DLS1 requires that the earth-sensor angle and sensor-solar angle be well known to compensate for them. It is not possible to estimate direct and diffuse irradiance components from a single measurement, so, with a single sensor these components have to be estimated using a time series of data. These angles can be estimated with photogrammetry if the angles of the DLS sensor to camera are statically mounted to each other, but it still requires the appropriate complicated post processing to turn this into a time series estimate. If the DLS1 and camera are not statically mounted, the best accuracy we can get is from the IMU in the DLS. This accuracy is useful, if the system is installed and calibrated properly, but bad installations or poor calibration of the magnetometer can significantly impact the results.
DLS2 directly measures both the sun to sensor angle and the direct and diffuse irradiance components. It then uses these to compensate for these angles. Because these measurements are present real-time, we've been able to move the complicated post-processing into the camera firmware, and the irradiance estimates are written to the image metadata instead of simply writing raw measurements. This greatly reduces the need for post-processing and makes the measurements more accurate. In addition we have both irradiance components, so we are better suited to understand the light on various surfaces. the diffuse component allows us to estimate the light present in shadowed areas. We have also substantially improved the absolute radiometric accuracy of DLS2 over DLS1.