The Heal node can reconstruct damaged, low res, or 8-bit data with 16-bit, high resolution fidelity. The Heal node can take quantized images with extreme banding of any shape, and clean up the terrain into a naturalistic output.
Heal can consume 8-bit data, (visually) damaged images, or low quality satellite data - even with bad JPEG artifacts - and turn it into a reasonably clean heightfield. It is also useful for upscaling lower resolution data such as Google map extracts.
Obviously, not all data can be reconstructed, but you can get a fair bit of quality back. Below is an example where we downsample a 16-bit heightfield to 8-bit, and then reconstruct it using the Heal node.
Properties
| Heal | |
| Reconstruction | Controls how aggressively missing or damaged areas are rebuilt from surrounding data (higher values preserve continuity but can soften detail). |
| Passes | Number of healing iterations to run; more passes produce a cleaner, more continuous surface but may blur fine detail. |
using AngryMonkey 