fix: catch all exceptions when importing cupy, not just ImportError

An installed-but-broken cupy (e.g. incompatible with NumPy 2.5, which
removed the 'bool8' alias) raises a TypeError during its own import, not
an ImportError. The narrow `except ImportError` guard let that propagate
and crashed the entire node import chain.

Broaden the guard to `except Exception` in all three CUDA-kernel modules
so any import-time failure disables cupy and falls back to the
pure-PyTorch implementations.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

bim_vfi_arch/costvol.py

@@ -4,6 +4,7 @@ import collections
 import os
 import re
 import torch
+import torch.nn.functional as F
 import typing
 
 cupy = None
@@ -14,12 +15,11 @@ def _ensure_cupy():
         try:
             import cupy as _cupy
             cupy = _cupy
-        except ImportError:
-            raise RuntimeError(
-                "cupy is required for BIM-VFI. Install it with:\n"
-                "  pip install cupy-cuda12x  (or cupy-cuda11x for CUDA 11)\n"
-                "Or run install.py from the ComfyUI-Tween directory."
-            )
+        except Exception:
+            # Broad catch: an installed-but-broken cupy (e.g. incompatible
+            # NumPy) raises non-ImportError exceptions at import time. Treat any
+            # failure as "cupy unavailable"; the PyTorch fallback will be used.
+            pass
 
 
 ##########################################################
@@ -246,6 +246,44 @@ def cuda_launch(strKey:str):
 # end
 
 
+def _pytorch_costvol_impl(tenOne, tenTwo, intKernelSize):
+    """Pure-PyTorch local cost volume via unfold + dot product."""
+    B, C, H, W = tenOne.shape
+    pad = (intKernelSize - 1) // 2
+
+    # Pad tenTwo so out-of-bounds yields 0 (matches CUDA kernel)
+    tenTwo_padded = F.pad(tenTwo, [pad, pad, pad, pad])
+
+    # Unfold into patches: (B, C, H, W, K, K)
+    patches = tenTwo_padded.unfold(2, intKernelSize, 1).unfold(3, intKernelSize, 1)
+    # Reshape to (B, C, H, W, K*K)
+    patches = patches.contiguous().view(B, C, H, W, intKernelSize * intKernelSize)
+
+    # Dot product over C dimension: (B, H, W, K*K)
+    tenOut = (tenOne.unsqueeze(-1) * patches).sum(dim=1)
+
+    # Permute to (B, K*K, H, W) to match CUDA output layout
+    tenOut = tenOut.permute(0, 3, 1, 2).contiguous()
+
+    return tenOut
+
+
+_costvol_fn = None
+
+def _pytorch_costvol(tenOne, tenTwo, intKernelSize):
+    global _costvol_fn
+    if _costvol_fn is None:
+        try:
+            _costvol_fn = torch.compile(_pytorch_costvol_impl)
+        except Exception:
+            _costvol_fn = _pytorch_costvol_impl
+    try:
+        return _costvol_fn(tenOne, tenTwo, intKernelSize)
+    except Exception:
+        _costvol_fn = _pytorch_costvol_impl
+        return _costvol_fn(tenOne, tenTwo, intKernelSize)
+
+
 ##########################################################
 
 
@@ -253,6 +291,8 @@ class costvol_func(torch.autograd.Function):
     @staticmethod
     @torch.amp.custom_fwd(device_type='cuda', cast_inputs=torch.float32)
     def forward(self, tenOne, tenTwo, intKernelSize):
+        _ensure_cupy()
+        if tenOne.is_cuda and cupy is not None:
             tenOut = tenOne.new_empty([tenOne.shape[0], intKernelSize ** 2, tenOne.shape[2], tenOne.shape[3]])
 
             cuda_launch(cuda_kernel('costvol_out', '''
@@ -302,6 +342,8 @@ class costvol_func(torch.autograd.Function):
                 args=[cuda_int32(tenOut.shape[0] * tenOut.shape[2] * tenOut.shape[3]), tenOne.data_ptr(), tenTwo.data_ptr(), intKernelSize, tenOut.data_ptr()],
                 stream=collections.namedtuple('Stream', 'ptr')(torch.cuda.current_stream().cuda_stream)
             )
+        else:
+            tenOut = _pytorch_costvol(tenOne, tenTwo, intKernelSize)
 
