{ "v": 1, "nodes": [ { "id": "c-intent", "type": "comment", "x": 80, "y": -60, "fields": { "text": "Generates a still image from a prompt, then animates it into a short video clip.", "color": "yellow" }, "w": 340 }, { "id": "n1", "type": "text", "x": -36.63060546098497, "y": 80.63198528663212, "fields": { "text": "a tiny sailboat on a glowing bioluminescent sea, night, stars" }, "name": "the image being" }, { "id": "n2", "type": "image", "x": 400, "y": 100, "fields": { "model": "qwen-image-3", "size": "1024x1024", "variations": "1" }, "sizes": { "prompt": 0 } }, { "id": "n3", "type": "ivideo", "x": 820, "y": 100, "fields": { "model": "ltx-2.3-quality", "prompt": "slow push-in, gentle waves, camera drifts forward", "resolution": "480p", "aspect": "16:9", "duration": "5", "modelOpts": {}, "loras": [ { "url": "", "strength": "" } ] }, "sizes": { "prompt": 0 } }, { "id": "n4", "type": "llm", "x": 414.8945610165405, "y": 605.2625832387628, "fields": { "model": "openai/gpt-5.6-terra", "prompt": "" }, "sizes": { "system": 0, "prompt": 0 } }, { "id": "n5", "type": "text", "x": -190.00000305175774, "y": 679.9999633789063, "fields": { "text": "You are animating this image for ltx-2.3.\nPromptig guide:\n1. Be More Specific. The Engine Can Handle It.\nLTX-2.3 includes a larger, more capable text connector. It interprets complex prompts more accurately, especially when they include:\nMultiple subjects\nSpatial relationships\nStylistic constraints\nDetailed actions\nPreviously, simplifying prompts improved consistency.\nNow, specificity wins.\nInstead of:\nA woman in a café\n\nTry:\nA woman in her 30s sits by the window of a small Parisian café. Rain runs down the glass behind her. Warm tungsten interior lighting. She slowly stirs her coffee while glancing at her phone. Background softly out of focus.\n0:00 / 0:06\n\nThe creative engine drifts less. Use that.\n2. Direct the Scene, Don’t Just Describe It\nLTX-2.3 is better at respecting spatial layout and relationships.\nBe explicit about:\nLeft vs right\nForeground vs background\nFacing toward vs away\nDistance between subjects\nInstead of:\nTwo people talking outside\n\nTry:\nTwo people stand facing each other on a quiet suburban sidewalk. The taller man stands on the left, hands in pockets. The woman stands on the right, holding a bicycle. Houses blurred in the background.\n\nBlock the scene like a director.\n3. Describe Texture and Material\nWith a rebuilt latent space and updated VAE, fine detail is sharper across resolutions.\nSo describe:\nFabric types\nHair texture\nSurface finish\nEnvironmental wear\nEdge detail\nExample:\nClose-up of wind moving through fine, curly hair. Individual strands visible. Soft afternoon backlight catching edge detail.\n\nYou should need less compensation in post.\n4. For Image-to-Video, Use Verbs\nOne of the biggest upgrades in 2.3 is reduced freezing and more natural motion.\nBut motion still needs clarity.\nAvoid:\nThe scene comes alive\n\nInstead:\nThe camera slowly pushes forward as the subject turns their head and begins walking toward the street. Cars pass.\nSpecify:\nWho moves\nWhat moves\nHow they move\nWhat the camera does\nMotion is driven by verbs.\n5. Avoid Static, Photo-Like Prompts\nIf your prompt reads like a still image, the output may behave like one.\nInstead of:\nA dramatic portrait of a man standing\n\nTry:\nA man stands on a windy rooftop. His coat flaps in the wind. He adjusts his collar and steps forward as the camera tracks right.\n\nAction reduces static outputs.\n6. Design for Native Portrait\nLTX-2.3 supports native vertical video up to 1080x1920, trained on vertical data.\nWhen generating portrait content, compose for vertical intentionally.\nExample:\nInfluencer vlogging while on holiday.\n\nDon’t treat vertical as cropped landscape. Frame for it.\n7. Be Clear About Audio\nThe new vocoder improves reliability and alignment.\nIf you want sound, describe it:\nEnvironmental audio\nTone and intensity\nDialogue clarity\nExample:\nA low, pulsing energy hum radiates from the glowing orb. A sharp, intermittent alarm blares in the background, metallic and urgent, echoing through the spacecraft interior.\n\nSpecific inputs produce more controlled outputs.\n8. Unlock More Complex Shots\nEarlier checkpoints rewarded simplicity.\nLTX-2.3 rewards direction.\nWith significantly stronger prompt adherence and improved visual quality, you can now design more ambitious scenes with confidence.\n\nYou can:\nLayer multiple actions within a single shot\nCombine detailed environments with character performance\nIntroduce precise stylistic constraints\nDirect camera movement alongside subject motion\nNo preamble or postamble", "optional": true }, "w": 418, "sizes": { "text": 354 }, "name": "motion system prompt" }, { "id": "n8", "type": "text", "x": -9.368380319129095, "y": 389.52823917260065, "fields": { "text": "sailing very fast in rough seas" }, "w": 254, "sizes": { "text": 98 }, "name": "movement" } ], "links": [ { "id": "l1", "from": { "node": "n1", "port": "text" }, "to": { "node": "n2", "port": "prompt" } }, { "id": "l2", "from": { "node": "n2", "port": "image" }, "to": { "node": "n3", "port": "image" } }, { "id": "l3", "from": { "node": "n2", "port": "image" }, "to": { "node": "n4", "port": "img1" } }, { "id": "l4", "from": { "node": "n5", "port": "text" }, "to": { "node": "n4", "port": "system" } }, { "id": "l7", "from": { "node": "n8", "port": "text" }, "to": { "node": "n4", "port": "prompt" } }, { "id": "l8", "from": { "node": "n4", "port": "text" }, "to": { "node": "n3", "port": "prompt" } } ], "nid": 9, "lid": 9, "view": { "panX": 206.7787288640572, "panY": 27.32052487753026, "scale": 1.0160530740967995 } }