Bit Depth, Color Space, and Resolution

< Previous | Contents | Manuals Home | Boris FX | Next >

‌Bit Depth, Color Space, and Resolution

Some care must be taken when creating image distortion maps (or references).

Eight-bit, ten-bit, and half-float image formats must not be used, as they are insufficiently precise to specify an exact pixel location . Sixteen-bit images are OK, but full floating-point images are best. (SynthEyes will only permit you to generate 16-bit or floating images, eliminating incompatible formats.)

Likewise, it is crucial to avoid any color-space processing when image maps are saved, and when the image map is read. (This may be impossible in Photoshop.)

Any color-space processing will create hidden distortions that may be very difficult to detect! Be sure to select the "linear" option.

In the distorted image above, note that the output image is no longer full frame (for illustration). The corrected image may also have tails in the corner. Generally the output image (and map) should be larger than the original undistorted image, so that there is no loss of effective resolution in the distortion and undistortion process. The SynthEyes Lens Workflow script manages this.

< Previous | Contents | Manuals Home | Boris FX | Next >