         self.save_for_backward(tenOne, tenTwo)
         self.intKernelSize = intKernelSize

docs/plans/2026-04-11-cupy-fallback.md

@@ -0,0 +1,297 @@
+# Pure-PyTorch Fallbacks for cupy Kernels
+
+> **For Claude:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.
+
+**Goal:** Make BIM-VFI, SGM-VFI, and GIMM-VFI work without cupy by adding pure-PyTorch fallback implementations of softsplat and costvol.
+
+**Architecture:** Each kernel file (`sgm_vfi_arch/softsplat.py`, `gimm_vfi_arch/.../softsplat.py`, `bim_vfi_arch/costvol.py`) gets a `_pytorch_*` fallback function. The `softsplat_func.forward()` and `costvol_func.forward()` methods dispatch to cupy when available, otherwise use the fallback. The `_check_cupy()` gate in `nodes.py` is removed so models can load on any backend.
+
+**Tech Stack:** PyTorch (`scatter_add_`, `F.unfold`, `F.pad`)
+
+---
+
+### Task 1: Add pure-PyTorch softsplat fallback to SGM-VFI
+
+**Files:**
+- Modify: `sgm_vfi_arch/softsplat.py`
+
+**Step 1: Add cupy availability flag and fallback function**
+
+At the top of `sgm_vfi_arch/softsplat.py`, change the hard `import cupy` to a try/except, and add the fallback function after the `cuda_launch` function (before the `softsplat()` function).
+
+Replace:
+```python
+import cupy
+```
+With:
+```python
+try:
+    import cupy
+except ImportError:
+    cupy = None
+```
+
+Add this fallback function (after `cuda_launch`, before `softsplat`):
+```python
+def _pytorch_softsplat(tenIn, tenFlow):
+    B, C, H, W = tenIn.shape
+    tenOut = tenIn.new_zeros(B, C, H, W)
+
+    # Build base grid: (x, y) for each pixel
+    grid_y, grid_x = torch.meshgrid(
+        torch.arange(H, device=tenIn.device, dtype=tenIn.dtype),
+        torch.arange(W, device=tenIn.device, dtype=tenIn.dtype),
+        indexing='ij',
+    )
+
+    # Target positions
+    flt_x = grid_x.unsqueeze(0) + tenFlow[:, 0, :, :]  # (B, H, W)
+    flt_y = grid_y.unsqueeze(0) + tenFlow[:, 1, :, :]
+
+    # Filter non-finite
+    valid = torch.isfinite(flt_x) & torch.isfinite(flt_y)
+    flt_x = torch.where(valid, flt_x, torch.zeros_like(flt_x))
+    flt_y = torch.where(valid, flt_y, torch.zeros_like(flt_y))
+
+    # Four neighbors (NW, NE, SW, SE)
+    nw_x = flt_x.floor().long()
+    nw_y = flt_y.floor().long()
+
+    # Bilinear weights
+    frac_x = flt_x - nw_x.float()
+    frac_y = flt_y - nw_y.float()
+    w_nw = (1.0 - frac_x) * (1.0 - frac_y)
+    w_ne = frac_x * (1.0 - frac_y)
+    w_sw = (1.0 - frac_x) * frac_y
+    w_se = frac_x * frac_y
+
+    # Zero out invalid pixels
+    w_nw = w_nw * valid
+    w_ne = w_ne * valid
+    w_sw = w_sw * valid
+    w_se = w_se * valid
+
+    # For each of the 4 neighbors, scatter into output
+    for dx, dy, w in [(0, 0, w_nw), (1, 0, w_ne), (0, 1, w_sw), (1, 1, w_se)]:
+        tx = nw_x + dx
+        ty = nw_y + dy
+        in_bounds = (tx >= 0) & (tx < W) & (ty >= 0) & (ty < H)
+        w_masked = w * in_bounds
+
+        # Flatten to 1D index for scatter_add
+        idx = (ty.clamp(0, H - 1) * W + tx.clamp(0, W - 1))  # (B, H, W)
+        idx = idx.unsqueeze(1).expand_as(tenIn)  # (B, C, H, W)
+        weighted = tenIn * w_masked.unsqueeze(1)  # (B, C, H, W)
+        tenOut.view(B, C, -1).scatter_add_(2, idx.reshape(B, C, -1), weighted.reshape(B, C, -1))
+
+    return tenOut
+```
+
+**Step 2: Update softsplat_func.forward to use fallback**
+
