Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

A different way of handling multiple images passed to SVD.

comfyanonymous61a123a1e083
diff --git a/comfy/model_base.py b/comfy/model_base.py
index 786c9cf47..253ea6667 100644
--- a/comfy/model_base.py
+++ b/comfy/model_base.py
@@ -303,7 +303,7 @@ class SVD_img2vid(BaseModel):
if latent_image.shape[1:] != noise.shape[1:]:
latent_image = utils.common_upscale(latent_image, noise.shape[-1], noise.shape[-2], "bilinear", "center")
- latent_image = utils.repeat_to_batch_size(latent_image, noise.shape[0])
+ latent_image = utils.resize_to_batch_size(latent_image, noise.shape[0])
out['c_concat'] = comfy.conds.CONDNoiseShape(latent_image)
diff --git a/comfy/utils.py b/comfy/utils.py
index 294bbb425..505577047 100644
--- a/comfy/utils.py
+++ b/comfy/utils.py
@@ -239,6 +239,26 @@ def repeat_to_batch_size(tensor, batch_size):
return tensor.repeat([math.ceil(batch_size / tensor.shape[0])] + [1] * (len(tensor.shape) - 1))[:batch_size]
return tensor
+def resize_to_batch_size(tensor, batch_size):
+ in_batch_size = tensor.shape[0]
+ if in_batch_size == batch_size:
+ return tensor
+
+ if batch_size <= 1:
+ return tensor[:batch_size]
+
+ output = torch.empty([batch_size] + list(tensor.shape)[1:], dtype=tensor.dtype, device=tensor.device)
+ if batch_size < in_batch_size:
+ scale = (in_batch_size - 1) / (batch_size - 1)
+ for i in range(batch_size):
+ output[i] = tensor[min(round(i * scale), in_batch_size - 1)]
+ else:
+ scale = in_batch_size / batch_size
+ for i in range(batch_size):
+ output[i] = tensor[min(math.floor((i + 0.5) * scale), in_batch_size - 1)]
+
+ return output
+
def convert_sd_to(state_dict, dtype):
keys = list(state_dict.keys())
for k in keys: