<h3 id="a-ready-to-run-comfyui-workflow-for-multi-shot-talking-character-video:-one-recurring-character-across-completely-different-locations-with-the-same-face-and-the-same-voice-audio-generated-together-with-the-picture-not-dubbed.">A ready-to-run ComfyUI workflow for multi-shot talking-character video: one recurring character across completely different locations, with the same face and the same voice — audio generated together with the picture, not dubbed.</h3>The showcase video is exactly what the workflow does: five shots, five locations, one woman, one voice. Nothing carries her between shots except a single identity sentence repeated byte-identically in every prompt, plus the cross-shot memory bank this graph wires up for you.Before you dive in — a word on expectations. This is a community-built, bleeding-edge pipeline: a 22B audio+video model with cross-shot memory, running on consumer hardware. It is powerful, but it is not one-click — your first clean render will likely take some tuning to YOUR machine (VRAM, system RAM and pagefile, and which model build). No two rigs behave identically. <code>INSTRUCTIONS.md</code> in the zip covers install, per-VRAM settings, and every failure mode reported so far. If you get stuck, comment here or open a GitHub issue — I answer, I troubleshoot, and most of the fixes in this pack exist because someone reported something. You will not be left hanging.<h2 id="quick-fixes-read-this-first">Quick fixes — read this first</h2>Nearly every problem reported with this pack is one of these.<ul><li>Renders crawl; VRAM pinned at 100% Why: Without <code>fp8_scaled_mm</code> the DiT runs in bf16 — ~40 GB staged. On a 32 GB card that streams over PCIe every step. Fix: Turn <code>fp8_scaled_mm</code> ON (with a bf16 or INT8 checkpoint) and <code>sequential_offload</code> OFF — ~22 GB resident, native fp8 matmul on RTX 40/50-series. It is off by default and is the single biggest speed setting in this pack.</li><li><code>fp8_scaled_mm</code> errors — “needs the bf16 checkpoint as its source” Why: The toggle downcasts the attention/FF linears itself, so it must start from bf16. An fp8 FILE would load every tensor as fp8 — norms, tables and adalns included — and crash the denoise pipeline. Fix: Point <code>model_file</code> at the bf16 (or INT8) build. Do not pair the fp8 file with this toggle.</li><li>The fp8 file didn't speed anything up Why: With the toggles off, an fp8 checkpoint upcasts to bf16 at load. It saves download size, not memory or time. Fix: Use bf16 or INT8 + <code>fp8_scaled_mm</code> for the actual win.</li><li>The INT8 file is slow Why: You ran it without <code>fp8_scaled_mm</code>. INT8 reconstructs to bf16 at load, so on its own it is the bf16 path plus ~40 s of reconstruction. Fix: Turn <code>fp8_scaled_mm</code> ON. INT8 then behaves exactly like bf16 as a source — same ~22 GB resident — at roughly 60% of the download.</li><li>Lip sync drifts apart ~10 s into every shot Why: The wrapper's video RoPE clock was hardcoded to 24 fps while audio RoPE runs in true seconds — a 25 fps render drifts ~4%/s. Looks like a model limit; it is not. Fix: Apply the multishot patch (Bug fix #0). No checkpoint can fix it. With the patch, 60–105 s multishot masters hold sync.</li><li><code>memory_size=0</code> every shot; a new face each shot Why: With <code>enable_audio_memory</code> off, the video memory-bank save was gated on the audio latent, so the bank never filled. Fix: Fixed in the patch (Bug fix #1). Console <code>memory_size=</code> should climb 0,1,2,… up to your cap.</li><li>Quality degrades over a long run — waxy skin, smearing by the late shots Why: The memory-bank trim was a no-op when <code>memory_max_size &lt;= num_fix_frames</code>, so the bank grew unbounded. Fix: Fixed in the patch (Bug fix #1b). <code>memory_size=</code> now freezes at your cap.</li><li>A GGUF errors about VAEs Why: A GGUF is DiT-only. Fix: Keep a full bf16 checkpoint in <code>checkpoint_path</code> — it supplies the VAEs, vocoder and connectors.</li><li>Burned-in subtitles or captions Why: The DMD pipeline has no CFG, so a plain negative does nothing. Fix: Use <code>negative_prompt_video</code> / <code>negative_scale_video</code> (~0.5). Above ~0.8 it locks every shot to shot 1's composition.</li><li>The voice comes out British Why: LTX drifts British unless told otherwise. Fix: Name it in the positive: “in a casual American accent”.</li></ul><h2 id="what's-in-the-zip">What's in the zip</h2><ul><li>The workflow (11 nodes, grouped and labeled: Prompt Source → Enhance &amp; Encode → Models &amp; Reference → Generate → Finish &amp; Save)</li><li>Built-in note blocks explaining which widgets to set, which to leave alone, and what to change on bigger cards</li><li>A 2-shot example prompt file demonstrating the identity-sentence rule</li><li>A README with setup, prompt-writing guidance, and performance notes</li></ul><h2 id="what-you-need-(not-included)">What you need (not included)</h2><ul><li>Node pack: RealRebelAI's <code>ComfyUI_JoyAI_Echo_GGUF_Nodes</code> plus <a target="_blank" rel="ugc" href="https://github.com/jlucasmcrell/joyai-echo-multishot-patch">my multishot patch</a> (v1.2 or newer required)</li><li>Checkpoint: the echoVid+ltxAud surgical merge — bf16 (43 GB) or the INT8 ConvRot build (27 GB)</li><li>Text encoder: any Gemma-3-12B-it, single-file safetensors or GGUF</li></ul><h2 id="setup-is-two-things">Setup is two things</h2><ul><li>Pick your Gemma in the Model Loader's <code>gemma_file</code> dropdown.</li><li>Drop <code>example_multishot.json</code> into <code>ComfyUI/input/joyecho_prompts/</code> and pick it in the Prompt Source node. Queue.</li></ul><h2 id="everything-else-i've-published">Everything else I've published</h2><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino">All models on Hugging Face</a></li><li><a target="_blank" rel="ugc" href="https://github.com/jlucasmcrell/joyai-echo-multishot-patch">JoyAI-Echo multishot patch</a> — nodes, releases, issues</li><li><a target="_blank" rel="ugc" href="https://github.com/jlucasmcrell/ltx23-multishot-lite">LTX-2.3 Multishot Lite</a> — the core-node workflow</li></ul><h2 id="support-this-work">Support this work</h2>Everything I publish — workflows, merges, quants, patches — is free and stays free. The models are non-commercial by license and I have no plans to change that.If something here saved you an evening:<ul><li><a target="_blank" rel="ugc" href="https://github.com/sponsors/jlucasmcrell">GitHub Sponsors</a> — recurring, no platform cut</li><li><a target="_blank" rel="ugc" href="https://ko-fi.com/joeygambino">Ko-fi</a> — one-off, no account needed</li><li><a target="_blank" rel="ugc" href="https://liberapay.com/joeygambino">Liberapay</a> — recurring, open-source and non-profit</li></ul>Bug reports are worth as much as money. Most of the fixes in these packages exist because somebody posted a log.