+In `softsplat_func.forward()`, replace the `elif tenIn.is_cuda != True: assert(False)` block so it dispatches to the fallback when cupy is unavailable or when not on CUDA:
+
+```python
+# Current:
+        if tenIn.is_cuda == True:
+            cuda_launch(cuda_kernel(...))(...) 
+        elif tenIn.is_cuda != True:
+            assert(False)
+
+# New:
+        if tenIn.is_cuda and cupy is not None:
+            cuda_launch(cuda_kernel(...))(...) 
+        else:
+            tenOut = _pytorch_softsplat(tenIn, tenFlow)
+```
+
+Also guard the `@cupy.memoize` decorator on `cuda_launch`:
+```python
+# Current:
+@cupy.memoize(for_each_device=True)
+def cuda_launch(strKey:str):
+
+# New:
+def cuda_launch(strKey:str):
+```
+(The function already has its own dict-based caching via `objCudacache`, and the memoize is redundant anyway. But the real issue is it crashes at import when cupy=None.)
+
+Wait - actually `cuda_launch` uses `cupy.RawKernel` inside, so it's only ever called on the cupy path. The `@cupy.memoize` decorator is the problem: it runs at import time. Replace it:
+
+```python
+# Replace @cupy.memoize(for_each_device=True) with a simple cache dict
+_cuda_launch_cache = {}
+
+def cuda_launch(strKey:str):
+    if strKey not in _cuda_launch_cache:
+        if 'CUDA_HOME' not in os.environ:
+            os.environ['CUDA_HOME'] = cupy.cuda.get_cuda_path()
+        _cuda_launch_cache[strKey] = cupy.RawKernel(
+            objCudacache[strKey]['strKernel'],
+            objCudacache[strKey]['strFunction'],
+            options=tuple(['-I ' + os.environ['CUDA_HOME'],
+                           '-I ' + os.environ['CUDA_HOME'] + '/include'])
+        )
+    return _cuda_launch_cache[strKey]
+```
+
+**Step 3: Commit**
+
+```bash
+git add sgm_vfi_arch/softsplat.py
+git commit -m "feat: add pure-PyTorch softsplat fallback for SGM-VFI"
+```
+
+---
+
+### Task 2: Add pure-PyTorch softsplat fallback to GIMM-VFI
+
+**Files:**
+- Modify: `gimm_vfi_arch/generalizable_INR/modules/softsplat.py`
+
+**Step 1: Add cupy availability flag and fallback function**
+
+Same pattern as Task 1. Replace `import cupy` with try/except. Add the same `_pytorch_softsplat()` function. Replace `@cupy.memoize(for_each_device=True)` on `cuda_launch` with a dict cache.
+
+The GIMM softsplat.py already has `@torch.compiler.disable()` on `cuda_launch` — keep that decorator.
+
+**Step 2: Update softsplat_func.forward dispatch**
+
+Same pattern: if `tenIn.is_cuda and cupy is not None` → cupy path, else → `_pytorch_softsplat`.
+
+**Step 3: Commit**
+
+```bash
+git add gimm_vfi_arch/generalizable_INR/modules/softsplat.py
+git commit -m "feat: add pure-PyTorch softsplat fallback for GIMM-VFI"
+```
+
+---
+
+### Task 3: Add pure-PyTorch costvol fallback to BIM-VFI
+
+**Files:**
+- Modify: `bim_vfi_arch/costvol.py`
+
+**Step 1: Add the fallback function**
+
+After the existing `cuda_launch` function, add:
+
+```python
+def _pytorch_costvol(tenOne, tenTwo, intKernelSize):
+    B, C, H, W = tenOne.shape
+    pad = (intKernelSize - 1) // 2
+
+    # Pad tenTwo with zeros so out-of-bounds accesses yield 0 (matches CUDA kernel)
+    tenTwo_padded = F.pad(tenTwo, [pad, pad, pad, pad])
+
+    # Unfold into (B, C, K*K, H, W) patches
+    patches = tenTwo_padded.unfold(2, intKernelSize, 1).unfold(3, intKernelSize, 1)
+    # patches shape: (B, C, H, W, K, K)
+    patches = patches.contiguous().view(B, C, H, W, intKernelSize * intKernelSize)
+    # -> (B, C, H, W, K^2)
+
+    # Dot product: sum over C
+    # tenOne: (B, C, H, W) -> (B, C, H, W, 1)
+    tenOut = (tenOne.unsqueeze(-1) * patches).sum(dim=1)
+    # tenOut: (B, H, W, K^2)
+
+    # Permute to (B, K^2, H, W) to match CUDA output layout
+    tenOut = tenOut.permute(0, 3, 1, 2).contiguous()
+
+    return tenOut
+```
+
+Add `import torch.nn.functional as F` at the top if not already present.
+
+**Step 2: Update costvol_func.forward dispatch**
+
+The current forward unconditionally calls `cuda_launch(cuda_kernel(...))`. Change to:
+
+```python
+@staticmethod
+@torch.amp.custom_fwd(device_type='cuda', cast_inputs=torch.float32)
+def forward(self, tenOne, tenTwo, intKernelSize):
+    if tenOne.is_cuda and cupy is not None:
+        # existing cupy code (unchanged)
+        tenOut = tenOne.new_empty([tenOne.shape[0], intKernelSize ** 2, tenOne.shape[2], tenOne.shape[3]])
+        cuda_launch(cuda_kernel(...))(...) 
+    else:
+        tenOut = _pytorch_costvol(tenOne, tenTwo, intKernelSize)
+
+    self.save_for_backward(tenOne, tenTwo)
+    self.intKernelSize = intKernelSize
+    return tenOut
+```
+
+**Step 3: Commit**
+
+```bash
+git add bim_vfi_arch/costvol.py
+git commit -m "feat: add pure-PyTorch costvol fallback for BIM-VFI"
+```
+
+---
+
+### Task 4: Remove _check_cupy gate from nodes.py
+
+**Files:**
+- Modify: `nodes.py`
+
+**Step 1: Remove the _check_cupy function and all its call sites**
+
+Delete the `_check_cupy()` function definition (lines 22-41). Remove the three calls:
+- Line 209: `_check_cupy("BIM-VFI")` (in BIM-VFI load)
+- Line 1377: `_check_cupy("SGM-VFI")` (in SGM-VFI load)
+- Line 1804: `_check_cupy("GIMM-VFI")` (in GIMM-VFI load)
+
+**Step 2: Commit**
+
+```bash
+git add nodes.py
+git commit -m "feat: remove cupy requirement gate, models now fallback to pure PyTorch"
+```
+
+---
+
+### Task 5: Make install.py not force cupy installation
+
+**Files:**
+- Modify: `install.py`
+
+**Step 1: Change cupy from required to optional**
+
+Make cupy a soft dependency — try to install it but don't fail if it can't be installed (ROCm users, no CUDA toolkit, etc.). Change `install()`:
+
+```python
+def install():
+    # Install core requirements first
+    requirements_path = os.path.join(os.path.dirname(__file__), "requirements.txt")
+    subprocess.check_call([
+        sys.executable, "-m", "pip", "install", "-r", requirements_path
+    ])
+
+    # Try to install cupy for NVIDIA users (optional, improves performance)
+    cupy_pkg = get_cupy_package()
+    if cupy_pkg:
+        try:
+            subprocess.check_call([
+                sys.executable, "-m", "pip", "install", cupy_pkg
+            ])
+            print(f"[Tween] cupy installed successfully ({cupy_pkg})")
+        except subprocess.CalledProcessError:
+            print(f"[Tween] WARNING: Could not install {cupy_pkg}. "
+                  f"BIM-VFI, SGM-VFI, and GIMM-VFI will use slower PyTorch fallback.")
+    else:
+        print("[Tween] cupy not available (no NVIDIA CUDA). "
+              "BIM-VFI, SGM-VFI, and GIMM-VFI will use PyTorch fallback.")
+```
+
+Also stop writing cupy into `requirements.txt` — remove the `update_requirements` call and function.
+
+**Step 2: Commit**
+
+```bash
+git add install.py
+git commit -m "feat: make cupy optional in install.py"
+```

