No lora in workflow
Bro i tried the workflow also checked the results(attached workflow) there are no talking head lora being use! In the workflow there is the lora name ohwxperson_av_lora but the result's workflow was created with no lora use!
download this lora https://huggingface.co/elix3r/LTX-2.3-22b-AV-LoRA-talking-head/blob/main/LTX-2.3-22b-AV-LoRA-talking-head-v1.safetensors and attach it replacing the ohwxperson_av_lora in the power lora node
No , i tried that already but the result is totally mismatch with the audio i evn add the transcript in the prompt as well. My point is that those vid in result section is really looking good but when i tried importing them to comfy it's showing no lora been used to generate those results. So are those created without using the lora
those were created w/o any lora. they were used in the training dataset
I see, so I've tried everything the lora, multiple settings, putting transcription in prompt etc.. but i don't what's happening the output is totally out of sync
Really sorry for that. Can you please share the reference image and prompt you're using, I will check and report back
I see, so I've tried everything the lora, multiple settings, putting transcription in prompt etc.. but i don't what's happening the output is totally out of sync
The video combine node at the very end, does it have same FPS as elsewhere in the workflow? One that is easy to overlook ;-) The FPS needs to be same at all inputs.
Been using the lora a bit my end, and seem to work great my end at least...
If you are using custom audio input, it might help to transcribe the spoken words into the text prompt
Just a quick one with Fish Audio TTS as the audio input: