Blender Motion Tracking Solving Camera.

Support Technical Support
1

Hi fellow artists.

I do a lot of motion tracking with the integrated tomato tracker.

I’ve ran into some problems where I need some help.

I have a clip starting from frame 1 and ends at frame 130.
And a solve of 0.2380 which is almost perfect.

But still the camera solve is sloppy, even with reconstructing it doesn’t help with the use of origin and set floor.

Does anyone know how to make a good camera solve? in a case like this.

This footage did work for me (made by me) http://www.youtube.com/watch?v=x9B6T0T7ci0&feature=BFa&list=UUYRS9H6A89ru-DRDbTL64HQ&lf=plcp

here are some images to visualize.

Attachments

Region capture 2.jpg1440×839 271 KB

Region capture 4.png1440×841 49.4 KB

Region capture 1.png1440×900 259 KB

2

Well it does seem strange that you’re getting such a good error rate but your solve is not working. The only thing that jumps out at me from your screenshots is that the sensor size is set at 35mm, which is unlikely unless your footage is from a cine camera or one bada$$ dslr. Finding out a truer value there might help. You also don’t have a lot of trackers by the looks of things. A few more might also tidy the solve up for you.

3

Thank you for your reply. I calculated the sensor size by this math Shot footage 1920 x 1080, image size 1300 x 731.

1920/1300 =1.47. is this correct. because my footage was shot with a Sony Cybershot and I can’t find the exact specs I want. Like the sensor size.

I added more tracking points but the error doubled to 0.6.

I also want to put an emphasis on this image. where the tracking points don’t match the image.

Check above (internet is being funky).

4

Thank you for your reply. I calculated the sensor size by this math Shot footage 1920 x 1080, image size 1300 x 731.
1920/1300 =1.47. is this correct.
Absolutely not. Depending on the camera model the sensor width may only be 5-6mm

5

Alright, thanks.

I fixed it between those values.

Could you please take a look at the above images, the middle one.

Because the tracking points do not match the footage.