Troubleshooting

• poetry install does not work First, ensure that poetry is installed correctly as descibed on the [Poetry Website](https://python-poetry.org/docs/).Secondly, ensure the correct python version is installed on your system, as specified in the file pyproject.tomlThird, this can be caused by a proxy which pip does not handle correctly. Manually setting the proxy in the Windows settings or even adding a system variable https_proxy = http://YOUR_PROXY:PORT can resolve this.

• Decoding takes a long time

  This is most likely due to the just-in-time compiler Numba, which is used for this computationally expensive function: During the first execution, an initial compilation is executed. This can take several tens of seconds up to single digit minutes, depending on your CPU and energy settings. However, for any subsequent execution, the compiled code is cached and the code of the function runs much faster, approaching the speeds of code written in C.

• My decoded coordinates show lots of noise

  • Make sure the exposure of your camera is adjusted so that the fringe patterns show up with maximum contrast. Avoid under- and overexposure during acquisition.

  • Try using more, sets K and/or shifts N.

  • If the decoded modulation is much lower than the decoded brightness, try to use larger wavelengths l resp. smaller number of periods v.

  • If the decoded modulation remains low even with very large wavelengths (less than five periods per screen length), and you are conducting a deflectometric mesurement, the surface under test is probably too rough. Since deflectometry is for specular and glossy surfaces only, it isn’t suited for scattering ones. You should consider a different measurement technique, e.g. fringe projection.

• My decoded coordinates show systematic offsets

  1. Ensure that the unambiguous measurement range is larger than the pattern length, i.e. UMR \(\ge\) L. If not, adjust the used wavelengths l resp. number of periods v accordingly.

  2. Ensure that the correct frames were captured while acquiring the fringe pattern sequence. If the timings are not set correctly, the sequence may be a frame off.

  3. This might occur if either the camera or the display used have a gamma value very different from 1.

  1. Typically, screens have a gamma value of 2.2; therefore compensate by setting the inverse value \(\gamma^{-1} = 1 / 2.2 \approx 0.45\) to the gamma attribute of the Fringes instance. Alternatively, change the gamma value of the light source or camera directly.

  2. You can use the static method gamma_auto_correct() to automatically estimate and apply the gamma correction factor to linearize the display/camera response curve.

  3. You might also use more shifts N to compensate for the dominant harmonics of the gamma-nonlinearities.