The alignment file setting used in the Convert tab does not work.

[Satou Suzuki]

First, these sentences are machine translated.
If you have any difficulty understanding them, please ask me again.

Hello, I am currently in the process of doing my first swap.

With the help of past questions on this forum and Google Gemini, I finally completed the Extract and Train steps. The next step is Convert.

I am currently stuck on an error message during the conversion process saying that the alignment file cannot be found. Please help.

This time, the following files are available:
-Unedited video file A showing person 1 (Video I want to swap this time)
-Unedited video file B showing person 2
-Video file C, which is a cut of file A except for the part showing person 1's face (file used in the Extract process)
-Video file D, which is a cut of file B except for the part showing person 2's face (file used in the Extract process)
-fsa file generated during extraction of video files C and D

Here's what I do:
-Specify file A as Input.
-Specify the fsa file generated from file C as Alignments.
-Specify file C as Reference Video.
-Specify the folder generated by training the combination of person 1 and person 2 as Model Dir.

If you start Convert in this state, the following error message will be displayed on every frame and the process will be completed in a very short time.
-No alignment found for (File name of file A)_******.mp4, skipping

The output directory will contain a file with the same content as file A, without any conversion.

As an experiment, I tried converting with the On The Fly setting turned on and without setting Alignment and Reference Video, and the conversion process was carried out, although the quality was very low.

That's all I could think of.
I searched the forums but couldn't find any similar cases.
Gemini was not very helpful, at least with the way I asked the question.

For reference, in my country we mainly deal with multi-byte characters, so I tried to make sure there were no multi-byte characters in the file path, but it had no effect on the results.

Can anyone tell me what's wrong?
If there's any information I'm missing, please let me know.

[torzdf]

In your example:

A should be the input video. The alignments file for A is the one you should use. No need to provide a reference video, as A is already a video.

[DeliciousCaramels]

The error indicates to me that the alignment file provided doesn't match the video.

Did you run a complete extraction pass (eg, all frames ran through the detector/ aligner) including the mask you trained your model with on video A?
Sometimes I only partially extract faces for data and later on I forget it is only partially done and "test" convert it to see how it looks, and it gives this error.