autoforge 1.9.0__tar.gz → 1.9.2__tar.gz

{autoforge-1.9.0 → autoforge-1.9.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: AutoForge
-Version: 1.9.0
+Version: 1.9.2
 Summary: AutoForge is a Python tool for generating 3D printed layered models from an input image.
 Author-email: Hendric Voss <hvoss@techfak.uni-bielefeld.de>
 License: CC BY-NC-SA 4.0

{autoforge-1.9.0 → autoforge-1.9.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "AutoForge"
-version="1.9.0"
+version="1.9.2"
 description = "AutoForge is a Python tool for generating 3D printed layered models from an input image."
 authors = [{ name = "Hendric Voss", email = "hvoss@techfak.uni-bielefeld.de" }]
 readme = "README.md"

{autoforge-1.9.0 → autoforge-1.9.2}/src/AutoForge.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: AutoForge
-Version: 1.9.0
+Version: 1.9.2
 Summary: AutoForge is a Python tool for generating 3D printed layered models from an input image.
 Author-email: Hendric Voss <hvoss@techfak.uni-bielefeld.de>
 License: CC BY-NC-SA 4.0

{autoforge-1.9.0 → autoforge-1.9.2}/src/autoforge/Helper/Heightmaps/ChristofidesHeightMap.py

@@ -1,4 +1,3 @@
-import os
 import random
 from typing import Optional
 
@@ -613,7 +612,8 @@ def run_init_threads(
     background_tuple,
     eps=1e-6,
     random_seed=None,
-    num_threads=64,
+    num_threads=4,
+    num_runs=32,
     init_method="kmeans",
     cluster_layers=None,
     material_colors=None,
@@ -625,26 +625,46 @@ def run_init_threads(
         random_seed = np.random.randint(1e6)
     lab_space = True
 
-    tasks = [
-        delayed(init_height_map)(
-            target,
-            max_layers,
-            h,
-            background_tuple,
-            eps,
-            random_seed + i,
-            init_method=init_method,
-            cluster_layers=cluster_layers,
-            lab_space=lab_space,
-            material_colors=material_colors,
-            focus_map=focus_map,
-            focus_boost=focus_boost,
-        )
-        for i in range(num_threads)
-    ]
-
-    # Execute tasks in parallel; adjust n_jobs to match your available cores
-    results = Parallel(n_jobs=os.cpu_count(), verbose=10)(tasks)
+    if num_threads > 1:
+
+        tasks = [
+            delayed(init_height_map)(
+                target,
+                max_layers,
+                h,
+                background_tuple,
+                eps,
+                random_seed + i,
+                init_method=init_method,
+                cluster_layers=cluster_layers,
+                lab_space=lab_space,
+                material_colors=material_colors,
+                focus_map=focus_map,
+                focus_boost=focus_boost,
+            )
+            for i in range(num_runs)
+        ]
+
+        # Execute tasks in parallel; adjust n_jobs to match your available cores
+        results = Parallel(n_jobs=num_threads, verbose=10)(tasks)
+
+    else:
+        results = [
+            init_height_map(
+                target,
+                max_layers,
+                h,
+                background_tuple,
+                eps,
+                random_seed + i,
+                init_method=init_method,
+                cluster_layers=cluster_layers,
+                lab_space=lab_space,
+                material_colors=material_colors,
+                focus_map=focus_map,
+                focus_boost=focus_boost,
+            ) for i in range(num_runs)
+        ]
 
     metrics = [(r[2] / r[3]) / (r[4] + 1e-6) for r in results]
     mean_metric = np.mean(metrics)

