Bingo export

[Erik_Vest]

We are the moment evaluating this great software with special attention on the Bingo export format and in that respect have a request for us to be able to use it. Is it somehow possible to also include the photo measurements in a separate “Bingo” file when you use the Bingo export tool? The required Bingo export format is kind of similar to the bundler format although the view list should be structured a little different and in a different file.
 
Photo measurement should be given in a long list where the first line states the image name without extension of the first image and then the camera name. The following lines should state all the observations in that image in the second and third column while the first column states the linkage with the other images (see illustration below).

So for example point 236-00001_3 has been measured in image 236b-00001 with coordinate -12.781  15.569 while it has been measured in image 236b-00002 with coordinate -14.281  3.924.

Hope this makes sense. Otherwise I could send some example files if you think you could include this possibility in the software.
 
 <236b-00001>      <RCD105            (first camera/image)
236b-00001_3   -12.781  15.569
236b-00001_8   14.610  15.453
236b-00001_9   20.247  15.528
236b-00001_11   -14.997  13.236
236b-00002_34   20.294  13.869
236b-00002_43   -14.278  -0.753
236b-00002_44   -4.366  0.355
<236b-00002>  <RCD105>         (second camera/image)
236b-00001_2   -21.525  2.173
236b-00001_3   -14.281  3.924
236b-00001_4   -7.205  5.695
236b-00001_5   -8.347  6.674
236b-00001_6   2.655  5.498
236b-00002_57  -9.469  -9.925
236b-00002_58  -22.613  15.507
236b-00002_59  -10.679  -5.430
236b-00003_30  -21.865  10.460
236b-00003_31  4.605  10.194
236b-00003_32  15.220  7.323
.
.
And so on………….


Best regards,

Erik Vest S?rensen

[Alexey Pasumansky]

Hello Erik,

Will consider the possibility of such option in future updates. At the first look it seems that we do have all the necessary data, but if there's any format specification please send us the link.

Just a few questions:
- what is the "camera model"?
- are there any requirements for point naming (image_name + index, for example, or could it be just index)?

As an alternative I think it's possible to create quite simple Python script to perform such export, but it will be quite slow for huge point clouds, but if you are interested, please let me know.

[Erik_Vest]

Sounds promising although not a “yes” J I have attached a small text that describes the formats required for photo coordinates more explicitly. There is no requirement for point naming other than max. 15 characters. The example I have given you is from a project where the “point base name” had the same name as the image name + a number. This makes it easier if manually editing points. This is however not a must. Point names could be just numbers if that is easier for you (see the text). Please note that I forgot to state that last line of photo is indicated by -99. It would also be good to have one blank character in front of the point names as stated in the text.
Regarding camera model, we normally use frame cameras that have been calibrated using a test-field. Calibration and imaging is with fixed lenses focused and fixed at infinity. So we have good control of the distortion that is normally available as distortion pairs in mm or deg. If however distortion is given as K1. K2, K3, P1, P2 etc. as in Agisoft we can?t use it directly. We use a model with an additional K0 parameter. I have attached a document describing the model which could be a further nice thing to consider?. However if the raw image measurements are exported in the suggested Bingo format I think that would be useful for us and "slow" python scripts is okay. 

Cheers,

Erik 
 

[geologen]

Hi Alexey,

Just to follow up on my last post. Do you think it is something you will implement and if yes what is the approximately timeframe?

Cheers,

Erik

[Alexey Pasumansky]

Hello Erik,

I think it's possible (at least from Python). This week I'm on the Intergeo'13, but next week I'll be back in the office and will try to provide you a working script for output evaluation.

[geologen]

Great 

[Alexey Pasumansky]

Hello Eric,

Few questions regarding the export format:

1. should be coordinates of points in mm or in pixels? If the values should be in mm than it is required to fill up "sensor pixel size" field in the Camera Calibration window before using the script (in case this data is missing in the image EXIF data).

2. if the maximal number of symbols in camera_name+tie_point_number depend on the number of points in cloud and in some cases the full name will be longer than 15 symbols. What is the recommended way to deal with such cases? Omitting photo_name will be fine if there is huge amount of points in the cloud?

[geologen]

Hi Alexey,

Coordinates in mm is fine. If it is the raw observations without correction for lens distortion etc. that is even better 

It would be fine to omit photo_name in case of huge amount of points

Cheers,

Erik

[Alexey Pasumansky]

Hello Erik,

PMd you the script, let me know how it works.

[balyoa]

Hello,

I'm interested for the export of the tie-points (coordinates of images observation), i, bingo format or PAT-B format.

Is it of function of photoscan or a special script ?

Thanks in advance.

Best Regards.

[Alexey Pasumansky]

Yet another update: next version of PhotoScan will support the following tie points export formats - BINGO, PATB and ORIMA.

[Mihai Burlacu]

Hello,

I am interested in using the exported matches from PhotoScan. However, I need them in pixel coordinates and in order to do the conversion it is not clear to me what the coordinates in the BINGO file are referred to. Could anyone let me know how "raw" they are, whether:
1. they are corrected for distortion or not? (I would assume they are not)
2. they are referred to the principal point (in case I inputted precalibration values) or to the center of the sensor?

Enabling exporting in pixel coordinates would eliminate the ambiguities regarding the principal point.

Regards,
Mihai

[Alexey Pasumansky]

Hello Mihai,

Do you think that the custom script will solve the problem for you? If so, that what format do you expect?