I was reading a paper titled
A single-image linear calibration method for camera
https://www.sciencedirect.com/science/article/pii/S0263224118307243

been having trouble finding (printing, and buying) checker boards so we changed our program to read charuco.
Doing charuco intrinsic and an aruco extrinsic. At first we were printing them off our printer and everything was unicorns and rainbows, all the numbers were very close to what we have physically measured and/or calculated. Had some professionally printed that turned out to be trash. The problem now is when we print them in house our fx fy is off which is causing our z distance to be off several mm from actual measured distance. nothing has changed in the program or process since starting to use charuco besides using newly printed boards (same printer, same paper, same brand of ink, printing from the same file and program). We now just cant get a “good” board. Looking for any insights, ideas, theories, just something that may help us get back to a board that is in our tolerance. What could we be overlooking? Included a couple images of our matrix we find that in the 1620-1630 range we get a z distance closer to measured[enter image description here](https://i.sstatic.net/oT8wnFcA.jpg)