From X ¡ Motion Transfer
Motion Transfer Prompts
Curated motion control and motion transfer video prompts collected from public posts on X. Try them in Oraclip with attribution to the original creators.
đˇď¸ How I created the Spider Swing trend đ 1ď¸âŁ Create your character refe
đˇď¸ How I created the Spider Swing trend đ 1ď¸âŁ Create your character reference sheet 2ď¸âŁ Upload your character sheet + Depth Previs reference video into Seedance 2.5 3ď¸âŁ Generate the final video in @pixazoai That's it. One character. One motion reference. One cinematic superhero shot. đ¸ď¸ đŹ Comment "Marvel" and I'll send you the prompt + assets. #aitrend #trendingnow #spiderman
AI just removed the biggest barrier in faceless content đ Watched this G
AI just removed the biggest barrier in faceless content đ Watched this Glorify demo and the Motion Control feature is actually insane: You upload any short motion reference video (guy in a cap waving hands) any character photo (the girl) = AI generates a full realistic video where the new character perfectly copies every movement, timing and micro-expression. No filming. No studio. No showing your face. No expensive editor. This is exactly how solopreneurs are quietly building $5kâ$15k+/mo faceless channels in 2026: ⢠AI business / motivation Shorts ⢠Product ad creatives for ecom brands ⢠Educational explainers ⢠Story-driven Reels You can now create a weekâs worth of high-quality video content in one afternoon instead of a full week of shooting and editing. The production cost just dropped to almost zero while output went up 10x. If youâre still thinking âI need to be on camera to make money with videoâ â this kind of tool just proved that wrong. Whatâs the first video you would create with Motion Control? Drop your niche below đ
Kling Motion Control | 1 Video, 1 Photo One reference video + one refere
Kling Motion Control | 1 Video, 1 Photo One reference video + one reference photo. Now your photo can follow the exact movement from the reference video. Prompt Keep the character perfectly synced with the reference video's body movements. Keep the camera completely static. Preserve the original scale and position. No zoom, no distortion.
Here's exactly how to film yourself in a field and end up on a private b
Here's exactly how to film yourself in a field and end up on a private beach. step 1. shoot one clean video of yourself. a simple location is best, an empty field, a parking lot, a blank wall. good daylight, steady camera, your full body in frame. walk, talk, gesture naturally. this is your motion reference, so the movement matters more than the background. step 2. go to an AI video tool that supports video-to-video with a reference (OpenArt, Runway, Kling, Higgsfield all do this). upload your clip. step 3. tell it to keep your motion but rebuild the world. the prompt is the whole trick: "keep the exact body motion, timing and gestures from the reference video. replace the entire environment with a luxury tropical beach, white sand, turquoise water, a white modern villa and a yacht in the background, bright sunny daylight, cinematic." your movement stays, the field disappears. step 4. match the lighting or it looks fake. if you filmed in flat daylight, ask for "bright midday sun" so the shadows on you line up with the new scene. mismatched light is the number one thing that gives it away. step 5. generate, then fix. first result is usually 80 percent there. push back: "more realistic sand texture," "keep my face consistent," "remove the warping on my hands." 2 to 3 passes and it's clean. step 6. export, cut to 15 seconds, post. the swap between your real field and the fake beach IS the hook, so show both. the whole thing is one principle: film the performance once, generate the world as many times as you want. one take, infinite locations.
