Write a premise. Get a finished clip back - picture and sound in one pass, as one take of any length or a multi-shot scene whose joins you cannot see, from a 16 GB card up.Two ComfyUI canvases and the node pack for the Joy-LTX 2.5 models: JoyAI-Echo's acting on LTX-2.5's engine. LTX-2.5 renders picture and audio together, in one generation, for as long as the card's tokens last; a bundled writer turns your idea into the shot prompts the model wants; a planner sizes the render to your card; the multishot sampler joins shots so a two-minute scene costs the VRAM of one shot. No reference photo, no LoRA, no second pass of your own.A 2-minute, four-shot bunker monologue rendered with the Multishot canvas was shown blind to a video-understanding model: one person, one voice, "a single continuous take, stable lighting", six seconds of dead air in two minutes. The joins measured below the clip's own frame-to-frame motion - the model did not find them either.<h2 id="guides">Guides</h2><ul><li>START HERE - every canvas has a note in the top-left that is the whole how-to: which box to type in, which panel to touch, what to leave alone. Read it once.</li><li>SETTINGS.md in the zip - every dial on every node, what it does and what breaks if you move it.</li></ul><h2 id="what-this-is">What this is</h2>A node pack (<code>ComfyUI-JoyLTX25</code>) and two workflows that drive the Joy-LTX 2.5 models - LTX-2.5 with JoyAI-Echo's video attention / feed-forward delta transplanted in - through the stock LTX-2.5 sampling path: two passes (a draft at render size, a refine after a latent upscale), the LTX-2.5 VAEs, the Gemma-4 text encoder. The pack adds what the stock graph lacks: a planner that sizes the passes to your card, a multishot sampler with seamless joins and an identity anchor, keyframes and image folders, and a loader that takes <code>.safetensors</code> and <code>.gguf</code> alike. The writer that turns your premise into shot prompts ships in the same zip, with our writing rules inside.<h2 id="which-canvas">Which canvas</h2><ul><li>JoyLTX25_Take - one prompt, one take, one generation. Type a scene idea; the writer writes the shot; MASTER CONTROLS sizes it. Optional lanes, each behind an on/off gate: start from a picture (I2V), end on a picture (an END keyframe at the last frame), or both. Up to 60 s in one generation on the 24 GB path (single pass, 960x544).</li><li>JoyLTX25_Multishot - one story, several shots, one video. The writer writes N shot prompts; the sampler renders each with the same two-pass pipeline and joins them. Same people in every shot without reference photos. Optional lanes: a start image for shot 1, and START / END frame folders - one image per shot, every shot opens and/or closes on its picture. VRAM stays at one shot however many shots you chain.</li></ul><h2 id="how-the-joins-work">How the joins work</h2><ul><li>continue - the previous shot's last frames and its sound are pinned into the next shot as raw latents, through a noise mask, in both passes. Pinning the second pass is what made the joins invisible: with pass 1 pinned alone the join diffs measured 6-7x the clip's own motion; pinned in both passes they sit at or below motion (0.0007-0.002 vs 0.002-0.006 in the 2-minute tests). The regenerated head is trimmed on decode.</li><li>cut - only the sound is pinned. The voice carries across a picture cut; the picture follows the new shot prompt. Real reframes when the writer gives each shot a different action.</li><li>fresh - independent shots, one file.