gimm_vfi_arch/generalizable_INR/modules/softsplat.py

@@ -9,7 +9,13 @@
 # --------------------------------------------------------
 
 import collections
+try:
     import cupy
+except Exception:
+    # Broad catch: an installed-but-broken cupy (e.g. incompatible NumPy)
+    # raises non-ImportError exceptions at import time. Treat any failure as
+    # "cupy unavailable" and fall back to the pure-PyTorch implementation.
+    cupy = None
 import os
 import re
 import torch
@@ -260,31 +266,94 @@ def cuda_kernel(strFunction: str, strKernel: str, objVariables: typing.Dict):
 # end
 
 
-@cupy.memoize(for_each_device=True)
+_cuda_launch_cache = {}
+
 @torch.compiler.disable()
 def cuda_launch(strKey: str):
-    try:
-        os.environ.setdefault("CUDA_HOME", cupy.cuda.get_cuda_path())
-    except Exception:
+    if strKey not in _cuda_launch_cache:
         if "CUDA_HOME" not in os.environ:
-            raise RuntimeError("'CUDA_HOME' not set, unable to find cuda-toolkit installation.")
-    
+            try:
+                cuda_path = cupy.cuda.get_cuda_path()
+            except Exception:
+                cuda_path = None
+            if cuda_path is None:
+                cuda_path = "/usr/local/cuda"
+            os.environ["CUDA_HOME"] = cuda_path
         strKernel = objCudacache[strKey]["strKernel"]
         strFunction = objCudacache[strKey]["strFunction"]
-    
-    return cupy.RawModule(
+        _cuda_launch_cache[strKey] = cupy.RawModule(
             code=strKernel,
             options=(
                 "-I " + os.environ["CUDA_HOME"],
                 "-I " + os.environ["CUDA_HOME"] + "/include",
             ),
         ).get_function(strFunction)
+    return _cuda_launch_cache[strKey]
 
 
 
 ##########################################################
 
 
+def _pytorch_softsplat_impl(tenIn, tenFlow):
+    """Pure-PyTorch forward warp via bilinear splatting (scatter_add)."""
+    B, C, H, W = tenIn.shape
+    tenOut = tenIn.new_zeros(B, C, H, W)
+
+    grid_y, grid_x = torch.meshgrid(
+        torch.arange(H, device=tenIn.device, dtype=tenIn.dtype),
+        torch.arange(W, device=tenIn.device, dtype=tenIn.dtype),
+        indexing='ij',
+    )
+
+    flt_x = grid_x.unsqueeze(0) + tenFlow[:, 0, :, :]
+    flt_y = grid_y.unsqueeze(0) + tenFlow[:, 1, :, :]
+
+    valid = torch.isfinite(flt_x) & torch.isfinite(flt_y)
+    flt_x = torch.where(valid, flt_x, torch.zeros_like(flt_x))
+    flt_y = torch.where(valid, flt_y, torch.zeros_like(flt_y))
+
+    nw_x = flt_x.floor().long()
+    nw_y = flt_y.floor().long()
+    frac_x = flt_x - nw_x.to(flt_x.dtype)
+    frac_y = flt_y - nw_y.to(flt_y.dtype)
+
+    w_nw = (1.0 - frac_x) * (1.0 - frac_y) * valid
+    w_ne = frac_x * (1.0 - frac_y) * valid
+    w_sw = (1.0 - frac_x) * frac_y * valid
+    w_se = frac_x * frac_y * valid
+
+    out_flat = tenOut.view(B, C, -1)
+
+    for dx, dy, w in [(0, 0, w_nw), (1, 0, w_ne), (0, 1, w_sw), (1, 1, w_se)]:
+        tx = nw_x + dx
+        ty = nw_y + dy
+        in_bounds = (tx >= 0) & (tx < W) & (ty >= 0) & (ty < H)
+        w_masked = w * in_bounds
+        idx = (ty.clamp(0, H - 1) * W + tx.clamp(0, W - 1))
+        idx = idx.unsqueeze(1).expand_as(tenIn)
+        weighted = tenIn * w_masked.unsqueeze(1)
+        out_flat.scatter_add_(2, idx.reshape(B, C, -1), weighted.reshape(B, C, -1))
+
+    return tenOut
+
+
+_softsplat_fn = None
+
+def _pytorch_softsplat(tenIn, tenFlow):
+    global _softsplat_fn
+    if _softsplat_fn is None:
+        try:
+            _softsplat_fn = torch.compile(_pytorch_softsplat_impl)
+        except Exception:
+            _softsplat_fn = _pytorch_softsplat_impl
+    try:
+        return _softsplat_fn(tenIn, tenFlow)
+    except Exception:
+        _softsplat_fn = _pytorch_softsplat_impl
+        return _softsplat_fn(tenIn, tenFlow)
+
+
 @torch.compiler.disable()
 def softsplat(tenIn, tenFlow, tenMetric, strMode, return_norm=False):
     assert strMode.split("-")[0] in ["sum", "avg", "linear", "softmax"]
@@ -366,7 +435,7 @@ class softsplat_func(torch.autograd.Function):
             [tenIn.shape[0], tenIn.shape[1], tenIn.shape[2], tenIn.shape[3]]
         )
 
