Uploads from VictoryCat

Currently Playing: Make AI Characters Talk with Any Audio [LTX2.3]

In this video we will go over how we can use LTX 2.3 to generate long form talking head videos. We will be using Eleven Labs to generate the images and the v4 voice audio for these videos. This is our Eleven Labs affiliate link: https://try.elevenlabs.io/k00erxy9jlh1 If you do decide to check them out it would go a long way to supporting this channel if you choose to use this affiliate link. The workflows that we use in this video can be found here: https://github.com/VictoryCATiii3/ComfyUI-Workflows/tree/main/LTX_Audio If you want to try runpod, please use our affiliate link which will give you an extra 5 dollars on your first 5 dollar deposit. https://github.com/VictoryCATiii3/ComfyUI-Workflows/tree/main/LTX_Audio We also have a video up on how to set up runpod: https://youtu.be/PeYl2LoZKd0?si=V3OHP3aK9YZgXg-1 If you would prefer to use Wan 2.1, we have this video up: https://youtu.be/gU0RQT_Oerc Wan 2.1 in general will have better image quality lip sync, but worse motion quality if you want the characters or background to do anything during your video. Timeline: 0:00 - Intro 1:56 - Our runpod setup 4:14 - Setting up the first workflow 4:49 - Generating the image in Eleven Labs 6:30 - Eleven labs v4 voice, and how quality voices affect video quality 8:14 - Fixing the prompt based on image and audio 10:50 - First video output 11:34 - Setting up the second workflow 14:12 - Same video, second workflow 14:47 - Making a long audio in eleven labs 15:50 - Long audio output LTX 16:30 - LTX 2.3 vs Wan 2.1 for long audio 17:22 - Adding loras to your LTX videos The models necessary for this workflow are as follows: For both Workflows: file: ltx-2.3-22b-dev-fp8.safetensors location: ComfyUI/models/checkpoints source: https://huggingface.co/Lightricks/LTX-2.3-fp8/tree/main file: gemma_3_12B_it_fp4_mixed.safetensors location: ComfyUI/models/text_encoders source: https://huggingface.co/Comfy-Org/ltx-2/tree/main/split_files/text_encoders file: ltx-2.3-spatial-upscaler-x2-1.1.safetensors location: ComfyUI/models/latent_upscale_models source: https://huggingface.co/Lightricks/LTX-2.3/tree/main *** ComfyUI workflow only *** file: ltx-2.3-id-lora-talkvid-3k.safetensors location: ComfyUI/models/loras source: https://huggingface.co/Comfy-Org/ltx-2.3/tree/main/split_files/loras file: ltx_2.3_distilled_1.1_lora_dynamic_fro09_avg_rank_111_bf16.safetensors location: ComfyUI/models/loras source: https://huggingface.co/Comfy-Org/ltx-2.3/tree/main/split_files/loras *** RuneXX Workflow only *** file: ltx-2.3-22b-distilled-lora-384.safetensors location: ComfyUI/models/loras source: https://huggingface.co/Lightricks/LTX-2.3/tree/main file: taeltx2_3.safetensors location: ComfyUI/models/vae source: https://huggingface.co/Kijai/LTX2.3_comfy/tree/main/vae file: MelBandRoformer_fp16.safetensors location: ComfyUI/models/diffusion_models source: https://huggingface.co/Kijai/MelBandRoFormer_comfy/tree/main If you want to try another lora to try to improve the results of your video I would recommend trying `ltx-2.3-22b-ic-lora-lipdub-0.9.safetensors` https://huggingface.co/buckets/raubatz/ltx/tree/Lightricks/LTX-2.3-22b-IC-LoRA-LipDub Or the official lightricks lora: https://huggingface.co/Lightricks/LTX-2.3-22b-IC-LoRA-DubIt/tree/main Here are all of RuneXX's LTX workflows: https://huggingface.co/RuneXX/LTX-2.3-Workflows #localai #educational #tutorial #technology #aivideogeneratorfree