</li></ul><h2 id="how-identity-holds-with-no-reference-photos">How identity holds with no reference photos</h2><code>identity_ref</code> attaches a frame of shot 1 to the later shots as an in-context keyframe (appended tokens, cropped away after pass 1), at <code>identity_strength</code>. Measured on a four-shot diner scene at the same seed: without it, four different faces and four different rooms; with it at 0.3-0.9, the same two people and the same booth in every shot. The cost is composition: a t=0 keyframe also says where the camera is, so <code>cuts only</code> gives you faces across cuts with the framing locked to shot 1 unless the writer moves people around; START frame folders are how you get real angle changes with the same people. <code>all shots</code> also anchors a long continue-mode take: the model's texture drifts by roughly +40-60% across four 30-second shots on its own; at 0.6 the anchor holds it flat (measured 1.00 0.88 1.11 1.10 vs 1.00 1.23 1.42 1.59 unanchored).<h2 id="master-controls-and-the-dials-that-matter">Master controls, and the dials that matter</h2>One panel: render size, take length (or seconds per shot), beat length for the writer, upscale, and an optional VRAM figure. Shipped defaults: 960x544, 8 s, <code>auto</code>.<ul><li>upscale - <code>auto</code> picks x2 / x1.5 / none from your card and the token count and prints which; <code>x2 / x1.5 / none</code> force it (stream and wait if you like). <code>none</code> is the single-pass switch - the sampler's two-pass follows it. 8 s x2 or ~14 s x1.5 at 960x544 is the 24 GB comfort zone; 30 s single pass at 1280x736 fits 24 GB.</li><li>join must match the writer's join style: continue ↔ extend take, cut ↔ continuous scene with cuts. The START HERE note lists the pairs.</li><li>identity_strength - 0.3-0.6 keeps the people and lets the framing move; 1.0 copies the framing too. 0.5 for 8 s shots, 0.6 for 30 s shots when anchoring a long take.</li><li>Fewer, longer shots - 4 x 30 s fills time better than 8 x 8 s (the writer's per-shot budget scales: a 2-minute, four-shot take came back with 6 s of dead air; eight 8-second shots had about 10 s per minute).</li><li>video_cfg on the sampler is a grade dial: 1.0 is the distilled default; 0.7-0.85 pulls saturation and contrast toward stock LTX-2.5 at the cost of the negative pass (~1.7x time).</li><li>seed_per_shot - leave it on.</li><li>RTX 50: add <code>--enable-triton-backend</code> to your launch line for comfy-native files (about 2x faster). GGUF is unaffected.</li></ul><h2 id="writing-a-premise">Writing a premise</h2>You write a premise, not a prompt. The writer writes the prompt, and its rules were each added after a specific failure: people talk in contractions; one beat per sentence; every sound named (LTX invents drones and music for anything left unsaid - "quiet room tone, a fridge hum, no music" is doing work); wardrobe and room stated every shot; dialogue sized to the shot length so speech is never crammed. Things that help from your side:<ul><li>Say who is in frame and what they wear; say where the camera is ("static medium two-shot from the end of the table").</li><li>Say "American voice" if you want one - LTX drifts toward British otherwise.</li><li>For a long continue-mode take, ask for one continuous conversation, no cuts, and let the writer pace it; for cut mode, give every shot a different action so the cuts have somewhere to go.</li><li>A camera asked to "follow behind" a talking runner tends to end up in front of them - a talking character faces the lens; if you want a follow shot, have them face away and not talk to camera.</li></ul><h2 id="quick-fixes">Quick fixes</h2><ul><li>Over-saturated, hard contrast, blown highlights - you have a v1 model file (names without <code>-v2-</code>). Those are withdrawn; download the v2 files below. The v2 files still sit a notch warmer than stock; <code>video_cfg</code> 0.7-0.85 brings them level.</li><li>The joins show - the writer's join style does not match the sampler's join (extend take ↔ continue). Or a 30-second-shot take without the anchor: set identity_ref to <code>all shots</code>, 0.6.