LTX-2 19B GGUF 12GB ComfyUI Workflows 5 TOTAL! t2v/i2v/v2v/ia2v/ta2v — I WANT ALL THE THINGSv2.0

This ComfyUI workflow generates LTX-2 video from text prompts, images, audio, or short video clips, with text-to-video, image-to-video, video-extension, text-and-audio-to-video, and image-and-audio-to-video modes.

Video extension takes a few seconds of video and a prompt to continue the video.

Two audio-enhancement nodes appear toward the end of the workflow.

Text-to-video uses a text prompt, image-to-video uses an image, and video extension uses a few seconds of video plus a prompt.

The audio modes use either a text prompt plus audio or an image plus audio; the image-and-audio mode is described as lip-syncing.

Plan for at least 12GB of VRAM and 48GB of system RAM for these GGUF workflows.

Suggestion · non vérifié

Update KJNodes and Comfy before use because a Comfy reorganization changed where the embeddings load.

Suggestion · non vérifié

The workflow names these node packs: audio-separation-nodes-comfyui, comfyui-kjnodes, comfyui-logic, and comfyui-videohelpersuite.

Named model files include gemma_3_12B_it_fp4_mixed.safetensors, ltx-2-19b-dev-Q4_K_M.gguf, ltx-2-19b-distilled-lora_resized_dynamic_fro09_avg_rank_175_fp8.safetensors, and ltx-2-19b-embeddings_connector_dev_bf16.safetensors.

Other named model files are LTX-2-Image2Vid-Adapter.safetensors, LTX2_audio_vae_bf16.safetensors, LTX2_video_vae_bf16.safetensors, and taeltx_2.safetensors.

Check the model selections for the FP4 Gemma text encoder, Tiny VAE previews, and Dev+LoRA combination before use.

Suggestion · non vérifié

Before running the workflow, verify that the named node packs are available.

Suggestion · non vérifié

Before running the workflow, verify that the named model files are available.

Suggestion · non vérifié