-        if tenIn.is_cuda == True:
+        if tenIn.is_cuda and cupy is not None:
             cuda_launch(
                 cuda_kernel(
                     "softsplat_out",
@@ -439,8 +508,8 @@ class softsplat_func(torch.autograd.Function):
                 ),
             )
 
-        elif tenIn.is_cuda != True:
-            assert False
+        else:
+            tenOut = _pytorch_softsplat(tenIn, tenFlow)
 
         # end
 

install.py

@@ -8,44 +8,39 @@ def get_cupy_package():
     try:
         import torch
         if not torch.cuda.is_available():
-            print("[Tween] WARNING: CUDA not available. cupy requires CUDA.")
             return None
         cuda_version = torch.version.cuda
         if cuda_version is None:
-            print("[Tween] WARNING: PyTorch has no CUDA version info.")
             return None
         major = int(cuda_version.split(".")[0])
         cupy_pkg = f"cupy-cuda{major}x"
-        print(f"[Tween] Detected CUDA {cuda_version}, will use {cupy_pkg}")
         return cupy_pkg
-    except Exception as e:
-        print(f"[Tween] WARNING: Could not detect CUDA version: {e}")
+    except Exception:
         return None
 
 
-def update_requirements(cupy_pkg):
-    """Write the correct cupy package into requirements.txt."""
-    requirements_path = os.path.join(os.path.dirname(__file__), "requirements.txt")
-    lines = []
-    if os.path.exists(requirements_path):
-        with open(requirements_path, "r") as f:
-            lines = [l.rstrip() for l in f if not l.strip().startswith("cupy")]
-    if cupy_pkg and cupy_pkg not in lines:
-        lines.append(cupy_pkg)
-    with open(requirements_path, "w") as f:
-        f.write("\n".join(lines) + "\n")
-
-
 def install():
-    cupy_pkg = get_cupy_package()
-    if cupy_pkg:
-        update_requirements(cupy_pkg)
-
+    # Install core requirements first
     requirements_path = os.path.join(os.path.dirname(__file__), "requirements.txt")
     subprocess.check_call([
         sys.executable, "-m", "pip", "install", "-r", requirements_path
     ])
 
+    # Try to install cupy for NVIDIA users (optional, improves performance)
+    cupy_pkg = get_cupy_package()
+    if cupy_pkg:
+        try:
+            subprocess.check_call([
+                sys.executable, "-m", "pip", "install", cupy_pkg
+            ])
+            print(f"[Tween] cupy installed ({cupy_pkg}) — fast CUDA kernels enabled")
+        except subprocess.CalledProcessError:
+            print(f"[Tween] WARNING: Could not install {cupy_pkg}. "
+                  f"BIM-VFI, SGM-VFI, and GIMM-VFI will use slower PyTorch fallback.")
+    else:
+        print("[Tween] cupy skipped (no NVIDIA CUDA). "
+              "BIM-VFI, SGM-VFI, and GIMM-VFI will use PyTorch fallback.")
+
 
 if __name__ == "__main__":
     install()

nodes.py

@@ -19,26 +19,6 @@ from .gimm_vfi_arch import clear_gimm_caches
 logger = logging.getLogger("Tween")
 
 
-def _check_cupy(model_name):
-    """Raise a clear error if cupy is not installed."""
-    try:
-        import cupy  # noqa: F401
-    except ImportError:
-        try:
-            cuda_ver = torch.version.cuda or "unknown"
-            major = int(cuda_ver.split(".")[0])
-            cupy_pkg = f"cupy-cuda{major}x"
-        except Exception:
-            cuda_ver = "unknown"
-            cupy_pkg = "cupy-cuda12x  # adjust to your CUDA version"
-        raise RuntimeError(
-            f"{model_name} requires cupy but it is not installed.\n\n"
-            f"Your PyTorch CUDA version: {cuda_ver}\n\n"
-            f"Install it with:\n"
-            f"  pip install {cupy_pkg}\n\n"
-            f"If you are unsure of your CUDA version, run:\n"
-            f"  python -c \"import torch; print(torch.version.cuda)\""
-        )
 
 
 def _get_system_ram_gb():
@@ -206,7 +186,6 @@ class LoadBIMVFIModel:
     CATEGORY = "video/BIM-VFI"
 
     def load_model(self, model_path, auto_pyr_level, pyr_level):
-        _check_cupy("BIM-VFI")
         full_path = os.path.join(MODEL_DIR, model_path)
 
         if not os.path.exists(full_path):
@@ -1374,7 +1353,6 @@ class LoadSGMVFIModel:
     CATEGORY = "video/SGM-VFI"
 
     def load_model(self, model_path, tta, num_key_points):
-        _check_cupy("SGM-VFI")
         full_path = os.path.join(SGM_MODEL_DIR, model_path)
 
         if not os.path.exists(full_path):
@@ -1801,7 +1779,6 @@ class LoadGIMMVFIModel:
     CATEGORY = "video/GIMM-VFI"
 
     def load_model(self, model_path, ds_factor):
-        _check_cupy("GIMM-VFI")
         full_path = os.path.join(GIMM_MODEL_DIR, model_path)
 
         # Auto-download main model if missing

sgm_vfi_arch/softsplat.py

@@ -1,7 +1,13 @@
 #!/usr/bin/env python
 
 import collections
+try:
     import cupy
+except Exception:
+    # Broad catch: an installed-but-broken cupy (e.g. incompatible NumPy)
+    # raises non-ImportError exceptions at import time. Treat any failure as
+    # "cupy unavailable" and fall back to the pure-PyTorch implementation.
+    cupy = None
 import os
 import re
 import torch
@@ -216,20 +222,91 @@ def cuda_kernel(strFunction:str, strKernel:str, objVariables:typing.Dict):
 # end
 
 
-@cupy.memoize(for_each_device=True)
-def cuda_launch(strKey:str):
-    if 'CUDA_HOME' not in os.environ:
-        os.environ['CUDA_HOME'] = cupy.cuda.get_cuda_path()
-    # end
+_cuda_launch_cache = {}
 
-    return cupy.RawKernel(objCudacache[strKey]['strKernel'], objCudacache[strKey]['strFunction'],
-                          options=tuple(['-I ' + os.environ['CUDA_HOME'], '-I ' + os.environ['CUDA_HOME'] + '/include']))
+def cuda_launch(strKey:str):
+    if strKey not in _cuda_launch_cache:
+        if 'CUDA_HOME' not in os.environ:
+            try:
+                cuda_path = cupy.cuda.get_cuda_path()
+            except Exception:
+                cuda_path = None
+            if cuda_path is None:
+                cuda_path = '/usr/local/cuda'
+            os.environ['CUDA_HOME'] = cuda_path
+        _cuda_launch_cache[strKey] = cupy.RawKernel(
+            objCudacache[strKey]['strKernel'],
+            objCudacache[strKey]['strFunction'],
+            options=tuple(['-I ' + os.environ['CUDA_HOME'],
+                           '-I ' + os.environ['CUDA_HOME'] + '/include'])
+        )
+    return _cuda_launch_cache[strKey]
 # end
 
