Lightcrafter Intrinsic Parameters

Hello Rafael,

Unfortunately the calibration data you've obtained is not correct. Thankfully I can tell you the issues that tell me the data is incorrect, but I'm not sure you'll be able to fix it with your calibration tool. 

The root cause of the incorrect data is that your calibration tool assumes the DMD has an orthogonal array when it actually has a diamond array. Please read the DLP3000 data sheet and reference Figure 3 to see what the array looks like physically.

The first sign of incorrect data is the focal length measurements. The lens used in the light engine for the LightCrafter have equal focal length in both x and y directions. Your measurement shows the y focal length as almost twice the x focal length which is incorrect.

The second sign of incorrect data is the principal point. The light engine used in the LightCrafter have 0% horizontal offset and about 100% vertical offset. So for a measurement I would expect the horizontal focal point to be half of the horizontal resolution and the vertical focal point to be the resolution. 

For example, the horizontal resolution of the DLP3000 DMD is 604 pixels and given the light engine design of 0% horizontal offset I would expect the principle point to be 302. Your measurement is 306, which is very close and is a number I would expect given manufacturing tolerances. 

The vertical resolution of the DLP3000 DMD is complicated because of the diamond pixel array. While the vertical pixel resolution is 684, its physical size in pixel units (compared to the horizontal resolution) is half of that equaling 342. Considering that the projection of the LightCrafter has an aspect ratio of about 16:10, it makes sense that the physical aspect ratio of the DMD is 604:342 which just about equals 16:10. 

To obtain correct calibration data you need to supply the calibration board feature point locations in physical space rather the pixel space. In pixel space the DLP3000 is more tall than it is wide, however in physical space it is more wide than it is tall.

As I said earlier, I'm not sure that your calibration tool will be able to account for the diamond array of the DMD. Thankfully, we are releasing a 3D Machine Vision reference design this week which does have the correct calibration procedures to accommodate the diamond pixel array! The reference design uses the LightCrafter 4500, but the calibration process is the same and I can help you modify the design if needed.

If I remember correctly I think your tool requires that you supply a chessboard image. I think that you are currently supplying a chessboard image that is 604x684 pixels to match the resolution of the LightCrafter. Try scaling the chessboard image to 604x342 and running the calibration again. 

Hope this helps!

Best regards,

Blair