Working with one of our openSFDI systems, we were getting some strange results related to demodulation of the AC images. Sometimes the demodulation would work well, other times there would be stripes in the image at the same spatial frequency as the projection.
On examination of the data, we found that one of the AC images had a higher DC offset than the other two, which is what was causing stripes in the demodulated images.. What seems to be happening is that if any of the images have regions of saturation (pixel counts of 65530) none of the wavelengths would demodulate properly. Reducing the exposure time of the offending wavelength such that there was no saturation in any of the images solved this problem.
A more permanent solution might be to change the order that images are acquired. The way the control code is currently designed, it displays a pattern on the DMD and collects images for each wavelength before switching patterns. Changing it so all the patterns for a wavelength are collected before the wavelength changes, might prevent a single image with saturation from corrupting the entire dataset.
I’m not sure if this is an issue with the camera, the software, or something else. Let me know if you have thoughts in the comments (or via e-mail).