</li><li>Dead air - longer shots; and write dialogue into the premise.</li><li>Validation error, the DiT name is "not in list" - pick your file in the loader's dropdown. The loader resolves a moved file by name when it is unique; it cannot invent one you have not downloaded.</li><li>LLM API error 404 / 503 - the writer's <code>model_name</code> is not pulled on the endpoint it points at, or a cloud model is overloaded; pick another tag you have.</li><li>Red or missing nodes - the writer pack or ComfyUI-GGUF is not installed (see Requirements). Everything else is this pack plus stock ComfyUI.</li><li>x1.5 upsampler crashes on partial load - the pack patches it at import; if you see <code>BlurDownsample</code> in a traceback, the pack did not load - check the console for <code>[JoyLTX]</code> lines.</li></ul><h2 id="the-prompt-writer-needs-a-model-you-actually-have">The prompt writer needs a model you actually have</h2>The canvases ship pointed at a local Ollama (<code>http://localhost:11434/v1</code>) with a cloud model tag in <code>model_name</code>. Pick a tag you have (<code>ollama list</code>), or any OpenAI-compatible endpoint in <code>base_url</code>. A hosted model writes better than anything that fits next to the video model; if you run the writer locally, turn on <code>unload_model_after</code> so it frees the card the moment the script is written. No LLM at all? Paste your own shot prompts into the sampler's <code>prompts</code> box, one per shot, separated by <code>---</code> lines.<h2 id="requirements">Requirements</h2><ul><li>Always: ComfyUI 0.32+ (native LTX-2.5), this pack, a Joy-LTX 2.5 DiT file, the LTX-2.5 video and audio VAEs, the Gemma-4 text encoder, the LTX-2.5 x2 latent upscaler.</li><li>The writer: <code>writer_pack\ComfyUI_JoyAI_Echo_GGUF_Nodes\</code> from this zip (RealRebelAI's pack with our prompts bundled; the Manager copy does not have them).</li><li>GGUF files (RTX 30/40): ComfyUI-GGUF.</li><li>24 GB cards, optional: <code>ltx-2.3-spatial-upscaler-x1.5-1.0.safetensors</code> from LTX-2.3, for the planner's x1.5 step (the pack patches its partial-load crash).</li></ul><ol><li>Unzip. Copy <code>ComfyUI-JoyLTX25\</code> into <code>custom_nodes\</code> - or Manager → Install via Git URL → <a target="_blank" rel="ugc" href="https://github.com/jlucasmcrell/ComfyUI-JoyLTX25">github.com/jlucasmcrell/ComfyUI-JoyLTX25</a>.</li><li>Copy <code>writer_pack\ComfyUI_JoyAI_Echo_GGUF_Nodes\</code> into <code>custom_nodes\</code>.</li><li>GGUF? Install ComfyUI-GGUF.</li><li>Models (below). Restart ComfyUI.</li><li>Open <code>ComfyUI-JoyLTX25\workflows\JoyLTX25_Take.json</code> or <code>_Multishot.json</code>, read START HERE, pick your file in the loader, Run.</li></ol><h2 id="models">Models</h2>One DiT file by card, into <code>models\diffusion_models\</code>. All on Hugging Face - <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/joyai-echo-ltx25-echoVid-gguf">GGUF</a> for RTX 30/40 (4-8x faster there than any 4-bit comfy-native file), <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/joyai-echo-ltx25-echoVid-comfy-native">comfy-native</a> for RTX 50 - and on Civitai as Joy-LTX 2.5, versions by VRAM tier. 070T30 is the default dose; 100T50 (same names) is livelier and a touch hotter.<ul><li>16 GB: <code>LTX25dist-echoVid-070T30-v2-DiT-Q4_K_S.gguf</code> (RTX 30/40) or <code>...-comfy-w4a8.safetensors</code> (RTX 50). 12 GB: <code>...-Q3_K_M.gguf</code>.</li><li>24 GB: <code>LTX25dist-echoVid-070T30-v2-DiT-Q5_K_M.gguf</code> (the default) or <code>...-comfy-mix4x8-17.0GB.safetensors</code>.</li><li>32 GB: <code>LTX25dist-echoVid-070T30-v2-DiT-comfy-int8.safetensors</code> (fastest on RTX 50) or <code>...-Q8_0.gguf</code>.</li><li>Want to apply your own distilled LoRA at your own strength? The plain <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/joyai-echo-ltx25-echoVid-dev">DEV merges</a> (also on Civitai as Joy-LTX 2.5 DEV merges): drop a Load LoRA (model only) after the loader.