Przejdź do głównej zawartości

AI Upscaling

Heightcraft's AI Upscaling is a powerful feature that allows you to increase the resolution and fidelity of your heightmaps. It uses deep learning models trained on high-resolution terrain data to hallucinate plausible details when upscaling.

How it Works​

The upscaling pipeline performs two key actions:

  1. Super-Resolution: Increases the pixel count (resolution) by 2x, 3x, or 4x.
  2. Bit Depth Expansion: Intelligently interpolates values to convert 8-bit inputs into 16-bit or 32-bit outputs, reducing "stair-stepping" artifacts common in low-bit-depth terrains.

Usage​

To enable upscaling, use the --upscale flag.

Basic Upscaling​

heightcraft input.png --upscale --output_path output.png

Custom Factor​

By default, the upscale factor is 2. You can change it to 3 or 4:

heightcraft input.png --upscale --upscale_factor 4 --output_path output.png

Increasing Bit Depth​

A common workflow is to take a standard 8-bit image and upscale it to a 16-bit heightmap for game engines.

heightcraft input_8bit.png --upscale --bit_depth 16 --output_path output_16bit.png

Training Custom Models​

Heightcraft supports training custom upscaling models on your own datasets.

heightcraft --train --dataset_path /path/to/high_res_tiffs --epochs 50

This will generate a trained_model.h5 file that you can use with the --pretrained_model flag.

Arguments:

  • --dataset_path: Path to a folder containing high-res images (PNG, JPG, TIF).
  • --pretrained_model: Path to save the trained model (default: trained_model.h5).

Hyperparameters:

  • --epochs (Default: 10): How many times the AI sees the entire dataset.
    • Effect: Higher values (e.g., 50-100) generally lead to better quality but take longer. If set too high, the model might "memorize" the training data (overfitting) and perform poorly on new maps.
  • --batch_size (Default: 16): How many images are processed at once.
    • Effect: Larger batches (32, 64) speed up training but require more GPU memory. Smaller batches (4, 8) are slower but offer more stable updates.
  • --learning_rate (Default: 0.0001): How fast the AI adapts its internal weights.
    • Effect: A higher rate (e.g., 0.001) learns faster but might miss the optimal solution (unstable). A lower rate (e.g., 0.00001) is more precise but takes much longer to converge.