DreamTechAI · axisrow · Jun 13, 2025 · Jun 14, 2025 · Jun 14, 2025
diff --git a/config.yaml b/config.yaml
@@ -0,0 +1,15 @@
+vae:
+  target: direct3d.models.vae.VAE
+  params: {}
+dit:
+  target: direct3d.models.dit.DiT
+  params: {}
+semantic_encoder:
+  target: direct3d.models.condition.SemanticEncoder
+  params: {}
+pixel_encoder:
+  target: direct3d.models.condition.PixelEncoder
+  params: {}
+scheduler:
+  target: direct3d.models.condition.Scheduler
+  params: {}
diff --git a/direct3d.egg-info/PKG-INFO b/direct3d.egg-info/PKG-INFO
@@ -0,0 +1,11 @@
+Metadata-Version: 2.1
+Name: direct3d
+Version: 1.0.0
+Summary: Direct3D: Scalable Image-to-3D Generation via 3D Latent Diffusion Transformer
+Requires-Python: >=3.10
+License-File: LICENSE
+Requires-Dist: torch
+Requires-Dist: numpy
+Requires-Dist: cython
+Requires-Dist: trimesh
+Requires-Dist: diffusers
diff --git a/direct3d.egg-info/SOURCES.txt b/direct3d.egg-info/SOURCES.txt
@@ -0,0 +1,8 @@
+LICENSE
+README.md
+setup.py
+direct3d.egg-info/PKG-INFO
+direct3d.egg-info/SOURCES.txt
+direct3d.egg-info/dependency_links.txt
+direct3d.egg-info/requires.txt
+direct3d.egg-info/top_level.txt
diff --git a/direct3d.egg-info/dependency_links.txt b/direct3d.egg-info/dependency_links.txt
@@ -0,0 +1 @@
+
diff --git a/direct3d.egg-info/requires.txt b/direct3d.egg-info/requires.txt
@@ -0,0 +1,5 @@
+torch
+numpy
+cython
+trimesh
+diffusers
diff --git a/direct3d.egg-info/top_level.txt b/direct3d.egg-info/top_level.txt
@@ -0,0 +1 @@
+
diff --git a/direct3d/models/dit.py b/direct3d/models/dit.py
@@ -68,7 +68,8 @@ def get_2d_sincos_pos_embed(
     grid = np.stack(grid, axis=0)
 
     grid = grid.reshape([2, 1, grid_size[0], grid_size[1]])
-    pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid)
+    grid = torch.from_numpy(grid)
+    pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid, output_type='pt')
     if cls_token and extra_tokens > 0:
         pos_embed = np.concatenate([np.zeros([extra_tokens, embed_dim]), pos_embed], axis=0)
     return pos_embed
@@ -110,7 +111,7 @@ def __init__(
         pos_embed = get_2d_sincos_pos_embed(
             embed_dim, (self.height, self.width), base_size=(self.height, self.width), interpolation_scale=self.interpolation_scale
         )
-        self.register_buffer("pos_embed", torch.from_numpy(pos_embed).float().unsqueeze(0), persistent=False)
+        self.register_buffer("pos_embed", pos_embed.float().unsqueeze(0), persistent=False)
 
     def forward(self, latent):
         height, width = latent.shape[-2] // self.patch_size, latent.shape[-1] // self.patch_size

diff --git a/direct3d/pipeline.py b/direct3d/pipeline.py
@@ -156,4 +156,3 @@ def __call__(
         outputs = {"meshes": meshes, "latents": latents}
 
         return outputs
-
diff --git a/requirements.txt b/requirements.txt
@@ -6,4 +6,7 @@ einops
 numpy
 transformers==4.40.2
 diffusers
-rembg
+rembg
+omegaconf
+onnxruntime
+torchvision
Original file line number	Diff line number	Diff line change
Expand Up		@@ -156,4 +156,3 @@ def __call__(
		outputs = {"meshes": meshes, "latents": latents}

		return outputs