{autoforge-1.9.0 → autoforge-1.9.2}/src/autoforge/auto_forge.py

@@ -15,6 +15,7 @@ The implementation intentionally keeps side-effects (disk writes / prints) order
 preserve prior behavior. Helper functions are factored out for readability; no functional
 behavior should have changed relative to the previous monolithic version.
 """
+
 import argparse
 import sys
 import os
@@ -154,7 +155,6 @@ def parse_args() -> argparse.Namespace:
 
     parser.add_argument(
         "--visualize",
-        type=bool,
         default=True,
         help="Enable visualization during optimization",
         action=argparse.BooleanOptionalAction,
@@ -435,9 +435,7 @@ def _auto_select_background_color(
                 f.write(f"chosen_filament_color={chosen_hex}\n")
                 f.write(f"closest_filament_index={closest_idx}\n")
                 if getattr(args, "background_material_name", None):
-                    f.write(
-                        f"closest_filament_name={args.background_material_name}\n"
-                    )
+                    f.write(f"closest_filament_name={args.background_material_name}\n")
         except Exception:
             traceback.print_exc()
     else:
@@ -447,7 +445,10 @@ def _auto_select_background_color(
 
 
 def _prepare_background_and_materials(
-    args, device: torch.device, material_colors_np: np.ndarray, material_TDs_np: np.ndarray
+    args,
+    device: torch.device,
+    material_colors_np: np.ndarray,
+    material_TDs_np: np.ndarray,
 ) -> Tuple[Tuple[int, int, int], torch.Tensor, torch.Tensor, torch.Tensor]:
     """Create torch tensors for materials & background color.
 
@@ -566,11 +567,12 @@ def _initialize_heightmap(
                 args.layer_height,
                 bgr_tuple,
                 random_seed=random_seed,
-                num_threads=args.num_init_rounds,
+                num_threads=4,
                 init_method="kmeans",
                 cluster_layers=args.num_init_cluster_layers,
                 material_colors=material_colors_np,
                 focus_map=None,
+                num_runs=args.num_init_rounds,
             )
         )
     return pixel_height_logits_init, global_logits_init, pixel_height_labels
@@ -648,7 +650,9 @@ def _build_optimizer(
     return optimizer
 
 
-def _run_optimization_loop(optimizer: FilamentOptimizer, args, device: torch.device) -> None:
+def _run_optimization_loop(
+    optimizer: FilamentOptimizer, args, device: torch.device
+) -> None:
     """Execute the main gradient-based optimization iterations.
 
     Features:
@@ -729,26 +733,6 @@ def _post_optimize_and_export(
     optimizer.pixel_height_labels = torch.tensor(
         pixel_height_labels, dtype=torch.int32, device=device
     )
-    if hasattr(optimizer, "height_residual"):
-        with torch.no_grad():
-            hr = optimizer.height_residual.detach()
-            full_h, full_w = optimizer.pixel_height_logits.shape[:2]
-            if hr.shape != (full_h, full_w):
-                hr4d = hr.unsqueeze(0).unsqueeze(0)
-                hr_full = (
-                    torch.nn.functional.interpolate(
-                        hr4d,
-                        size=(full_h, full_w),
-                        mode="bilinear",
-                        align_corners=False,
-                    )
-                    .squeeze(0)
-                    .squeeze(0)
-                )
-                optimizer.height_residual.data = hr_full.to(device)
-            optimizer.best_params["height_residual"] = (
-                optimizer.height_residual.detach().clone()
-            )
     if focus_map_proc is not None and focus_map_full is not None:
         optimizer.focus_map = focus_map_full
 
@@ -932,8 +916,10 @@ def start(args) -> float:
     )
 
     # Prepare background color tensor and material tensors
-    bgr_tuple, background, material_colors, material_TDs = _prepare_background_and_materials(
-        args, device, material_colors_np, material_TDs_np
+    bgr_tuple, background, material_colors, material_TDs = (
+        _prepare_background_and_materials(
+            args, device, material_colors_np, material_TDs_np
+        )
     )
 
     # Compute sizes
@@ -951,12 +937,14 @@ def start(args) -> float:
     focus_map_full = _load_priority_mask(args, output_img_np, device)
 
     # Initialize heightmap
-    pixel_height_logits_init, global_logits_init, pixel_height_labels = _initialize_heightmap(
-        args,
-        output_img_np,
-        bgr_tuple,
-        material_colors_np,
-        random_seed,
+    pixel_height_logits_init, global_logits_init, pixel_height_labels = (
+        _initialize_heightmap(
+            args,
+            output_img_np,
+            bgr_tuple,
+            material_colors_np,
+            random_seed,
+        )
     )
 
     # Prepare processing targets and focus map (processing-res)