asperry In the Configuring Stereo Depth tutorial, it mentions that there exists a trade-off between "fill-rate" and accuracy when choosing a stereo depth confidence threshold. What exactly does "fill-rate" refer to? It seems to me that the word "rate" is perhaps being used incorrectly here and maybe it should be "fill density"? I'm not sure how the concept of time (implied by the use of the word "rate") would factor into my choice of a depth confidence threshold.
jakaskerl Hi @asperry I think the term was used from a viewpoint of the algorithm iterating through the image pixel-by-pixel. So a higher fill-rate would mean the pixel gets accepted more often than when using high accuracy mode where the opposite is true. Perhaps frequency would be a better term to use. Thanks, Jaka
asperry Thanks for the answer @jakaskerl. But to confirm, my choice of stereo depth confidence threshold does not affect performance for producing the disparity/depth frames, correct? The algorithm still has to check the threshold for every pixel.
