Release/v0.2.0

.github/workflows/release.yml

@@ -9,8 +9,26 @@ permissions:
   contents: write
 
 jobs:
+  # ─── Create a single release ────────────────────────────────────────────────
+  create-release:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Create release
+        run: |
+          gh release create ${{ github.ref_name }} \
+            --title "Modly Beta ${{ github.ref_name }}" \
+            --generate-notes \
+            --latest \
+            --repo ${{ github.repository }}
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
   # ─── Windows build ────────────────────────────────────────────────────────────
   build-windows:
+    needs: create-release
     runs-on: windows-latest
     steps:
       - name: Checkout
@@ -29,18 +47,26 @@ jobs:
         run: node scripts/download-python-embed.js
 
       - name: Build & Package (Windows)
+        shell: bash
         run: npx electron-vite build && npx electron-builder --win --publish never
         env:
           CSC_IDENTITY_AUTO_DISCOVERY: false
 
-      - name: Upload Windows artifact
-        uses: actions/upload-artifact@v4
-        with:
-          name: windows-installer
-          path: dist/*.exe
+      - name: Upload assets to release
+        shell: bash
+        run: |
+          gh release upload ${{ github.ref_name }} \
+            dist/*.exe \
+            dist/*.exe.blockmap \
+            dist/latest.yml \
+            --repo ${{ github.repository }} \
+            --clobber
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
   # ─── Linux build ─────────────────────────────────────────────────────────────
   build-linux:
+    needs: create-release
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
@@ -55,42 +81,15 @@ jobs:
       - name: Install dependencies
         run: npm install
 
-      - name: Install Linux build dependencies
-        run: sudo apt-get install -y rpm
-
       - name: Build & Package (Linux)
         run: npx electron-vite build && npx electron-builder --linux --publish never
 
-      - name: Upload Linux artifact
-        uses: actions/upload-artifact@v4
-        with:
-          name: linux-installer
-          path: dist/*.AppImage
-
-  # ─── Create GitHub Release ───────────────────────────────────────────────────
-  release:
-    needs: [build-windows, build-linux]
-    runs-on: ubuntu-latest
-    steps:
-      - name: Download Windows artifact
-        uses: actions/download-artifact@v4
-        with:
-          name: windows-installer
-          path: artifacts/windows
-
-      - name: Download Linux artifact
-        uses: actions/download-artifact@v4
-        with:
-          name: linux-installer
-          path: artifacts/linux
-
-      - name: Create GitHub Release
+      - name: Upload assets to release
         run: |
-          gh release create ${{ github.ref_name }} \
-            artifacts/windows/*.exe \
-            artifacts/linux/*.AppImage \
-            --title "Modly Beta ${{ github.ref_name }}" \
-            --generate-notes \
-            --repo ${{ github.repository }}
+          gh release upload ${{ github.ref_name }} \
+            dist/*.AppImage \
+            dist/latest-linux.yml \
+            --repo ${{ github.repository }} \
+            --clobber
         env:
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}

api/main.py

@@ -5,7 +5,8 @@
 from contextlib import asynccontextmanager
 from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.staticfiles import StaticFiles
+from fastapi.responses import FileResponse
+from fastapi import HTTPException
 
 from routers import generation, model, optimize, status, settings, extensions, export
 
@@ -22,7 +23,7 @@ async def lifespan(app: FastAPI):
 
 app = FastAPI(
     title="Modly API",
-    version="0.1.3",
+    version="0.2.0",
     lifespan=lifespan,
 )
 
@@ -41,6 +42,11 @@ async def lifespan(app: FastAPI):
 app.include_router(extensions.router, prefix="/extensions")
 app.include_router(export.router,     prefix="/export")
 
-# Serve generated GLB files from workspace
-from services.generator_registry import WORKSPACE_DIR
-app.mount("/workspace", StaticFiles(directory=str(WORKSPACE_DIR)), name="workspace")
+# Serve generated files from workspace — dynamic so path changes take effect immediately
+@app.get("/workspace/{full_path:path}")
+async def serve_workspace_file(full_path: str):
+    import services.generator_registry as reg
+    file_path = reg.WORKSPACE_DIR / full_path
+    if not file_path.exists() or not file_path.is_file():
+        raise HTTPException(status_code=404, detail="File not found")
+    return FileResponse(str(file_path))

api/routers/generation.py

@@ -4,7 +4,7 @@
 import uuid
 from typing import Dict
 from fastapi import APIRouter, File, Form, UploadFile, HTTPException, BackgroundTasks
-from services.generators.base import smooth_progress
+from services.generators.base import smooth_progress, GenerationCancelled
 
 import re as _re
 from services.generator_registry import generator_registry, WORKSPACE_DIR
@@ -13,6 +13,8 @@
 router = APIRouter(tags=["generation"])
 
