diff --git a/comfy/model_base.py b/comfy/model_base.py
index 5c2668ba9..1c7695761 100644
--- a/comfy/model_base.py
+++ b/comfy/model_base.py
@@ -578,8 +578,8 @@ class Stable_Zero123(BaseModel):
     def __init__(self, model_config, model_type=ModelType.EPS, device=None, cc_projection_weight=None, cc_projection_bias=None):
         super().__init__(model_config, model_type, device=device)
         self.cc_projection = comfy.ops.manual_cast.Linear(cc_projection_weight.shape[1], cc_projection_weight.shape[0], dtype=self.get_dtype(), device=device)
-        self.cc_projection.weight.copy_(cc_projection_weight)
-        self.cc_projection.bias.copy_(cc_projection_bias)
+        self.cc_projection.weight = torch.nn.Parameter(cc_projection_weight.clone())
+        self.cc_projection.bias = torch.nn.Parameter(cc_projection_bias.clone())
 
     def extra_conds(self, **kwargs):
         out = {}
diff --git a/comfy/model_management.py b/comfy/model_management.py
index bcf1399c4..3b39d6080 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -1801,7 +1801,7 @@ def debug_memory_summary():
         return torch.cuda.memory.memory_summary()
     return ""
 
-class InterruptProcessingException(Exception):
+class InterruptProcessingException(BaseException):
     pass
 
 interrupt_processing_mutex = threading.RLock()
diff --git a/comfy/model_patcher.py b/comfy/model_patcher.py
index 93d19d6fe..ee56f8523 100644
--- a/comfy/model_patcher.py
+++ b/comfy/model_patcher.py
@@ -685,9 +685,9 @@ class ModelPatcher:
                         sd.pop(k)
             return sd
 
-    def patch_weight_to_device(self, key, device_to=None, inplace_update=False, return_weight=False):
+    def patch_weight_to_device(self, key, device_to=None, inplace_update=False, return_weight=False, force_cast=False):
         weight, set_func, convert_func = get_key_weight(self.model, key)
-        if key not in self.patches:
+        if key not in self.patches and not force_cast:
             return weight
 
         inplace_update = self.weight_inplace_update or inplace_update
@@ -695,7 +695,7 @@ class ModelPatcher:
         if key not in self.backup and not return_weight:
             self.backup[key] = collections.namedtuple('Dimension', ['weight', 'inplace_update'])(weight.to(device=self.offload_device, copy=inplace_update), inplace_update)
 
-        temp_dtype = comfy.model_management.lora_compute_dtype(device_to)
+        temp_dtype = comfy.model_management.lora_compute_dtype(device_to) if key in self.patches else None
         if device_to is not None:
             temp_weight = comfy.model_management.cast_to_device(weight, device_to, temp_dtype, copy=True)
         else:
@@ -703,9 +703,10 @@ class ModelPatcher:
         if convert_func is not None:
             temp_weight = convert_func(temp_weight, inplace=True)
 
-        out_weight = comfy.lora.calculate_weight(self.patches[key], temp_weight, key)
+        out_weight = comfy.lora.calculate_weight(self.patches[key], temp_weight, key) if key in self.patches else temp_weight
         if set_func is None:
-            out_weight = comfy.float.stochastic_rounding(out_weight, weight.dtype, seed=comfy.utils.string_to_seed(key))
+            if key in self.patches:
+                out_weight = comfy.float.stochastic_rounding(out_weight, weight.dtype, seed=comfy.utils.string_to_seed(key))
             if return_weight:
                 return out_weight
             elif inplace_update:
@@ -1584,7 +1585,7 @@ class ModelPatcherDynamic(ModelPatcher):
                     key = key_param_name_to_key(n, param_key)
                     if key in self.backup:
                         comfy.utils.set_attr_param(self.model, key, self.backup[key].weight)
-                    self.patch_weight_to_device(key, device_to=device_to)
+                    self.patch_weight_to_device(key, device_to=device_to, force_cast=True)
                     weight, _, _ = get_key_weight(self.model, key)
                     if weight is not None:
                         self.model.model_loaded_weight_memory += weight.numel() * weight.element_size()
@@ -1609,6 +1610,10 @@ class ModelPatcherDynamic(ModelPatcher):
                             m._v = vbar.alloc(v_weight_size)
                         allocated_size += v_weight_size
 
