<h1 id="ltx-2.5-image+audio-to-video-(lip-sync)"><strong>LTX-2.5 Image+Audio-to-Video (Lip Sync)</strong></h1><p>A ComfyUI workflow for <strong>LTX-2.5</strong> that adds driving-audio lip sync to the stock image-to-video template. The official 2.5 I2V workflow generates its own audio from an empty latent (no lip sync); this version encodes a real input audio track and locks it in as fixed conditioning across both the base and refine/upscale sampling passes, so the generated mouth movement follows your audio.</p><p>Built on top of Comfy's own LTX-2.5 template nodes — <span style="color:rgb(18, 184, 134)"><strong><em>no custom node packs required.</em></strong></span></p><h2 id="what-it-does"><strong>What it does</strong></h2><ul><li><p>Takes a starting image + a driving audio clip (voice/dialogue) and outputs a video with lip-synced motion.</p></li><li><p>Uses <code>LTXVAudioVAEEncode</code> to encode the input audio into a latent, then <code>SetLatentNoiseMask</code> (fully unmasked) to hold it fixed so the sampler treats it as a given rather than something to generate.</p></li><li><p>That same fixed audio latent is fed into <strong>both</strong> the base pass and the refine/upscale pass, so the second pass doesn't drift from the driving audio.</p></li></ul><h2 id="requirements"><strong>Requirements</strong></h2><ul><li><p>ComfyUI with the LTX-2.5 nodes (<code>comfy-core</code>, no extra custom nodes)</p></li><li><p>LTX-2.5 checkpoint/model files as used by the stock LTX-2.5 template</p></li><li><p>An audio VAE checkpoint (loaded via the existing <code>VAELoader</code> node in the graph)</p></li></ul><h2 id="usage"><strong>Usage</strong></h2><ol><li><p>Load the workflow into ComfyUI.</p></li><li><p>In <code>LoadImage</code>, upload your starting frame.</p></li><li><p>In <code>LoadAudio</code>, upload your driving audio (voice track).</p></li><li><p>Queue the prompt.</p></li></ol><h2 id="notes-tuning"><strong>Notes / tuning</strong></h2><ul><li><p><code>SolidMask</code> value is <code>0</code> (fully preserve the input audio latent, no noise added to it).</p></li><li><p><code>LTXVDualCFGGuider</code> is set to <code>video_cfg: 1</code>, <code>audio_cfg: 1</code> on both passes.</p></li><li><p>If sync drifts, double-check the audio latent is still wired into <strong>both</strong> <code>LTXVConcatAVLatent</code> nodes (base and refine) — this was the main bug in earlier iterations of this workflow, where the refine pass pulled audio from the base pass's own output instead of the original encoded audio.</p></li></ul><h2 id="credit"><strong>Credit</strong></h2><p>Adapted from Comfy's stock LTX-2.5 image-to-video template, with the audio-conditioning path ported over from the LTX-2.3 image+audio-to-video template.</p>

Files with SHA256: 160caafa362c46f582c1429f3db223f0653cd710fb23e587c02cceee441e300f