 _jobs: Dict[str, JobStatus] = {}
+_cancelled: set = set()
+_cancel_events: Dict[str, threading.Event] = {}
 
 
 @router.post("/from-image")
@@ -71,6 +73,7 @@ async def generate_from_image(
 
     job = JobStatus(job_id=job_id, status="pending", progress=0)
     _jobs[job_id] = job
+    _cancel_events[job_id] = threading.Event()
 
     background_tasks.add_task(_run_generation, job_id, image_bytes, params, collection)
 
@@ -86,6 +89,19 @@ async def job_status(job_id: str):
     return job
 
 
+@router.post("/cancel/{job_id}")
+async def cancel_job(job_id: str):
+    job = _jobs.get(job_id)
+    if not job:
+        raise HTTPException(404, f"Job {job_id} not found")
+    _cancelled.add(job_id)
+    if job_id in _cancel_events:
+        _cancel_events[job_id].set()
+    if job.status in ("pending", "running"):
+        job.status = "cancelled"
+    return {"cancelled": True}
+
+
 async def _run_generation(job_id: str, image_bytes: bytes, params: dict, collection: str = "Default") -> None:
     job = _jobs[job_id]
     job.status = "running"
@@ -118,20 +134,36 @@ def progress_cb(pct: int, step: str = "") -> None:
         else:
             gen = await loop.run_in_executor(None, generator_registry.get_active)
 
+        if job_id in _cancelled:
+            return
+
         # Direct output to the collection subfolder
         coll_dir = WORKSPACE_DIR / collection
         coll_dir.mkdir(parents=True, exist_ok=True)
         gen.outputs_dir = coll_dir
 
+        cancel_event = _cancel_events.get(job_id)
+        import inspect
+        supports_cancel = "cancel_event" in inspect.signature(gen.generate).parameters
         output_path = await loop.run_in_executor(
             None,
-            lambda: gen.generate(image_bytes, params, progress_cb),
+            lambda: gen.generate(image_bytes, params, progress_cb, cancel_event)
+                    if supports_cancel
+                    else gen.generate(image_bytes, params, progress_cb),
         )
+
+        if job_id in _cancelled:
+            return
+
         job.status     = "done"
         job.progress   = 100
         job.output_url = f"/workspace/{collection}/{output_path.name}"
 
+    except GenerationCancelled:
+        job.status = "cancelled"
     except Exception as exc:
+        if job_id in _cancelled:
+            return
         tb = traceback.format_exc()
         print(f"[Generation ERROR] {exc}\n{tb}")
         job.status = "error"

api/routers/model.py

@@ -39,6 +39,23 @@ async def switch_model(model_id: str):
         raise HTTPException(400, str(e))
 
 
+@router.post("/unload-all")
+async def unload_all_models():
+    """Unloads all models from memory to free VRAM/RAM."""
+    generator_registry.unload_all()
+    # Force Python to release memory back to the OS
+    import gc
+    gc.collect()
+    try:
+        import ctypes, sys
+        if sys.platform == "win32":
+            k32 = ctypes.windll.kernel32
+            k32.SetProcessWorkingSetSizeEx(k32.GetCurrentProcess(), -1, -1, 0)
+    except Exception:
+        pass
+    return {"unloaded": True}
+
+
 @router.post("/unload/{model_id}")
 async def unload_model(model_id: str):
     """Unloads a model from memory so its files can be safely deleted."""
@@ -51,19 +68,30 @@ async def unload_model(model_id: str):
 
 
 @router.get("/hf-download")
-async def hf_download(repo_id: str, model_id: str):
+async def hf_download(repo_id: str, model_id: str, skip_prefixes: Optional[str] = None):
     """
     Streams a HuggingFace Hub model download via SSE.
     Downloads into MODELS_DIR / model_id applying the filtering
     declared in the extension manifest (hf_skip_prefixes).
 
