Why the weird limit ?
Started testing the model on comfyUI while it might be pretty good, I did come across the weirdest thing, if you run the model at any other resolution apart from 1024 and at a 1:1 aspect ratio, it throws an error. Initially I assumed and was almost convinced this error was exclusive to int8 convrot but apparently it's the same when using the full weights bf16 model as well.
So please enlighten me why is there such a strange limit on an editing model ? I remember the devs were so fixated and hell bent when they hadn't received native support on comfyUI but a weird limitation as such on a model that's released in 2026 is just strange.
I kind of already know the kind of reply I'd have to expect for this message of mine, it'll likely be blamed on the comfyUI implementation.
Hi, I’d be happy to help you. But I haven’t been able to reproduce your issue (all these ratios work fine on my machine).
Could you provide a more detailed description, such as the actual workflow you’re using?
Hi, I’d be happy to help you. But I haven’t been able to reproduce your issue (all these ratios work fine on my machine).
Could you provide a more detailed description, such as the actual workflow you’re using?
I've been working with a real basic workflow tbh,
Node threw an error during execution.
ComfyUI Error Report
Error Details
- Node ID: 178
- Node Type: KSampler
- Exception Type: ValueError
- Exception Message: ValueError: JoyImageEdit: reference latent spatial/temporal shape (1, 16, 1, 144, 112) must match noise (1, 16, 1, 128, 128).
Stack Trace
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\execution.py", line 543, in execute
output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\execution.py", line 342, in get_output_data
return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\execution.py", line 316, in _async_map_node_over_list
await process_inputs(input_dict, i)
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\execution.py", line 304, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\nodes.py", line 1607, in sample
return common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\nodes.py", line 1571, in common_ksampler
samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\sample.py", line 74, in sample
samples = sampler.sample(noise, positive, negative, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\custom_nodes\ComfyUI-TiledDiffusion\utils.py", line 51, in KSampler_sample
return orig_fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\custom_nodes\ComfyUI-SeFi\sefi\integration\sampling.py", line 303, in _patched_ksampler_sample
return _ORIGINAL_KSAMPLER_SAMPLE(self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1444, in sample
return sample(self.model, noise, positive, negative, cfg, self.device, sampler, sigmas, self.model_options, latent_image=latent_image, denoise_mask=denoise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1334, in sample
return cfg_guider.sample(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1316, in sample
output = executor.execute(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1254, in outer_sample
output = self.inner_sample(noise, latent_image, device, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1229, in inner_sample
samples = executor.execute(self, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\custom_nodes\ComfyUI-TiledDiffusion\utils.py", line 34, in KSAMPLER_sample
return orig_fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 999, in sample
samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\venv_py312\Lib\site-packages\torch\utils\_contextlib.py", line 124, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\k_diffusion\sampling.py", line 205, in sample_euler
denoised = model(x, sigma_hat * s_in, **extra_args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 639, in __call__
out = self.inner_model(x, sigma, model_options=model_options, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1202, in __call__
return self.outer_predict_noise(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "Z:\AI\Comfy_UI_Standalone\ComfyUI\comfy\samplers.py", line 1209, in outer_predict_noise... (26 KB left)
This is what I get exactly when running the model at any other resolution apart from 1024 nd 1:1 aspect ratio.
I'm sorry, but I still haven't been able to reproduce your issue. I've followed your workflow and connected everything step by step, and it still works perfectly on my machine.
I'll send you the workflow file; please try running it again in your environment.
From your error message, it appears that the input latent sizes for Positive/Negative and KSampler are inconsistent. However, I haven't found any issues in this workflow. I guess it comes from your ComfyUI-TiledDiffusion or ComfyUI-SeFi, maybe.
Alternatively, you can try this workflow from our repository, which I'm certain works correctly.
I'm sorry, but I still haven't been able to reproduce your issue. I've followed your workflow and connected everything step by step, and it still works perfectly on my machine.
I'll send you the workflow file; please try running it again in your environment.
From your error message, it appears that the input latent sizes for Positive/Negative and KSampler are inconsistent. However, I haven't found any issues in this workflow. I guess it comes from your ComfyUI-TiledDiffusion or ComfyUI-SeFi, maybe.
Alternatively, you can try this workflow from our repository, which I'm certain works correctly.
Cool, I'll give this a try, at this point I'm guessing the error I get could only be exclusive to the int8 convrot I've been trying to use.
int8 convrot also works, the issue might come from your environment. I will also keep an eye out to see if anyone else encounters this problem. If it is confirmed to be a compatibility issue, I will note it in the model card. 🙂
int8 convrot also works, the issue might come from your environment. I will also keep an eye out to see if anyone else encounters this problem. If it is confirmed to be a compatibility issue, I will note it in the model card. 🙂
strange, I'll try and run the model on my other installation then to check if it works.