+                    for param in params:
+                        if param not in ("weight", "bias"):
+                            force_load_param(self, param, device_to)
+
                 else:
                     for param in params:
                         key = key_param_name_to_key(n, param)
diff --git a/comfy_api_nodes/nodes_veo2.py b/comfy_api_nodes/nodes_veo2.py
index 084b086a8..2ff75d9b2 100644
--- a/comfy_api_nodes/nodes_veo2.py
+++ b/comfy_api_nodes/nodes_veo2.py
@@ -393,8 +393,8 @@ class Veo3VideoGenerationNode(IO.ComfyNode):
         model="veo-3.0-generate-001",
         generate_audio=False,
     ):
-        if "lite" in model and resolution == "4k":
-            raise Exception("4K resolution is not supported by the veo-3.1-lite model.")
+        if resolution == "4k" and ("lite" in model or "3.0" in model):
+            raise Exception("4K resolution is not supported by the veo-3.1-lite or veo-3.0 models.")
 
         model = MODELS_MAP[model]
 
diff --git a/comfy_extras/frame_interpolation_models/film_net.py b/comfy_extras/frame_interpolation_models/film_net.py
new file mode 100644
index 000000000..cf4f6e1e1
--- /dev/null
+++ b/comfy_extras/frame_interpolation_models/film_net.py
@@ -0,0 +1,258 @@
+"""FILM: Frame Interpolation for Large Motion (ECCV 2022)."""
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+import comfy.ops
+
+ops = comfy.ops.disable_weight_init
+
+
+class FilmConv2d(nn.Module):
+    """Conv2d with optional LeakyReLU and FILM-style padding."""
+
+    def __init__(self, in_channels, out_channels, size, activation=True, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.even_pad = not size % 2
+        self.conv = operations.Conv2d(in_channels, out_channels, kernel_size=size, padding=size // 2 if size % 2 else 0, device=device, dtype=dtype)
+        self.activation = nn.LeakyReLU(0.2) if activation else None
+
+    def forward(self, x):
+        if self.even_pad:
+            x = F.pad(x, (0, 1, 0, 1))
+        x = self.conv(x)
+        if self.activation is not None:
+            x = self.activation(x)
+        return x
+
+
+def _warp_core(image, flow, grid_x, grid_y):
+    dtype = image.dtype
+    H, W = flow.shape[2], flow.shape[3]
+    dx = flow[:, 0].float() / (W * 0.5)
+    dy = flow[:, 1].float() / (H * 0.5)
+    grid = torch.stack([grid_x[None, None, :] + dx, grid_y[None, :, None] + dy], dim=3)
+    return F.grid_sample(image.float(), grid, mode="bilinear", padding_mode="border", align_corners=False).to(dtype)
+
+
+def build_image_pyramid(image, pyramid_levels):
+    pyramid = [image]
+    for _ in range(1, pyramid_levels):
+        image = F.avg_pool2d(image, 2, 2)
+        pyramid.append(image)
+    return pyramid
+
+
+def flow_pyramid_synthesis(residual_pyramid):
+    flow = residual_pyramid[-1]
+    flow_pyramid = [flow]
+    for residual_flow in residual_pyramid[:-1][::-1]:
+        flow = F.interpolate(flow, size=residual_flow.shape[2:4], mode="bilinear", scale_factor=None).mul_(2).add_(residual_flow)
+        flow_pyramid.append(flow)
+    flow_pyramid.reverse()
+    return flow_pyramid
+
+
+def multiply_pyramid(pyramid, scalar):
+    return [image * scalar[:, None, None, None] for image in pyramid]
+
+
+def pyramid_warp(feature_pyramid, flow_pyramid, warp_fn):
+    return [warp_fn(features, flow) for features, flow in zip(feature_pyramid, flow_pyramid)]
+
+
+def concatenate_pyramids(pyramid1, pyramid2):
+    return [torch.cat([f1, f2], dim=1) for f1, f2 in zip(pyramid1, pyramid2)]
+
+
+class SubTreeExtractor(nn.Module):
+    def __init__(self, in_channels=3, channels=64, n_layers=4, device=None, dtype=None, operations=ops):
+        super().__init__()
+        convs = []
+        for i in range(n_layers):
+            out_ch = channels << i
+            convs.append(nn.Sequential(
+                FilmConv2d(in_channels, out_ch, 3, device=device, dtype=dtype, operations=operations),
+                FilmConv2d(out_ch, out_ch, 3, device=device, dtype=dtype, operations=operations)))
+            in_channels = out_ch
+        self.convs = nn.ModuleList(convs)
+
+    def forward(self, image, n):
+        head = image
+        pyramid = []
+        for i, layer in enumerate(self.convs):
+            head = layer(head)
+            pyramid.append(head)
+            if i < n - 1:
+                head = F.avg_pool2d(head, 2, 2)
+        return pyramid
+
+
+class FeatureExtractor(nn.Module):
+    def __init__(self, in_channels=3, channels=64, sub_levels=4, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.extract_sublevels = SubTreeExtractor(in_channels, channels, sub_levels, device=device, dtype=dtype, operations=operations)
+        self.sub_levels = sub_levels
+
+    def forward(self, image_pyramid):
+        sub_pyramids = [self.extract_sublevels(image_pyramid[i], min(len(image_pyramid) - i, self.sub_levels))
+                        for i in range(len(image_pyramid))]
+        feature_pyramid = []
+        for i in range(len(image_pyramid)):
+            features = sub_pyramids[i][0]
+            for j in range(1, self.sub_levels):
+                if j <= i:
+                    features = torch.cat([features, sub_pyramids[i - j][j]], dim=1)
+            feature_pyramid.append(features)
+            # Free sub-pyramids no longer needed by future levels
+            if i >= self.sub_levels - 1:
+                sub_pyramids[i - self.sub_levels + 1] = None
+        return feature_pyramid
+
+
+class FlowEstimator(nn.Module):
+    def __init__(self, in_channels, num_convs, num_filters, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self._convs = nn.ModuleList()
+        for _ in range(num_convs):
+            self._convs.append(FilmConv2d(in_channels, num_filters, 3, device=device, dtype=dtype, operations=operations))
+            in_channels = num_filters
+        self._convs.append(FilmConv2d(in_channels, num_filters // 2, 1, device=device, dtype=dtype, operations=operations))
+        self._convs.append(FilmConv2d(num_filters // 2, 2, 1, activation=False, device=device, dtype=dtype, operations=operations))
+
+    def forward(self, features_a, features_b):
+        net = torch.cat([features_a, features_b], dim=1)
+        for conv in self._convs:
+            net = conv(net)
+        return net
+
+
+class PyramidFlowEstimator(nn.Module):
+    def __init__(self, filters=64, flow_convs=(3, 3, 3, 3), flow_filters=(32, 64, 128, 256), device=None, dtype=None, operations=ops):
+        super().__init__()
+        in_channels = filters << 1
+        predictors = []
+        for i in range(len(flow_convs)):
+            predictors.append(FlowEstimator(in_channels, flow_convs[i], flow_filters[i], device=device, dtype=dtype, operations=operations))
+            in_channels += filters << (i + 2)
+        self._predictor = predictors[-1]
+        self._predictors = nn.ModuleList(predictors[:-1][::-1])
+
+    def forward(self, feature_pyramid_a, feature_pyramid_b, warp_fn):
+        levels = len(feature_pyramid_a)
+        v = self._predictor(feature_pyramid_a[-1], feature_pyramid_b[-1])
+        residuals = [v]
+        # Coarse-to-fine: shared predictor for deep levels, then specialized predictors for fine levels
+        steps = [(i, self._predictor) for i in range(levels - 2, len(self._predictors) - 1, -1)]
+        steps += [(len(self._predictors) - 1 - k, p) for k, p in enumerate(self._predictors)]
+        for i, predictor in steps:
+            v = F.interpolate(v, size=feature_pyramid_a[i].shape[2:4], mode="bilinear").mul_(2)
+            v_residual = predictor(feature_pyramid_a[i], warp_fn(feature_pyramid_b[i], v))
+            residuals.append(v_residual)
+            v = v.add_(v_residual)
+        residuals.reverse()
+        return residuals
+
+
+def _get_fusion_channels(level, filters):
+    # Per direction: multi-scale features + RGB image (3ch) + flow (2ch), doubled for both directions
+    return (sum(filters << i for i in range(level)) + 3 + 2) * 2
+
+
+class Fusion(nn.Module):
+    def __init__(self, n_layers=4, specialized_layers=3, filters=64, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.output_conv = operations.Conv2d(filters, 3, kernel_size=1, device=device, dtype=dtype)
+        self.convs = nn.ModuleList()
+        in_channels = _get_fusion_channels(n_layers, filters)
+        increase = 0
+        for i in range(n_layers)[::-1]:
+            num_filters = (filters << i) if i < specialized_layers else (filters << specialized_layers)
+            self.convs.append(nn.ModuleList([
+                FilmConv2d(in_channels, num_filters, 2, activation=False, device=device, dtype=dtype, operations=operations),
+                FilmConv2d(in_channels + (increase or num_filters), num_filters, 3, device=device, dtype=dtype, operations=operations),
+                FilmConv2d(num_filters, num_filters, 3, device=device, dtype=dtype, operations=operations)]))
+            in_channels = num_filters
+            increase = _get_fusion_channels(i, filters) - num_filters // 2
+
+    def forward(self, pyramid):
+        net = pyramid[-1]
+        for k, layers in enumerate(self.convs):
+            i = len(self.convs) - 1 - k
+            net = layers[0](F.interpolate(net, size=pyramid[i].shape[2:4], mode="nearest"))
+            net = layers[2](layers[1](torch.cat([pyramid[i], net], dim=1)))
+        return self.output_conv(net)
+
+
+class FILMNet(nn.Module):
+    def __init__(self, pyramid_levels=7, fusion_pyramid_levels=5, specialized_levels=3, sub_levels=4,
+                 filters=64, flow_convs=(3, 3, 3, 3), flow_filters=(32, 64, 128, 256), device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.pyramid_levels = pyramid_levels
+        self.fusion_pyramid_levels = fusion_pyramid_levels
+        self.extract = FeatureExtractor(3, filters, sub_levels, device=device, dtype=dtype, operations=operations)
+        self.predict_flow = PyramidFlowEstimator(filters, flow_convs, flow_filters, device=device, dtype=dtype, operations=operations)
+        self.fuse = Fusion(sub_levels, specialized_levels, filters, device=device, dtype=dtype, operations=operations)
+        self._warp_grids = {}
+
+    def get_dtype(self):
+        return self.extract.extract_sublevels.convs[0][0].conv.weight.dtype
+
+    def _build_warp_grids(self, H, W, device):
+        """Pre-compute warp grids for all pyramid levels."""
+        if (H, W) in self._warp_grids:
+            return
+        self._warp_grids = {}  # clear old resolution grids to prevent memory leaks
+        for _ in range(self.pyramid_levels):
+            self._warp_grids[(H, W)] = (
+                torch.linspace(-(1 - 1 / W), 1 - 1 / W, W, dtype=torch.float32, device=device),
+                torch.linspace(-(1 - 1 / H), 1 - 1 / H, H, dtype=torch.float32, device=device),
+            )
+            H, W = H // 2, W // 2
+
+    def warp(self, image, flow):
+        grid_x, grid_y = self._warp_grids[(flow.shape[2], flow.shape[3])]
+        return _warp_core(image, flow, grid_x, grid_y)
+
+    def extract_features(self, img):
+        """Extract image and feature pyramids for a single frame. Can be cached across pairs."""
+        image_pyramid = build_image_pyramid(img, self.pyramid_levels)
+        feature_pyramid = self.extract(image_pyramid)
+        return image_pyramid, feature_pyramid
+
+    def forward(self, img0, img1, timestep=0.5, cache=None):
+        # FILM uses a scalar timestep per batch element (spatially-varying timesteps not supported)
+        t = timestep.mean(dim=(1, 2, 3)).item() if isinstance(timestep, torch.Tensor) else timestep
+        return self.forward_multi_timestep(img0, img1, [t], cache=cache)
+
+    def forward_multi_timestep(self, img0, img1, timesteps, cache=None):
+        """Compute flow once, synthesize at multiple timesteps. Expects batch=1 inputs."""
+        self._build_warp_grids(img0.shape[2], img0.shape[3], img0.device)
+
+        image_pyr0, feat_pyr0 = cache["img0"] if cache and "img0" in cache else self.extract_features(img0)
+        image_pyr1, feat_pyr1 = cache["img1"] if cache and "img1" in cache else self.extract_features(img1)
+
+        fwd_flow = flow_pyramid_synthesis(self.predict_flow(feat_pyr0, feat_pyr1, self.warp))[:self.fusion_pyramid_levels]
+        bwd_flow = flow_pyramid_synthesis(self.predict_flow(feat_pyr1, feat_pyr0, self.warp))[:self.fusion_pyramid_levels]
+
+        # Build warp targets and free full pyramids (only first fpl levels needed from here)
+        fpl = self.fusion_pyramid_levels
+        p2w = [concatenate_pyramids(image_pyr0[:fpl], feat_pyr0[:fpl]),
+               concatenate_pyramids(image_pyr1[:fpl], feat_pyr1[:fpl])]
+        del image_pyr0, image_pyr1, feat_pyr0, feat_pyr1
+
+        results = []
+        dt_tensors = torch.tensor(timesteps, device=img0.device, dtype=img0.dtype)
+        for idx in range(len(timesteps)):
+            batch_dt = dt_tensors[idx:idx + 1]
+            bwd_scaled = multiply_pyramid(bwd_flow, batch_dt)
+            fwd_scaled = multiply_pyramid(fwd_flow, 1 - batch_dt)
+            fwd_warped = pyramid_warp(p2w[0], bwd_scaled, self.warp)
+            bwd_warped = pyramid_warp(p2w[1], fwd_scaled, self.warp)
+            aligned = [torch.cat([fw, bw, bf, ff], dim=1)
+                       for fw, bw, bf, ff in zip(fwd_warped, bwd_warped, bwd_scaled, fwd_scaled)]
+            del fwd_warped, bwd_warped, bwd_scaled, fwd_scaled
+            results.append(self.fuse(aligned))
+            del aligned
+        return torch.cat(results, dim=0)
diff --git a/comfy_extras/frame_interpolation_models/ifnet.py b/comfy_extras/frame_interpolation_models/ifnet.py
new file mode 100644
index 000000000..03cb34c50
--- /dev/null
+++ b/comfy_extras/frame_interpolation_models/ifnet.py
@@ -0,0 +1,128 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+import comfy.ops
+
+ops = comfy.ops.disable_weight_init
+
+
+def _warp(img, flow, warp_grids):
+    B, _, H, W = img.shape
+    base_grid, flow_div = warp_grids[(H, W)]
+    flow_norm = torch.cat([flow[:, 0:1] / flow_div[0], flow[:, 1:2] / flow_div[1]], 1).float()
+    grid = (base_grid.expand(B, -1, -1, -1) + flow_norm).permute(0, 2, 3, 1)
+    return F.grid_sample(img.float(), grid, mode="bilinear", padding_mode="border", align_corners=True).to(img.dtype)
+
+
+class Head(nn.Module):
+    def __init__(self, out_ch=4, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.cnn0 = operations.Conv2d(3, 16, 3, 2, 1, device=device, dtype=dtype)
+        self.cnn1 = operations.Conv2d(16, 16, 3, 1, 1, device=device, dtype=dtype)
+        self.cnn2 = operations.Conv2d(16, 16, 3, 1, 1, device=device, dtype=dtype)
+        self.cnn3 = operations.ConvTranspose2d(16, out_ch, 4, 2, 1, device=device, dtype=dtype)
+        self.relu = nn.LeakyReLU(0.2, True)
+
+    def forward(self, x):
+        x = self.relu(self.cnn0(x))
+        x = self.relu(self.cnn1(x))
+        x = self.relu(self.cnn2(x))
+        return self.cnn3(x)
+
+
+class ResConv(nn.Module):
+    def __init__(self, c, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.conv = operations.Conv2d(c, c, 3, 1, 1, device=device, dtype=dtype)
+        self.beta = nn.Parameter(torch.ones((1, c, 1, 1), device=device, dtype=dtype))
+        self.relu = nn.LeakyReLU(0.2, True)
+
+    def forward(self, x):
+        return self.relu(torch.addcmul(x, self.conv(x), self.beta))
+
+
+class IFBlock(nn.Module):
+    def __init__(self, in_planes, c=64, device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.conv0 = nn.Sequential(
+            nn.Sequential(operations.Conv2d(in_planes, c // 2, 3, 2, 1, device=device, dtype=dtype), nn.LeakyReLU(0.2, True)),
+            nn.Sequential(operations.Conv2d(c // 2, c, 3, 2, 1, device=device, dtype=dtype), nn.LeakyReLU(0.2, True)))
+        self.convblock = nn.Sequential(*(ResConv(c, device=device, dtype=dtype, operations=operations) for _ in range(8)))
+        self.lastconv = nn.Sequential(operations.ConvTranspose2d(c, 4 * 13, 4, 2, 1, device=device, dtype=dtype), nn.PixelShuffle(2))
+
+    def forward(self, x, flow=None, scale=1):
+        x = F.interpolate(x, scale_factor=1.0 / scale, mode="bilinear")
+        if flow is not None:
+            flow = F.interpolate(flow, scale_factor=1.0 / scale, mode="bilinear").div_(scale)
+            x = torch.cat((x, flow), 1)
+        feat = self.convblock(self.conv0(x))
+        tmp = F.interpolate(self.lastconv(feat), scale_factor=scale, mode="bilinear")
+        return tmp[:, :4] * scale, tmp[:, 4:5], tmp[:, 5:]
+
+
+class IFNet(nn.Module):
+    def __init__(self, head_ch=4, channels=(192, 128, 96, 64, 32), device=None, dtype=None, operations=ops):
+        super().__init__()
+        self.encode = Head(out_ch=head_ch, device=device, dtype=dtype, operations=operations)
+        block_in = [7 + 2 * head_ch] + [8 + 4 + 8 + 2 * head_ch] * 4
+        self.blocks = nn.ModuleList([IFBlock(block_in[i], channels[i], device=device, dtype=dtype, operations=operations) for i in range(5)])
+        self.scale_list = [16, 8, 4, 2, 1]
+        self.pad_align = 64
+        self._warp_grids = {}
+
+    def get_dtype(self):
+        return self.encode.cnn0.weight.dtype
+
+    def _build_warp_grids(self, H, W, device):
+        if (H, W) in self._warp_grids:
+            return
+        self._warp_grids = {}  # clear old resolution grids to prevent memory leaks
+        grid_y, grid_x = torch.meshgrid(
+            torch.linspace(-1.0, 1.0, H, device=device, dtype=torch.float32),
+            torch.linspace(-1.0, 1.0, W, device=device, dtype=torch.float32), indexing="ij")
+        self._warp_grids[(H, W)] = (
+            torch.stack((grid_x, grid_y), dim=0).unsqueeze(0),
+            torch.tensor([(W - 1.0) / 2.0, (H - 1.0) / 2.0], dtype=torch.float32, device=device))
+
+    def warp(self, img, flow):
+        return _warp(img, flow, self._warp_grids)
+
+    def extract_features(self, img):
+        """Extract head features for a single frame. Can be cached across pairs."""
+        return self.encode(img)
+
+    def forward(self, img0, img1, timestep=0.5, cache=None):
+        if not isinstance(timestep, torch.Tensor):
+            timestep = torch.full((img0.shape[0], 1, img0.shape[2], img0.shape[3]), timestep, device=img0.device, dtype=img0.dtype)
+
+        self._build_warp_grids(img0.shape[2], img0.shape[3], img0.device)
+
+        B = img0.shape[0]
+        f0 = cache["img0"].expand(B, -1, -1, -1) if cache and "img0" in cache else self.encode(img0)
+        f1 = cache["img1"].expand(B, -1, -1, -1) if cache and "img1" in cache else self.encode(img1)
+        flow = mask = feat = None
+        warped_img0, warped_img1 = img0, img1
+        for i, block in enumerate(self.blocks):
+            if flow is None:
+                flow, mask, feat = block(torch.cat((img0, img1, f0, f1, timestep), 1), None, scale=self.scale_list[i])
+            else:
+                fd, mask, feat = block(
+                    torch.cat((warped_img0, warped_img1, self.warp(f0, flow[:, :2]), self.warp(f1, flow[:, 2:4]), timestep, mask, feat), 1),
+                    flow, scale=self.scale_list[i])
+                flow = flow.add_(fd)
+            warped_img0 = self.warp(img0, flow[:, :2])
+            warped_img1 = self.warp(img1, flow[:, 2:4])
+        return torch.lerp(warped_img1, warped_img0, torch.sigmoid(mask))
+
+
+def detect_rife_config(state_dict):
+    head_ch = state_dict["encode.cnn3.weight"].shape[1]  # ConvTranspose2d: (in_ch, out_ch, kH, kW)
+    channels = []
+    for i in range(5):
+        key = f"blocks.{i}.conv0.1.0.weight"
+        if key in state_dict:
+            channels.append(state_dict[key].shape[0])
+    if len(channels) != 5:
+        raise ValueError(f"Unsupported RIFE model: expected 5 blocks, found {len(channels)}")
+    return head_ch, channels
diff --git a/comfy_extras/nodes_frame_interpolation.py b/comfy_extras/nodes_frame_interpolation.py
new file mode 100644
index 000000000..a3b00d36e
--- /dev/null
+++ b/comfy_extras/nodes_frame_interpolation.py
@@ -0,0 +1,211 @@
+import torch
+from tqdm import tqdm
+from typing_extensions import override
+
+import comfy.model_patcher
+import comfy.utils
+import folder_paths
+from comfy import model_management
+from comfy_extras.frame_interpolation_models.ifnet import IFNet, detect_rife_config
+from comfy_extras.frame_interpolation_models.film_net import FILMNet
+from comfy_api.latest import ComfyExtension, io
+
+FrameInterpolationModel = io.Custom("INTERP_MODEL")
+
+
+class FrameInterpolationModelLoader(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FrameInterpolationModelLoader",
+            display_name="Load Frame Interpolation Model",
+            category="loaders",
+            inputs=[
+                io.Combo.Input("model_name", options=folder_paths.get_filename_list("frame_interpolation"),
+                               tooltip="Select a frame interpolation model to load. Models must be placed in the 'frame_interpolation' folder."),
+            ],
+            outputs=[
+                FrameInterpolationModel.Output(),
+            ],
+        )
+
+    @classmethod
+    def execute(cls, model_name) -> io.NodeOutput:
+        model_path = folder_paths.get_full_path_or_raise("frame_interpolation", model_name)
+        sd = comfy.utils.load_torch_file(model_path, safe_load=True)
+
+        model = cls._detect_and_load(sd)
+        dtype = torch.float16 if model_management.should_use_fp16(model_management.get_torch_device()) else torch.float32
+        model.eval().to(dtype)
+        patcher = comfy.model_patcher.ModelPatcher(
+            model,
+            load_device=model_management.get_torch_device(),
+            offload_device=model_management.unet_offload_device(),
+        )
+        return io.NodeOutput(patcher)
+
+    @classmethod
+    def _detect_and_load(cls, sd):
+        # Try FILM
+        if "extract.extract_sublevels.convs.0.0.conv.weight" in sd:
+            model = FILMNet()
+            model.load_state_dict(sd)
+            return model
+
+        # Try RIFE (needs key remapping for raw checkpoints)
+        sd = comfy.utils.state_dict_prefix_replace(sd, {"module.": "", "flownet.": ""})
+        key_map = {}
+        for k in sd:
+            for i in range(5):
+                if k.startswith(f"block{i}."):
+                    key_map[k] = f"blocks.{i}.{k[len(f'block{i}.'):]}"
+        if key_map:
+            sd = {key_map.get(k, k): v for k, v in sd.items()}
+        sd = {k: v for k, v in sd.items() if not k.startswith(("teacher.", "caltime."))}
+
+        try:
+            head_ch, channels = detect_rife_config(sd)
+        except (KeyError, ValueError):
+            raise ValueError("Unrecognized frame interpolation model format")
+        model = IFNet(head_ch=head_ch, channels=channels)
+        model.load_state_dict(sd)
+        return model
+
+
+class FrameInterpolate(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FrameInterpolate",
+            display_name="Frame Interpolate",
+            category="image/video",
+            search_aliases=["rife", "film", "frame interpolation", "slow motion", "interpolate frames", "vfi"],
+            inputs=[
+                FrameInterpolationModel.Input("interp_model"),