 
 ##########################################################
 
 
+def _pytorch_softsplat_impl(tenIn, tenFlow):
+    """Pure-PyTorch forward warp via bilinear splatting (scatter_add)."""
+    B, C, H, W = tenIn.shape
+    tenOut = tenIn.new_zeros(B, C, H, W)
+
+    grid_y, grid_x = torch.meshgrid(
+        torch.arange(H, device=tenIn.device, dtype=tenIn.dtype),
+        torch.arange(W, device=tenIn.device, dtype=tenIn.dtype),
+        indexing='ij',
+    )
+
+    flt_x = grid_x.unsqueeze(0) + tenFlow[:, 0, :, :]
+    flt_y = grid_y.unsqueeze(0) + tenFlow[:, 1, :, :]
+
+    valid = torch.isfinite(flt_x) & torch.isfinite(flt_y)
+    flt_x = torch.where(valid, flt_x, torch.zeros_like(flt_x))
+    flt_y = torch.where(valid, flt_y, torch.zeros_like(flt_y))
+
+    nw_x = flt_x.floor().long()
+    nw_y = flt_y.floor().long()
+    frac_x = flt_x - nw_x.to(flt_x.dtype)
+    frac_y = flt_y - nw_y.to(flt_y.dtype)
+
+    w_nw = (1.0 - frac_x) * (1.0 - frac_y) * valid
+    w_ne = frac_x * (1.0 - frac_y) * valid
+    w_sw = (1.0 - frac_x) * frac_y * valid
+    w_se = frac_x * frac_y * valid
+
+    out_flat = tenOut.view(B, C, -1)
+
+    for dx, dy, w in [(0, 0, w_nw), (1, 0, w_ne), (0, 1, w_sw), (1, 1, w_se)]:
+        tx = nw_x + dx
+        ty = nw_y + dy
+        in_bounds = (tx >= 0) & (tx < W) & (ty >= 0) & (ty < H)
+        w_masked = w * in_bounds
+        idx = (ty.clamp(0, H - 1) * W + tx.clamp(0, W - 1))
+        idx = idx.unsqueeze(1).expand_as(tenIn)
+        weighted = tenIn * w_masked.unsqueeze(1)
+        out_flat.scatter_add_(2, idx.reshape(B, C, -1), weighted.reshape(B, C, -1))
+
+    return tenOut
+
+
+_softsplat_fn = None
+
+def _pytorch_softsplat(tenIn, tenFlow):
+    global _softsplat_fn
+    if _softsplat_fn is None:
+        try:
+            _softsplat_fn = torch.compile(_pytorch_softsplat_impl)
+        except Exception:
+            _softsplat_fn = _pytorch_softsplat_impl
+    try:
+        return _softsplat_fn(tenIn, tenFlow)
+    except Exception:
+        _softsplat_fn = _pytorch_softsplat_impl
+        return _softsplat_fn(tenIn, tenFlow)
+# end
+
+
 def softsplat(tenIn:torch.Tensor, tenFlow:torch.Tensor, tenMetric:torch.Tensor, strMode:str):
     assert(strMode.split('-')[0] in ['sum', 'avg', 'linear', 'soft'])
 
@@ -281,7 +358,7 @@ class softsplat_func(torch.autograd.Function):
     def forward(self, tenIn, tenFlow):
         tenOut = tenIn.new_zeros([tenIn.shape[0], tenIn.shape[1], tenIn.shape[2], tenIn.shape[3]])
 
-        if tenIn.is_cuda == True:
+        if tenIn.is_cuda and cupy is not None:
             cuda_launch(cuda_kernel('softsplat_out', '''
                 extern "C" __global__ void __launch_bounds__(512) softsplat_out(
                     const int n,
@@ -345,8 +422,8 @@ class softsplat_func(torch.autograd.Function):
                 stream=collections.namedtuple('Stream', 'ptr')(torch.cuda.current_stream().cuda_stream)
             )
 
-        elif tenIn.is_cuda != True:
-            assert(False)
+        else:
+            tenOut = _pytorch_softsplat(tenIn, tenFlow)
 
         # end