External Publication
Visit Post

1st movie clip!

Hugging Face Forums [Unofficial] May 8, 2026
Source

worked so hard to figure this out and actually thought I’d done it

Created an image to video faceswap wf

load reference image

load video clip

reactor>Rife VFI>video combine

after tweaking with the settings I could replace the last 12 to 20 frames of a broken clip

it worked perfectly.. created 3 x 6 second clips, replaced the faces (last part of the clip… hardly noticed the transition)… all the clips were near perfect

Then it just stopped working - no changes - no more tweaks.. now it will only output the last 3 frames and no matter what I change I can’t get back to were it was. The output is a slit second long instead of 2 to 3 seconds before

I updated ComfyUI

I even rebuilt the w/f from scratch..

I then checked the background command window and spotted this

2026-05-08 07:08:07.6050653 [E:onnxruntime:Default, provider_bridge_ort.cc:2369 onnxruntime::TryGetProviderInfo_CUDA] E:_work\1\s\onnxruntime\core\session\provider_bridge_ort.cc:1962 onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : Error loading “D:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll” which depends on “cublasLt64_12.dll” which is missing. (Error 126: “The specified module could not be found.”)

No idea what it means, nothing changed.. please help as this really is a show stopper

Discussion in the ATmosphere

Loading comments...