LTX-2.3 22B GGUF / Safetensors 12GB VRAM ComfyUI Workflow T2V, I2V, A2V FLF, LipSync Audio #ltx23 #ltx #comfyui — v1.2

Use this ComfyUI workflow with LTXV 2.3 to create video through text-to-video, image-to-video, audio-to-video, first/last-frame, and lip-sync modes. It uses First Frame and Last Frame images plus an audio file when the relevant options are active, with video as the output.

Load a First Frame image and a Last Frame image.

Load an audio file in the audio loader; it is not used unless you activate the relevant options.

Install the correct version of llama-cpp-python before use; the instructions include a Windows batch-file option and pip commands for a wheel.

Before opening the workflow, update or download ComfyUI-RT-LTX2-RareTools, ComfyUI-KJNodes, ComfyUI-LTXVideo, ComfyUI-GGUF, ComfyUI-Easy-Use, and ComfyLiterals.

The listed model files include gemma_3_12B_it_fp8_e4m3fn.safetensors, gemma-3-12b-it-Q4_K_M.gguf, ltx-2-19b-distilled.safetensors, ltx-2.3-22b-distilled_transformer_only_fp8_input_scaled_v2.safetensors, and ltx-2.3-22b-distilled-Q8_0.gguf.

Other listed model files are ltx-2.3-spatial-upscaler-x2-1.1.safetensors, LTX23_audio_vae_bf16.safetensors, LTX23_video_vae_bf16.safetensors, taeltx2_3.safetensors, and ltx-2.3_text_projection_bf16.safetensors.

The listed node packs are ComfyLiterals, comfymath, comfyui-easy-use, ComfyUI-GGUF, comfyui-kjnodes, ComfyUI-LTXVideo, and comfyui-videohelpersuite.

The workflow is set to load random images for both First Frame and Last Frame.

It loads a random audio file in the audio loader; that file is not used unless the relevant options are activated.

Before use, verify that every listed model file is present under its exact filename.

Vorschlag · nicht geprüft

Before opening the workflow, verify that comfymath and ComfyUI-GGUF are installed as node packs.

Vorschlag · nicht geprüft