The size of tensor a (145) must match the size of tensor b (150) at non-singleton dimension 3

[bluusun]

Loading model /root/.cache/huggingface/hub/models--lllyasviel--control_v11f1e_sd15_tile/snapshots/3f877705c37010b7221c3d10743307d6b5b6efac/diffusion_pytorch_model.bin onto cuda backend...

  0%|          | 0/40 [00:00<?, ?it/s]

Traceback (most recent call last):
File "/usr/local/bin/imagine", line 7, in
imaginairy.cli.imagine.imagine_cmd()
File "/usr/local/lib/python3.10/site-packages/click/core.py", line 1130, in call
return self.main(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/click/core.py", line 1055, in main
rv = self.invoke(ctx)
File "/usr/local/lib/python3.10/site-packages/click/core.py", line 1404, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/usr/local/lib/python3.10/site-packages/click/core.py", line 760, in invoke
return __callback(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/click/decorators.py", line 26, in new_func
return f(get_current_context(), *args, **kwargs)
File "/usr/local/lib/python3.10/contextlib.py", line 79, in inner
return func(*args, **kwds)
File "/usr/local/lib/python3.10/site-packages/imaginairy/cli/imagine.py", line 187, in imagine_cmd
return _imagine_cmd(
File "/usr/local/lib/python3.10/site-packages/imaginairy/cli/shared.py", line 220, in _imagine_cmd
filenames = imagine_image_files(
File "/usr/local/lib/python3.10/site-packages/imaginairy/api/generate.py", line 90, in imagine_image_files
for result in imagine(
File "/usr/local/lib/python3.10/site-packages/imaginairy/api/generate.py", line 237, in imagine
result = generate_single_image(
File "/usr/local/lib/python3.10/site-packages/imaginairy/api/generate_refiners.py", line 336, in generate_single_image
x = sd(
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/imaginairy/vendored/refiners/foundationals/latent_diffusion/model.py", line 94, in forward
unconditional_prediction, conditional_prediction = self.unet(latents).chunk(2)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/usr/local/lib/python3.10/site-packages/imaginairy/vendored/refiners/fluxion/layers/chain.py", line 279, in forward
result = self._call_layer(layer, name, *intermediate_args)
File "/usr/local/lib/python3.10/site-packages/imaginairy/vendored/refiners/fluxion/layers/chain.py", line 273, in _call_layer
raise ChainError(message) from None
imaginairy.vendored.refiners.fluxion.layers.chain.ChainError:

RuntimeError:
The size of tensor a (145) must match the size of tensor b (150) at non-singleton dimension 3

(CHAIN)
├── Conv2d(in_channels=4, out_channels=320, kernel_size=(3, 3), padding=(1, 1))
├── >>> (RES) Residual() | SD1ControlnetAdapter.SD1ControlnetAdapter.SD1UNet.Controlnet_1.DownBlocks.Chain_1.Residual
│ ├── UseContext(context=controlnet, key=condition_details)
│ └── (CHAIN) ConditionEncoder()
│ ├── (CHAIN) #1 ...
│ ├── (CHAIN) #2 ...
│ ├── (CHAIN) #3 ...
│ ├── (CHAIN) #4 ...
│ └── Conv2d(in_channels=256, out_channels=320, kernel_size=(3, 3), padding=(1, 1))
└── (PASS)
0: Tensor(shape=(2, 320, 99, 150), dtype=float16, device=cuda:0, min=-4.19, max=3.85, mean=0.01, std=0.40, norm=1238.00, grad=False)

(CHAIN) DownBlocks(in_channels=4)
├── >>> (CHAIN) | SD1ControlnetAdapter.SD1ControlnetAdapter.SD1UNet.Controlnet_1.DownBlocks.Chain_1 #1
│ ├── Conv2d(in_channels=4, out_channels=320, kernel_size=(3, 3), padding=(1, 1))
│ ├── (RES) Residual()
│ │ ├── UseContext(context=controlnet, key=condition_details)
│ │ └── (CHAIN) ConditionEncoder() ...
│ └── (PASS)
│ ├── Conv2d(in_channels=320, out_channels=320, kernel_size=(1, 1))
│ └── Lambda(_store_residual(x: torch.Tensor))
├── (CHAIN) (x2) #2
│ ├── (SUM) ResidualBlock(in_channels=320, out_channels=320)
0: Tensor(shape=(2, 4, 99, 150), dtype=float16, device=cuda:0, min=-4.61, max=4.02, mean=-0.18, std=1.12, norm=389.50, grad=False)

(PASS) Controlnet(name=details)
├── (PASS) TimestepEncoder()
│ ├── UseContext(context=diffusion, key=timestep)
│ ├── (CHAIN) RangeEncoder(sinuosidal_embedding_dim=320, embedding_dim=1280)
│ │ ├── Lambda(compute_sinuosoidal_embedding(x: jaxtyping.Int[Tensor, '*batch 1']) -> jaxtyping.Float[Tensor, '*batch 1 embedding_dim'])
│ │ ├── Converter(set_device=False)
│ │ ├── Linear(in_features=320, out_features=1280) #1
│ │ ├── SiLU()
│ │ └── Linear(in_features=1280, out_features=1280) #2
│ └── SetContext(context=range_adapter, key=timestep_embedding_details)
├── Slicing(dim=1, end=4)
0: Tensor(shape=(2, 4, 99, 150), dtype=float16, device=cuda:0, min=-4.61, max=4.02, mean=-0.18, std=1.12, norm=389.50, grad=False)

(CHAIN) SD1UNet(in_channels=4)
├── >>> (PASS) Controlnet(name=details) | SD1ControlnetAdapter.SD1ControlnetAdapter.SD1UNet.Controlnet_1 #1
│ ├── (PASS) TimestepEncoder()
│ │ ├── UseContext(context=diffusion, key=timestep)
│ │ ├── (CHAIN) RangeEncoder(sinuosidal_embedding_dim=320, embedding_dim=1280) ...
│ │ └── SetContext(context=range_adapter, key=timestep_embedding_details)
│ ├── Slicing(dim=1, end=4)
│ ├── (CHAIN) DownBlocks(in_channels=4)
│ │ ├── (CHAIN) #1 ...
│ │ ├── (CHAIN) (x2) #2 ...
│ │ ├── (CHAIN) #3 ...
0: Tensor(shape=(2, 4, 99, 150), dtype=float16, device=cuda:0, min=-4.61, max=4.02, mean=-0.18, std=1.12, norm=389.50, grad=False)

[bluusun]

This seems to be related to --size 720p / --size 1080p. It seems the older specification with separate width/height did not have that limitation.

Now when using a control image smaller than the specified image the error occurs.

Leaving out sizes works (but images are much smaller)