search

Haoming02 / comfyui-old-photo-restoration

Custom Nodes for ComfyUI for Bringing Old Photo Back to Life
MIT License
14 stars 4 forks source link

OOM with the upscale even on 512 with 16G #4

Open Wprussell opened 2 days ago

Wprussell commented 2 days ago

Seems like there is no CPU failover implemented... Even though... 16 Is Huge for this.... I know you are limited by the architecture. But if you can somehow improve the memory management This would be fantastic. So much simpler and cleaner then BOPBTL

!!! Exception during processing !!! Allocation on device Traceback (most recent call last): File "C:\AI\ComfyUI\comfyui\execution.py", line 323, in execute output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\AI\ComfyUI\comfyui\execution.py", line 198, in get_output_data return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\AI\ComfyUI\comfyui\execution.py", line 169, in _map_node_over_list process_inputs(input_dict, i) File "C:\AI\ComfyUI\comfyui\execution.py", line 158, in process_inputs results.append(getattr(obj, func)(**inputs)) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\AI\ComfyUI\ComfyUI\custom_nodes\comfyui-old-photo-restoration\nodes.py", line 95, in test output: Image = global_test( ^^^^^^^^^^^^ File "C:\AI\ComfyUI\ComfyUI\custom_nodes\comfyui-old-photo-restoration\Global\test.py", line 139, in global_test generated = model.inference(input_image, mask) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\AI\ComfyUI\ComfyUI\custom_nodes\comfyui-old-photo-restoration\Global\models\mapping_model.py", line 439, in inference label_feat_map = self.mapping_net.inference_forward( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\AI\ComfyUI\ComfyUI\custom_nodes\comfyui-old-photo-restoration\Global\models\NonLocal_feature_mapping_model.py", line 192, in inference_forward x2 = self.NL_scale_1.inference_forward(x1, mask) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\AI\ComfyUI\ComfyUI\custom_nodes\comfyui-old-photo-restoration\Global\models\networks.py", line 772, in inference_forward concat_1=torch.cat((z,composed_fold,mask),dim=1) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ torch.OutOfMemoryError: Allocation on device

Got an OOM, unloading all loaded models. Prompt executed in 25.79 seconds

Haoming02 commented 1 day ago

Set gpu_id to -1 to use CPU

Also, I was able to process a 800x512 image on a RTX 3060 (12 GB VRAM) as shown in the README, so not sure what is happening for you.