Alibaba rethinks character animation with Wan-Animate-2 Anyone who has t
Alibaba rethinks character animation with Wan-Animate-2 Anyone who has tried animating a character from a reference video knows the trade-off: you either lose identity, lose fine motion detail, or the compute cost becomes unbearable. Tongyi Lab, Alibaba's research team, just published a paper that tries to tackle all three problems at once, with a name that's already a statement: Wan-Animate-2. đŹ The team frames the existing landscape around three paradigms and their flaws. Methods based on explicit motion representations suffer from extraction errors and identity drift, implicit motion features lose fine-grained dynamics through compression, and in-context learning avoids intermediate representations but incurs prohibitive computational costs. Wan-Animate-2 removes the intermediate motion extractor entirely, letting the driving video feed directly into a redesigned Diffusion Transformer. đ Architecturally, the framework conditions video generation on a reference image and a reference video through a dual-branch DiT, introducing Time-Align RoPE for temporal alignment between denoising video tokens and reference tokens, and Sparse-Ref Attention to selectively attend to the most informative reference features. đšď¸ One of the most interesting capabilities is text-driven viewpoint control, which decouples the output camera perspective from the driving video, a capability the authors describe as rarely supported by prior character animation methods that rely on explicit motion representations. ⥠For interactive use cases, the team introduces Wan-Animate-2-Lite, an efficient variant designed to push inference latency down to real-time thresholds through a three-stage training paradigm combining teacher forcing pretraining with an error buffer mechanism and Self-Forcing distillation with chunk-wise backpropagation. According to independent reporting, this lite version reportedly reaches real-time streaming at 24 fps, opening deployment scenarios like digital avatars and live-streaming hosts that were previously infeasible at this quality level. Why it matters: if it holds up outside the demo reel, this changes the equation for anyone working with digital avatars, live streaming, or low-budget video production. No complex 3D rig, no expensive motion capture setup, just a reference video and an image. For companies and independent creators, it means access to professional-grade animation tooling without the hardware and pipeline barriers of traditional workflows, especially now that the team says it will release the Wan-Animate-2-Base model weights publicly. As a Creative & AI Specialist, what strikes me most isn't the visual quality itself, it's the decision to rethink the whole pipeline with live use cases in mind from day one: too many animation models stay stuck in offline mode because they were built for quality alone, while here the real-time variant looks like a design priority rather than an afterthought, and if the open weight release lives up to the paper, the real test will be how easily it slots into the workflows creators already rely on. đą Want AI updates on WhatsApp? DM me "AI" and I'll send you access. đ Follow me for the last AI updates! #AInews #WanAnimate2 #Alibaba #CharacterAnimation #AIVideo
Hmm.. some parts came out really good. But some failed, also because the
Hmm.. some parts came out really good. But some failed, also because the image (see first comment) supplied isn't suitable for the prompt per se. Overall, super impressed. -- Prompt share: Prompt 1 (0â10 Seconds) Prompt: 10-second sequence featuring the alien creature from alien creature.jpg, split into four distinct 2.5-second cinematic shots with bottom-left overlay text: Shot 1: Leone-style Extreme Close-Up on the main turquoise eye, capturing iris details. Text overlay in bottom-left: "ECU - Main Eye". Shot 2: Apocalypse Now-inspired low-angle tracking camera following the spindly legs from behind as the creature steps forward. Text overlay in bottom-left: "Low-Angle Tracking". Shot 3: High-contrast The Third Man-style low-angle Dutch tilt looking up at the bulbous body. Text overlay in bottom-left: "Low-Angle Dutch Tilt". Shot 4: Dynamic text screen reading "ALIEN". The dot above the 'i' detaches and bounces off the letters, causing soft-body physical bending and distortion. Transitions out with digital glitch effects. Text overlay in bottom-left: "Text VFX - Soft Body Physics". Prompt 2 (10â20 Seconds) Prompt: 10-second sequence focusing on secondary anatomy and skin texture of the alien creature from alien creature.jpg, divided into four 2.5-second shots with bottom-left text overlay: Shot 1: Vertigo-style dolly-zoom (zolly shot) pushing in on the secondary stalk-like eyes. Text overlay in bottom-left: "Dolly-Zoom - Stalk Eyes". Shot 2: Towering Metropolis-style high-angle surveillance view looking down at the creature. Text overlay in bottom-left: "High-Angle Surveillance". Shot 3: Rapid action-style whip pan starting from a detailed leg joint and snapping to a full-body wide shot. Text overlay in bottom-left: "Whip-Pan Transition". Shot 4: Text frame displaying "CREATURE". The dot on the 'i' triggers a force-field distortion effect that melts and warps the typography. Post-effects feature subtle color banding and static noise. Text overlay in bottom-left: "Text VFX - Force Field Melt". Prompt 3 (20â30 Seconds) Prompt: 