+                io.Image.Input("images"),
+                io.Int.Input("multiplier", default=2, min=2, max=16),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
+
+    @classmethod
+    def execute(cls, interp_model, images, multiplier) -> io.NodeOutput:
+        offload_device = model_management.intermediate_device()
+
+        num_frames = images.shape[0]
+        if num_frames < 2 or multiplier < 2:
+            return io.NodeOutput(images)
+
+        model_management.load_model_gpu(interp_model)
+        device = interp_model.load_device
+        dtype = interp_model.model_dtype()
+        inference_model = interp_model.model
+
+        # Free VRAM for inference activations (model weights + ~20x a single frame's worth)
+        H, W = images.shape[1], images.shape[2]
+        activation_mem = H * W * 3 * images.element_size() * 20
+        model_management.free_memory(activation_mem, device)
+        align = getattr(inference_model, "pad_align", 1)
+
+        # Prepare a single padded frame on device for determining output dimensions
+        def prepare_frame(idx):
+            frame = images[idx:idx + 1].movedim(-1, 1).to(dtype=dtype, device=device)
+            if align > 1:
+                from comfy.ldm.common_dit import pad_to_patch_size
+                frame = pad_to_patch_size(frame, (align, align), padding_mode="reflect")
+            return frame
+
+        # Count total interpolation passes for progress bar
+        total_pairs = num_frames - 1
+        num_interp = multiplier - 1
+        total_steps = total_pairs * num_interp
+        pbar = comfy.utils.ProgressBar(total_steps)
+        tqdm_bar = tqdm(total=total_steps, desc="Frame interpolation")
+
+        batch = num_interp  # reduced on OOM and persists across pairs (same resolution = same limit)
+        t_values = [t / multiplier for t in range(1, multiplier)]
+
+        out_dtype = model_management.intermediate_dtype()
+        total_out_frames = total_pairs * multiplier + 1
+        result = torch.empty((total_out_frames, 3, H, W), dtype=out_dtype, device=offload_device)
+        result[0] = images[0].movedim(-1, 0).to(out_dtype)
+        out_idx = 1
+
+        # Pre-compute timestep tensor on device (padded dimensions needed)
+        sample = prepare_frame(0)
+        pH, pW = sample.shape[2], sample.shape[3]
+        ts_full = torch.tensor(t_values, device=device, dtype=dtype).reshape(num_interp, 1, 1, 1)
+        ts_full = ts_full.expand(-1, 1, pH, pW)
+        del sample
+
+        multi_fn = getattr(inference_model, "forward_multi_timestep", None)
+        feat_cache = {}
+        prev_frame = None
+
+        try:
+            for i in range(total_pairs):
+                img0_single = prev_frame if prev_frame is not None else prepare_frame(i)
+                img1_single = prepare_frame(i + 1)
+                prev_frame = img1_single
+
+                # Cache features: img1 of pair N becomes img0 of pair N+1
+                feat_cache["img0"] = feat_cache.pop("next") if "next" in feat_cache else inference_model.extract_features(img0_single)
+                feat_cache["img1"] = inference_model.extract_features(img1_single)
+                feat_cache["next"] = feat_cache["img1"]
+
+                used_multi = False
+                if multi_fn is not None:
+                    # Models with timestep-independent flow can compute it once for all timesteps
+                    try:
+                        mids = multi_fn(img0_single, img1_single, t_values, cache=feat_cache)
+                        result[out_idx:out_idx + num_interp] = mids[:, :, :H, :W].to(out_dtype)
+                        out_idx += num_interp
+                        pbar.update(num_interp)
+                        tqdm_bar.update(num_interp)
+                        used_multi = True
+                    except model_management.OOM_EXCEPTION:
+                        model_management.soft_empty_cache()
+                        multi_fn = None  # fall through to single-timestep path
+
+                if not used_multi:
+                    j = 0
+                    while j < num_interp:
+                        b = min(batch, num_interp - j)
+                        try:
+                            img0 = img0_single.expand(b, -1, -1, -1)
+                            img1 = img1_single.expand(b, -1, -1, -1)
+                            mids = inference_model(img0, img1, timestep=ts_full[j:j + b], cache=feat_cache)
+                            result[out_idx:out_idx + b] = mids[:, :, :H, :W].to(out_dtype)
+                            out_idx += b
+                            pbar.update(b)
+                            tqdm_bar.update(b)
+                            j += b
+                        except model_management.OOM_EXCEPTION:
+                            if batch <= 1:
+                                raise
+                            batch = max(1, batch // 2)
+                            model_management.soft_empty_cache()
+
+                result[out_idx] = images[i + 1].movedim(-1, 0).to(out_dtype)
+                out_idx += 1
+        finally:
+            tqdm_bar.close()
+
+        # BCHW -> BHWC
+        result = result.movedim(1, -1).clamp_(0.0, 1.0)
+        return io.NodeOutput(result)
+
+
+class FrameInterpolationExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            FrameInterpolationModelLoader,
+            FrameInterpolate,
+        ]
+
+
+async def comfy_entrypoint() -> FrameInterpolationExtension:
+    return FrameInterpolationExtension()
diff --git a/folder_paths.py b/folder_paths.py
index 9c96540e3..80f4b291a 100644
--- a/folder_paths.py
+++ b/folder_paths.py
@@ -52,6 +52,8 @@ folder_names_and_paths["model_patches"] = ([os.path.join(models_dir, "model_patc
 
