Defocus & Distortion Grids

In this series of tests, we will answer the most popular questions about camera tracking that arise among CG artists.

The goal of the experiment #4 is to determine the maximum acceptable amount of blur/defocus when shooting a distortion grid. Yes, we recommend always stopping the aperture down as much as possible to achieve maximum depth of field, but in some cases even this may not be sufficient.

At the same time, in some cases – especially when working with tele lenses – it may be more practical to shoot certain focus distances slightly out of sharp focus rather than moving the camera.

Knowing the acceptable limits of grid blur will not only reduce the number of camera repositioning operations but also help optimize the overall distortion-grid shooting process.

In this test, we used a Sony Burano camera in Open Gate mode and two lenses: Sigma 50 mm and Supreme Prime 25 mm.
For each lens, we filmed the grid at different scale, gradually changing the focus distance so that the image started in sharp focus and then smoothly transitioned into defocus.
Next, each grid was split into frames with a step of several frames, starting from the “sharp” grid and progressing to full, heavy defocus.
Then, in 3DEqualizer, we performed automatic snapping until it was no longer possible.
After that, we applied our mm_grid_prepare gizmo to the “non-working” grids to see how many additional blur steps could still be extended while remaining suitable for automatic snapping.
Supreme Prime 25mm
The next level of blur makes snapping impossible, even with a processed grid. The matrix partially locks on, but most of it has to be adjusted manually.
Sigma 50mm
Creating the matrix is only possible through the use of 3DEqualizer’s automatic distortion calculation tool.
Snapping works fine

Next, creating the matrix is only possible through the use of 3DEqualizer’s automatic distortion calculation tool.

In the case of blur, this tool detects lines and builds at least some kind of matrix, thus allowing you to calculate distortion.

So here it’s basically a choice between having nothing at all or getting at least some distortion data.

Now we know the maximum acceptable amount of grid blur at which 3DEqualizer can still automatically snap the grid.

When shooting blurred and unevenly lit grids, you can use the mm_grid_prepare gizmo to improve snapping quality.

In cases of strong blur, you can use the automatic matrix placement tool in 3DEqualizer during snapping to obtain at least an approximate matrix (better something than nothing).

For quick access to the automatic distortion calculation algorithm, you can use this script.

Curious to dive deeper or run your own analysis?

All detailed data and test materials are available for download


← Return