Merge 75bea7db20 into 2db3b0ff90

* Update amd portable for rocm 7.2

* Update Python patch version in release workflow

.github/workflows/release-stable-all.yml

@@ -20,7 +20,7 @@ jobs:
       git_tag: ${{ inputs.git_tag }}
       cache_tag: "cu130"
       python_minor: "13"
-      python_patch: "9"
+      python_patch: "11"
       rel_name: "nvidia"
       rel_extra_name: ""
       test_release: true
@@ -65,11 +65,11 @@ jobs:
       contents: "write"
       packages: "write"
       pull-requests: "read"
-    name: "Release AMD ROCm 7.1.1"
+    name: "Release AMD ROCm 7.2"
     uses: ./.github/workflows/stable-release.yml
     with:
       git_tag: ${{ inputs.git_tag }}
-      cache_tag: "rocm711"
+      cache_tag: "rocm72"
       python_minor: "12"
       python_patch: "10"
       rel_name: "amd"

comfy/ldm/wan/vae.py

@@ -479,10 +479,12 @@ class WanVAE(nn.Module):
 
     def encode(self, x):
         conv_idx = [0]
-        feat_map = [None] * count_conv3d(self.decoder)
         ## cache
         t = x.shape[2]
         iter_ = 1 + (t - 1) // 4
+        feat_map = None
+        if iter_ > 1:
+            feat_map = [None] * count_conv3d(self.decoder)
         ## 对encode输入的x，按时间拆分为1、4、4、4....
         for i in range(iter_):
             conv_idx = [0]
@@ -502,10 +504,11 @@ class WanVAE(nn.Module):
 
     def decode(self, z):
         conv_idx = [0]
-        feat_map = [None] * count_conv3d(self.decoder)
         # z: [b,c,t,h,w]
-
         iter_ = z.shape[2]
+        feat_map = None
+        if iter_ > 1:
+            feat_map = [None] * count_conv3d(self.decoder)
         x = self.conv2(z)
         for i in range(iter_):
             conv_idx = [0]

comfy_extras/nodes_custom_sampler.py

@@ -701,7 +701,14 @@ class Noise_EmptyNoise:
 
     def generate_noise(self, input_latent):
         latent_image = input_latent["samples"]
-        return torch.zeros(latent_image.shape, dtype=latent_image.dtype, layout=latent_image.layout, device="cpu")
+        if latent_image.is_nested:
+            tensors = latent_image.unbind()
+            zeros = []
+            for t in tensors:
+                zeros.append(torch.zeros(t.shape, dtype=t.dtype, layout=t.layout, device="cpu"))
+            return comfy.nested_tensor.NestedTensor(zeros)
+        else:
+            return torch.zeros(latent_image.shape, dtype=latent_image.dtype, layout=latent_image.layout, device="cpu")
 
 
 class Noise_RandomNoise: