STMAP SOLVE
Hello,
I tracked two versions of the same footage: the original distorted footage and a version with an STMap applied.
For the original distorted footage, tracking worked well when I entered the actual focal length used during shooting.
However, when I tracked the STMap-applied footage and entered the same real focal length, the tracking points continuously showed sliding behavior and the solve was unstable.
After repeated issues, I retried the tracking by entering the field of view value obtained from the successfully tracked distorted footage. In this case, the focal length value changed to approximately 32, but the tracking became perfectly stable and accurate.
Although the practical issue is resolved, I do not fully understand the underlying algorithm.
Is it incorrect to use the real, physical focal length on STMap-undistorted footage?Or is there a specific conversion or recalculation that should be applied when using STMap-processed footage?
Thanks for uploading your screenshots, they made it much easier to see what’s going on. This doesn’t appear to be an STMap issue. The main problem is that the camera sensor size isn’t set.
Without a sensor size, a real focal length has no proper context, since both values are needed to calculate the correct field of view. If you enter a focal length alone, PFTrack assumes a default sensor size (full-frame). If your footage was shot on a smaller sensor or cropped mode, that assumption will be wrong and can cause inaccuracies.
When you enter the FOV directly, PFTrack simply calculates the focal length needed to match it, which is why the value changes to ~32mm and the tracking becomes stable. This also ‘might’ indicate that the camera has a smaller format sensor or the focal length has been noted down inaccurately.
The key takeaway is that when working with STMaps, accurate and matching camera and lens settings, especially sensor size and focal length, are essential for accurate results.