 folder_names_and_paths["audio_encoders"] = ([os.path.join(models_dir, "audio_encoders")], supported_pt_extensions)
 
+folder_names_and_paths["frame_interpolation"] = ([os.path.join(models_dir, "frame_interpolation")], supported_pt_extensions)
+
 output_directory = os.path.join(base_path, "output")
 temp_directory = os.path.join(base_path, "temp")
 input_directory = os.path.join(base_path, "input")
diff --git a/main.py b/main.py
index 12b04719d..dbaf2745c 100644
--- a/main.py
+++ b/main.py
@@ -9,6 +9,8 @@ import folder_paths
 import time
 from comfy.cli_args import args, enables_dynamic_vram
 from app.logger import setup_logger
+setup_logger(log_level=args.verbose, use_stdout=args.log_stdout)
+
 from app.assets.seeder import asset_seeder
 from app.assets.services import register_output_files
 import itertools
@@ -27,8 +29,6 @@ if __name__ == "__main__":
     os.environ['HF_HUB_DISABLE_TELEMETRY'] = '1'
     os.environ['DO_NOT_TRACK'] = '1'
 
-setup_logger(log_level=args.verbose, use_stdout=args.log_stdout)
-
 faulthandler.enable(file=sys.stderr, all_threads=False)
 