JoyAI-Echo Multishot Workflow - one character, many shots, same face + voice

<pre><code>**v1.2 — A/V sync fix, AutoFinish, encoder picker, ready-to-run workflow**

This listing goes from v1.0 straight to v1.2. v1.1 was the workflow release, which now has its own listing.

**Breaking:** JoyEcho Model Loader's widgets were reorganized — all inputs are now optional, with file pickers above their manual-path fallbacks. Saved graphs from v1.0 will load this node with shifted values: delete and re-add the node (one output wire to reconnect), or start from the bundled workflow.

**Fixed**

- **Multi-shot A/V sync drift.** Each shot's generated audio runs ~30 ms shorter than its video, and the final assembly compounded that — audio ran ~120 ms early by shot 5, so lip sync visibly degraded through the piece. Each shot's wav is now padded to its own video duration before concat; sync is exact by construction. If multi-shot renders drifted out of sync late in a piece, this was why.
- **LLM Enhance no longer demands an API key for local endpoints.** When base_url is a local address (localhost, 127.0.0.1, 192.168.x, .local), a placeholder is sent automatically — Ollama, LM Studio and llama.cpp ignore it anyway. Cloud providers still raise a clear error naming the endpoint. An empty field previously errored even against a local server, which was the most common first-run stumble.
- **Reference Picker loads scripted refs for non-character scenes.** A scene's JSON `refs` block (`{"kelpie": "shot.png"}`) is now honored directly, so creature and location references load without the folder name having to appear in the prompt prose. Previously only names written in the prose matched.
- Hires refine no longer crashes on resolution retarget (cached seqlen mismatch).

**New**

- **AutoFinish node + worker** (missing from v1.0 entirely): after the render it automatically upscales the per-shot masters and assembles the audio-synced final MP4 — no manual upscale/concat step.
- **Model Loader: `gemma_file` dropdown** — pick the text encoder from `models/text_encoders` or `models/clip` instead of typing a path. Single-file .safetensors or .gguf encoders supported (v1.0 required the HF directory).
- **Prompt Source: `manual_path`** — point at any .txt/.json prompt file anywhere on disk, no folder conventions needed.
- **Conditioning cache** — the slow Gemma encode is cached per (prompt, encoder); re-renders with unchanged text skip the encode entirely.
- **Reference Picker:** character dropdown that survives node refreshes; default refs folder is now `input/joyecho_refs/` (one subfolder per character).
- **Writer system prompts** gained two hard-won sections: FRAMING (use shot-type nouns — "framed from the waist up" gets ignored or mis-read; keep speaking shots at medium close-up or tighter, because the VAE packs 32 px into one latent token and a wide shot puts the mouth below one token) and HUMAN MOVEMENT (state gait/turn/contact mechanics; only describe body parts that are in frame). The short-story prompt file is now included alongside the long-story one.

Canonical source and updates: https://huggingface.co/joeygambino/joyai-echo-multishot-workflow

</code></pre>

Files with SHA256: 10e09dd09c847b2e15766474f8d460dc9dd650a9c680131cba60ee7dd30497a0