report

1.) One to two sentences explaining the object you took pictures of.

2.) Take screenshots (windows button + PrintScrn - images saved to Pictures) and include in your report.

3.) Show a screenshot of VSfM dense model reconstruction.

4.) Show a screenshot of point cloud in CloudCompare.

Part 0: Installing VisualSfM for Windows 10

1) Download and install Microsoft Visual C++ 2010 SP1

2) Download the VisualSFM_windows_64bit.zip and extract to root C

3) Download the bundler_sfm-master.zip from Github and extract to root C

4) Download the CMVS-PMVS-master.zip from Github and extract to root C

5) Navigate to the Win60-VS2010 folder and copy contents.

6) Paste contents into the C:\VisualSFM_windows_64bit folder.

7) Open VisualSfM.

Part 1: Take pictures of a subject

  1. Find a subject that you'd liked to generate a 3-D model from.

    1. We only have a couple hours in lab, so do not take a lot of photos or at very high resolution that will take hours to process

    2. Take as many photos as you can at this point - you can sort them out to pick the best ones later. 

    3. Remember, you're looking to get between 60% and 80% overlap between photos.

  2. Create a folder where you can store your images and your SfM project

  3. Copy all of the photos to a folder.

  4. Discard any blurry photos or photos that are very close or identical to other photos.

Part 2: Open Visual SfM

Follow the instructions here: http://ccwu.me/vsfm/doc.html for "3D Reconstruction - Using the VisualSFM GUI"

Basic instruction:

  1. Click the "File->Open Multi Images" folder with the + to open your images.

    1. It should take a minute or two for all of your images to load.

  2. Click the "SfM->Pairwise Matching->Compute Missing Match" four coloured arrows to begin matching the images.

    1. This should take between five and twenty minutes 

  3. Click the "SfM->Reconstruct Sparse" red and blue arrows to begin the Sparse Reconstruction

    1. You should be able to watch photos be added to the reconstruction in real time.

    2. Click the "BA" to run the multi-core bundel adjustment

  4. Click "Sfm->Reconstruct Dense"  red green and blue button CMVS.

    1. Save the files to a model directory.

    2. This could tak a long time.

Open in CloudCompare

Open CloudCompare and load the .ply file output from VsfM. 

You can also load the bundle.rd.out file.

More Visual SFM Examples

References

Balcomb Thesis 2015

Correcting Fish-eye distortion from a GoPro lens in Adobe PremierPro (Creative Cloud)

http://www.tricedesigns.com/2014/06/26/lens-correction-for-gopro-video-footage-has-never-been-easier/

SfM From YouTube videos

https://ryanfb.github.io/etc/2014/10/22/drone_photogrammetry_workflow.html

Create still frames from video

The camera recorded video on its flight in HD (1920x1080p, 29.97 fps).

I used Adobe Creative Cloud Premiere Pro to convert the M4P video to .JPEG images at 5 frames per second (fps) and the highest resolution quality setting.

Open the Premiere Pro

Add the .MP4 file to the Projects

Click on File > Export 

In Export Settings select:

Format: JPEG

Preset: Custom

Output Name: [Select Folder for Archive]

In Basic Settings select:

Make sure the video is in the highest quality setting

Frame Rate: 1 or 5 per second - this needs to be tweaked to ensure you get enough focused frames

Maximum Render Quality is checked.

Orthophoto mosaic

I used Adobe CC Photoshop to convert the .JPEG images into a photo mosaic using the add images to Stack

File > Scripts > Load files into Stack.

Or 

File > Automate > PhotoMerge

 

I also created digital surface models (DSM) equivalent to Orthomosaics in Visual-SfM using the dense point reconstruction (CMVS)

In Cloud Compare I opened the '.bundle.out' file