</li></ul>Also needed (<a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2.5">Lightricks/LTX-2.5</a>): <code>ltx-2.5-video-vae-bf16</code> + <code>ltx-2.5-audio-vae-bf16</code> → <code>models\vae\</code>; <code>ltx-2.5-latent-spatial-upscaler-x2-bf16-1.0</code> → <code>models\latent_upscale_models\</code>; text encoder <code>gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot</code> → <code>models\text_encoders\</code> (16 GB cards: the 10.6 GB <code>gemma4-12b-ltx25-comfy-w4a8</code> from <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/LTX-2.5-Quantized">LTX-2.5-Quantized</a>).<h2 id="what-is-verified-and-what-is-not">What is verified, and what is not</h2><ul><li>Verified: both canvases rendered end to end from the shipped files on a 5090 and a 3090 (T2V, I2V, END keyframe, multishot continue / cut, start and end frame folders); every quantisation in the repos rendered (Q2_K through Q8_0, int8 / w4a8 / w4a4 / nvfp4 / mixed); a 2-minute 4 x 30 s continue take and a 6 x 20 s cut sequence reviewed blind; joins measured below motion; identity held at 0.3-0.9 on a four-shot cut scene; a 60 s single take at 960x544.</li><li>Known: the v2 files render a notch warmer than stock LTX-2.5 (saturation tracks stock, contrast +7-30% on the same seed); a 30-second-shot continue take softens skin over the shots without the anchor; cut mode with the identity anchor locks framing to shot 1 unless the shots differ in action; LTX reads a talking runner as a selfie. Speed: 8 s at 960x544 two-pass, ~2 min on a 5090 (int8), ~7 min on a 3090 (Q5_K_M); 30 s single pass at 1280x736, 5 / 17 min; a 4 x 30 s take ~40 min on a 3090.</li><li>Not claimed: lip-sync scores. Two video-understanding reviewers scored the merge and stock LTX-2.5 within noise of each other on an 8 s talking head; what the merge adds is visible in the acting, not reliably in a reviewer's number.</li></ul><h2 id="credits">Credits</h2>JoyAI-Echo by JD (jdopensource/JoyAI-Echo). LTX-2.5 by Lightricks. Writer pack by RealRebelAI (ComfyUI_JoyAI_Echo_GGUF_Nodes), with our prompts bundled and attributed. Nodes, canvases and the merge by me. Licensed under the LTX-2.x Community License (inherited from both parents). Questions: comment here or open an issue on GitHub - I answer.

Joy-LTX 2.5 — One Take or a Seamless Multishot, Picture and Sound in One Pass (ComfyUI)

v1.1.0 - adds the PLUS canvases (<code>JoyLTX25_Take_PLUS</code>, <code>JoyLTX25_Multishot_PLUS</code>): a LoRA stack (four slots on the DiT; on a DEV merge the official distilled LoRA at 0.4-0.6), Refs by Name (cast characters from <code>input/joyecho_refs/&lt;name&gt;/</code> photo folders - one photo per character, and from a character's second shot on their own rendered face and voice carry into every later shot), Prompts from File (your own shot list, single file or a folder walked by an incrementing episode index - queue once per file), and audio export (flac). The loader is now JoyLTX Load Model (takes .safetensors or .gguf, resolves a moved file by name). The plain Take / Multishot canvases are unchanged apart from the loader and the v2 model names. File: <code>Joy-LTX_2.5_v1.1.0.zip</code>. Models on Hugging Face: <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/joyai-echo-ltx25-echoVid-gguf">GGUF</a> · <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/joyai-echo-ltx25-echoVid-comfy-native">comfy-native</a> · <a target="_blank" rel="ugc" href="https://huggingface.co/joeygambino/joyai-echo-ltx25-echoVid-dev">DEV merges</a>. Default DiT in the canvases: <code>LTX25dist-echoVid-070T30-v2-DiT-comfy-int8.safetensors</code> - pick your own file in the loader if your card is smaller.

Files with SHA256: 6fce54d085e15146ccd4137a345672512043019c28c05d8ff8cef06dd49f4c65