Known cameral positions

[Karlsc]

>>

and you need to calibrate the setup once first

<<

how is that done in an open fixed location? I know how to calibrate a single camera but that doesn't help when they are separated by at least 5 meters.
the camera layout I have is set up so that any point has at least 4 cameras on it

there is no single original video. there are 8 separate videos each fairly heavy since it took some time to set up the cameras initially. I've given up pulling frames using the built in tool you guys have. It doesn't work very well: for example on a source I know to be 30fps I can only set to pull 25 FPS..

so I've moved to using FFMPEG and pulling a frame per frame. I know this means that the I-frames are being calculated and synthesized but when I inspect by hand I damned if I can tell which frame is a b-frame and which an I-frame

[Karlsc]

If your cameras are fixed, take a good dataset and export the camera internals and externals. You'll need Pro or Aerial in order to load photos from known parameters.

I'll be happy to help if you can send me the original data you are working with.

What do you mean by a "good dataset"? one where it has successfully constructed a model? Does that mean for the cameras individually? or in their fixed positions?

[Andrea Alessi]

Yes, you can save parameters of a successful reconstruction and use them for later reconstruction in a fixed rig scenario

[Karlsc]

Yes, you can save parameters of a successful reconstruction and use them for later reconstruction in a fixed rig scenario

right, but that doesn't help the initial fixed rig create a successful reconstruction
Is there a way to do that?

[Andrea Alessi]

Do you have a fixed rig? If so, simply reconstruct something by placing something easy to reconstruct, so you can save the parameters.

Once you have saved them use the load photos with known parameters (of your second scan) in pro or aerial. This will skip entirely the SfM phase and use the parameters of the previous project.

This is different than using constraints for cameras.

[Karlsc]

The rig is fixed to the boat itself. - I've been simulating it by putting the cameras on stands at fixed distances from the boat in an attempt to find the optimal configuration that will have all cameras included in the synthesis. if we take one side of the boat I'm currently looking at

1 camera on the bow looking aft

2 cameras on the stern (one each side offset by about 1m in my test and about 4 m in the full sized boat) looking forwards

1 camera midships offset from hull looking forwards diagonally

1 camera midships offset from hull looking aft diagonally

1 camera 3-4 meters above looking down.

each is a 4k 30fps "Sports" camera. I can add more cameras "within reason" since these are cheap. hence

I've been adding in the older "360 view" cameras in an attempt to find out if I need more cameras for synthesis.

The goal is to start with generating a 3D model statically and then use these cameras to capture dynamic frame by frame additions to create a Voxel based movie

so I have also tried using a 4k moving scan of the boat to create the baseline model. and then have the additional cameras added in. and use those as deltas against the baseline.

currently my testing model is a Laser so Front to Aft cameras are separated by about 4 meters

Side cameras are offset from centerline by about 2 meters See attached image layout

[Karlsc]

[Karlsc]

ignore the previous one

[Karlsc]

OK so I'm confused.. I created a fixed rig. and it still took me a while to get it to give me a good rendering. But the most recent run had me start with 23 cameras on the rig of which Zephyr light " recognized" 14"

So I pulled out those 14 images into a separate directory and re-ran Zephyr lite with exactly the same settings.. and this time it only recognized 5 of those 14 cameras.

And the Export Calibration exports a file of some sort, but it can't seem to be imported.

what am I doing wrong?

[Karlsc]

how do I get the system to reliably repeat recognizing the cameras for exactly the same images?

[Andrea Alessi]

Hi Karlsc,

you need to use the "load photos from known parameters function" which is available in 3DF Zephyr Pro and Aerial only. This is not possible in zephyr free or lite.

[Karlsc]

But if I present the tool with the same 14 images that it extracted the first time, shouldn't it logically extract the same structure from them as it did the first time? IE. I'm looking at repeatability

[Andrea Alessi]

I understand the question now, thank you for claryfing.

Yes, with the same settings you will obtain the same results. Zephyr is deterministic. (please note that if you set certain custom values and then switched to preset, zephyr will keep some of them, so make sure to reset the settings if this is the case).

However, due to floating point math approximation over GPU acceleration you may get slightly different results (which in turn, cause zephyr to fail) and this is usually due to bad photos. If you run it in CPU zephyr is 100% deterministic (albeit slower than in GPU),

[Karlsc]

I dunno what qualifies as bad photos. these are 4k images that are from static shots not video.