In the last article (Quantum Perlin Noise) we explored how quantum algo rithms, specifically Grover-Rudolph algorithm and Quantum Fourier Transform (QFT) interpolation, can generate smooth noise distributions in one dimension. This method could be a promising quantum alternative to classical Perlin noise, laying the foundations for a new era in procedural generation.
However, for the procedural creation of video-games worlds, textures and other visual environments, a single dimension is insufficient. The true power of procedural generation lies in its ability to construct planes and spaces. In this article, we take the next fundamental step: extending our quantum Perlin noise algorithm to two dimensions, demonstrating how we can move from generating a line of values to create a visually rich map.
The transition from one dimension to two dimensions introduces a new layer of complexity. While in one dimension the only concern is the smoothness between adjacent points in a line, in two dimensions we must be able to ensure gradual transitions in all directions: horizontal, vertical and diagonal.
Our quantum approach addresses the challenge in an elegant form. Instead of using one unique register to codify the position, we use two different registers: one for the X coordinate and the other one for Y coordinate. The key lies in how we handled these registers:
• Combined state preparation: The Grover-Rudolph algorithm is ap plied to the combined state of the X and Y registers. This situation creates an initial, non smoothed probability distribution across the entire 2D plane. Each basis state corresponds to a unique coordinate in our map (e.g: |010110⟩ corresponds to (X,Y ) = (2,6). An example of this non smoothed 2D distribution:
Figure 1: Non smoothed 2D distribution after GR algorithm
• Smoothing by each dimension: As we explained in the previous ar ticle, QFT interpolation is responsible for the smoothed transition. In the two dimensions case, we applied QFT interpolation independently to each dimensional register. This approach allow us to achieve smoothness between one pixel those above and below it.
Another important aspect is that the generalization to N dimension is straightforward! After applying the generalized Grover-Rudolph algo rithm, it is only needed to perform QFT interpolation independently on each of the N dimensional registers. This multidimensional scenario opens the possibility of create 3D worlds or even simulations that evolve smoothly over time, by designating one of these dimensional registers as the time dimension.
To implement this logic in the 2 dimension scenario we are going to use 10 qubits circuits, where we have two ancilla qubits and 3 register qubits for each coordinate. The circuit is:
Figure 2: Quantum Perlin Noise circuit
After executing this circuit and measuring the probabilities of each state, we obtain a matrix of values that represents our quantum noise in two dimensions. In the following figure we can observe a 32x32-pixel probability map in gray scale. The lighter areas represent the states (X,Y) with higher probability, and darker ones represent the lower probability states. The most important feature is the smoothness between each pixel, that is, a Perlin noise distribution obtained through quantum means.
Figure 3: Quantum Perlin Noise
The true potential of this noise map is revealed when we use it to generate a biome map, as in video-games. By assigning a color or terrain type to each probability threshold (e.g: lowest probability values correspond with deep water, intermediate values to meadows and forest, and the higher probability values correspond with snowy mountains), we obtain the following result:
Figure 4: Quantum Perlin Noise Biome
The figure above shows a biome map generated from our Perlin noise dis tribution. The result is an organic-looking landscape, with coasts, forest and mountains that looks natural, without abrupt transitions.
Wehave demonstrated successfully that quantum Perlin noise technique presented in the previous article can be extended form the one-dimensional case to multiple dimensions, paving the way for generation of maps, textures and other complex visual resources. This approach not only recreates the aesthetic qualities of Perlin noise, but does so using the fundamental principles of quantum computation, such as superposition and entanglement.
The path forward is exciting. Next steps could include extending the method to 3D maps, creating realistic clouds or other complex structures, incorporating a time dimension to create dynamic environments that can evolve, or testing the efficiency of this algorithm in a real quantum hardware. Quantum procedural generation is still in its infancy, but its potential to create virtual worlds that are richer, more complex, and more believable than ever is undeniable.
Do you have any questions?
Add new comment