 import comfy_aimdo.control
diff --git a/manager_requirements.txt b/manager_requirements.txt
index f770ec933..a079d3492 100644
--- a/manager_requirements.txt
+++ b/manager_requirements.txt
@@ -1 +1 @@
-comfyui_manager==4.1
+comfyui_manager==4.2.1
diff --git a/models/frame_interpolation/put_frame_interpolation_models_here b/models/frame_interpolation/put_frame_interpolation_models_here
new file mode 100644
index 000000000..e69de29bb
diff --git a/nodes.py b/nodes.py
index 299b3d758..bb38e07b8 100644
--- a/nodes.py
+++ b/nodes.py
@@ -2457,7 +2457,8 @@ async def init_builtin_extra_nodes():
         "nodes_number_convert.py",
         "nodes_painter.py",
         "nodes_curve.py",
-        "nodes_rtdetr.py"
+        "nodes_rtdetr.py",
+        "nodes_frame_interpolation.py",
     ]
 
     import_failed = []
diff --git a/requirements.txt b/requirements.txt
index ccdd47674..a25bc0667 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,5 +1,5 @@
 comfyui-frontend-package==1.42.14
-comfyui-workflow-templates==0.9.57
+comfyui-workflow-templates==0.9.59
 comfyui-embedded-docs==0.4.3
 torch
 torchsde