+    skip_prefixes: JSON-encoded list of path prefixes to exclude (passed from Electron).
+    Falls back to registry manifest if not provided.
+
     SSE format: data: {"percent": 0-100, "file": "...", "status": "..."}
     """
+    import json as _json
     dest_dir  = str(MODELS_DIR / model_id)
-    try:
-        skip_list = generator_registry.get_manifest(model_id).get("hf_skip_prefixes", [])
-    except KeyError:
-        skip_list = []
+    # Prefer skip_prefixes passed directly from the client (authoritative, no registry dep)
+    if skip_prefixes:
+        try:
+            skip_list = _json.loads(skip_prefixes)
+        except Exception:
+            skip_list = []
+    else:
+        try:
+            skip_list = generator_registry.get_manifest(model_id).get("hf_skip_prefixes", [])
+        except KeyError:
+            skip_list = []
 
     async def stream():
         loop = asyncio.get_running_loop()
@@ -105,7 +133,7 @@ def _dl(f=filename):
 
                 # Reserve 1-95 for file downloads, leave 95-100 for finalisation
                 pct = 1 + round((i + 1) / total * 94)
-                yield _fmt({"percent": pct, "file": filename})
+                yield _fmt({"percent": pct, "file": filename, "fileIndex": i + 1, "totalFiles": total})
 
             yield _fmt({"percent": 100, "status": "done"})
 

api/schemas/generation.py

@@ -4,7 +4,7 @@
 
 class JobStatus(BaseModel):
     job_id: str
-    status: Literal["pending", "running", "done", "error"]
+    status: Literal["pending", "running", "done", "error", "cancelled"]
     progress: int = 0              # 0–100
     step: Optional[str] = None    # Human-readable current step
     output_url: Optional[str] = None

api/services/generator_registry.py

@@ -327,6 +327,7 @@ def update_paths(self, models_dir: Optional[Path], workspace_dir: Optional[Path]
         import services.generator_registry as _self_module
 
         if models_dir is not None:
+            self.unload_all()
             models_dir.mkdir(parents=True, exist_ok=True)
             _self_module.MODELS_DIR = models_dir
             for model_id, gen in self._generators.items():

api/services/generators/base.py

@@ -7,6 +7,10 @@
 from typing import Callable, Optional
 
 
+class GenerationCancelled(Exception):
+    """Raised by generators when a cancel_event is set mid-generation."""
+
+
 def smooth_progress(
     progress_cb: Callable[[int, str], None],
     start: int,
@@ -73,6 +77,25 @@ def load(self) -> None:
     def unload(self) -> None:
         """Release memory. Can be overridden if needed."""
         self._model = None
+        import gc
+        gc.collect()
+        try:
+            import torch
+            if torch.cuda.is_available():
+                torch.cuda.empty_cache()
+        except ImportError:
+            pass
+        # Force the OS to reclaim unused memory from this process
+        try:
+            import ctypes
+            import sys
+            if sys.platform == "win32":
+                kernel32 = ctypes.windll.kernel32
+                kernel32.SetProcessWorkingSetSizeEx(
+                    kernel32.GetCurrentProcess(), -1, -1, 0
+                )
+        except Exception:
+            pass
 
     def is_loaded(self) -> bool:
         return self._model is not None
@@ -87,14 +110,21 @@ def generate(
         image_bytes: bytes,
         params: dict,
         progress_cb: Optional[Callable[[int, str], None]] = None,
+        cancel_event: Optional[threading.Event] = None,
     ) -> Path:
         """
         Starts 3D generation from an image.
         Returns the path to the generated .glb file.
         progress_cb(percent: int, step_label: str)
+        cancel_event: set this to interrupt generation between steps.
         """
         ...
 
+    def _check_cancelled(self, cancel_event: Optional[threading.Event]) -> None:
+        """Raises GenerationCancelled if cancel_event is set."""
+        if cancel_event and cancel_event.is_set():
+            raise GenerationCancelled()
+
     # ------------------------------------------------------------------ #
     # Parameter schema (for the UI)
     # ------------------------------------------------------------------ #

api/services/generators/hunyuan3d.py

@@ -69,15 +69,6 @@ def load(self) -> None:
         self._model = pipeline
         print(f"[Hunyuan3DGenerator] Loaded on {device}.")
 
-    def unload(self) -> None:
-        super().unload()
-        try:
-            import torch
-            if torch.cuda.is_available():
-                torch.cuda.empty_cache()
-        except ImportError:
-            pass
-
     # ------------------------------------------------------------------ #
     # Inference
     # ------------------------------------------------------------------ #