File size: 1,389 Bytes
6e204a5 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | """
Unit tests for Raster Diffusion, Renderer, and VQ-VAE Floorplan Previews.
"""
import numpy as np
import torch
import pytest
from floorgen.models.diffusion_core.raster_diffusion import (
VectorToRasterRenderer,
VectorConditionedVQVAE,
generate_raster_preview
)
def test_vector_to_raster_renderer():
sample_plan = {
"rooms": [
{"category": "living_room", "box": [20, 20, 120, 100]},
{"category": "master_bedroom", "box": [120, 20, 200, 100]},
{"category": "bathroom", "box": [120, 100, 180, 150]}
],
"doors": [{"pos": [120, 50]}],
"boundary": [20, 20, 200, 150]
}
renderer = VectorToRasterRenderer(canvas_size=256)
img = renderer.render(sample_plan)
assert isinstance(img, np.ndarray)
assert img.shape == (256, 256, 3)
assert img.dtype == np.uint8
# Should not be a blank white canvas
assert np.any(img != 255)
# Test convenience function
preview = generate_raster_preview(sample_plan)
assert preview.shape == (256, 256, 3)
def test_vqvae_forward_pass():
vqvae = VectorConditionedVQVAE(in_channels=3, latent_dim=32, num_embeddings=64)
dummy_img = torch.rand(2, 3, 256, 256)
recon, quantized = vqvae(dummy_img)
assert recon.shape == (2, 3, 256, 256)
assert quantized.shape == (2, 32, 32, 32)
assert not torch.isnan(recon).any()
|