10-second full-body kinematic motion sequence of the alien creature from alien creature.jpg, across four 2.5-second shots with bottom-left overlay text: Shot 1: 360-degree Matrix-style bullet-time Steadicam orbit surrounding the full bipedal structure. Text overlay in bottom-left: "360 Steadicam Orbit". Shot 2: Extreme macro tracking shot following the fluid motion of a coiling neural tentacle. Text overlay in bottom-left: "Macro Tracking - Tentacle". Shot 3: Requiem for a Dream-style 3-panel split-screen: ECU main eye, ECU secondary eye, and Medium Full shot. Text overlay in bottom-left: "Split-Screen Composite". Shot 4: Dynamic text frame with "SURVEILLANCE" wiggling with rubber-like soft body dynamics. A tentacle tip multiplies 30 times across the frame as motion graphics before fading through white noise. Text overlay in bottom-left: "Text VFX - Element Multiplication". -- Done using @FlowbyGoogle 's Omni
we live in a CRAZY ERA from now you can generate 100 videos a day in one
we live in a CRAZY ERA from now you can generate 100 videos a day in one click. with GENJUTSU on DAAS here's the exact prompt i used for this one: Create a photorealistic video using the uploaded image as the identity reference and the uploaded video as the motion reference. SUBJECT The subject is an adult woman aged 25 or older. Use the woman shown in the uploaded image as the visual identity reference. Preserve her facial features, face shape, eye color, nose shape, lip shape, skin tone, and natural hairstyle. HAIRSTYLE AND NAILS Her hair is straight and tied into a ponytail. Two blonde strands frame the front of her face. Her nails have a classic French manicure with a pale pink base and white tips. ACTION Reproduce the visible movements from the uploaded video in the same order and at the same pace. Match the head movements, hand gestures, facial expressions, body position, and direction of gaze. Do not add movements that are not present in the reference video. CAMERA Match the reference video's camera angle, framing, subject position, and duration. Keep the camera movement consistent with the original video. VISUAL CONSISTENCY Keep the same adult woman recognizable throughout the entire video. Maintain consistent facial features, hairstyle, clothing, body proportions, and nail design in every frame. IMAGE QUALITY Use realistic skin texture, natural lighting, accurate anatomy, anatomically correct hands, five fingers on each hand, realistic hair movement, and smooth transitions between frames. CLOTHING AND SCENE Keep the clothing and setting appropriate for a general audience. Preserve the original outfit when it is suitable for general audiences. Do not add revealing clothing, nudity, sexual content, or suggestive gestures. OUTPUT REQUIREMENTS Produce one continuous, photorealistic video. Follow the reference video's sequence of actions without adding unrelated scenes, extra people, text, logos, or watermarks. Prioritize identity consistency, accurate motion reproduction, and realistic visual quality.
10 camera directions for AI video prompts: Close-up shot Wide establishi
10 camera directions for AI video prompts: Close-up shot Wide establishing shot Slow tracking shot Aerial drone shot Low-angle shot Over-the-shoulder shot Handheld camera Dolly zoom Orbiting camera First-person POV Which one do you use most? https://t.co/7xHzBIVL0j
The weirdest way to make AI copy a dance better? Take the dancer out of
The weirdest way to make AI copy a dance better? Take the dancer out of the video. Sounds backwards. But this simple trick can dramatically change how motion transfer works in Seedance 2.0. When you give a video of someone dancing directly to a video model, youâre not giving it âmotion.â Youâre giving it everything: The person. Their face. Their clothes. The room. The lighting. The camera. The background. And somewhere inside all that noise â the choreography. Now ask the model to put that movement onto a completely different character. Youâre essentially asking it to solve ten problems at once. And then we wonder why the choreography drifts. The trick is to separate two things that shouldnât be mixed: WHO is moving and HOW they are moving. Hereâs the workflow: 1. Lock the character first. Create the character reference in GPT Image 2 â face, body, clothing, overall identity. The goal is simple: the characterâs identity should exist independently from the movement. 2. Strip the dancer out of the source video. Convert the dance footage into a depth map. Instead of giving the model the original performer, you give it spatial information â essentially a representation of where the body is in 3D space. No face. No clothes. No background. No visual identity. Just movement. 3. Give the model two separate references. Character sheet = identity Depth map = motion Now each input has one job. And thatâs the important part. 4. Donât let the references compete. The depth map shouldnât tell the model who is dancing. The character reference shouldnât tell it how to dance. Youâre removing the ambiguity instead of asking the model to resolve it. Thatâs why this works. Raw footage contains an enormous amount of information, and the model has to decide what matters. A depth map dramatically narrows the problem. Instead of: âWhat am I supposed to copy from this video?â the model gets something closer to: âHere is the movement. Put this character inside it.â And that distinction matters. Because better AI video isnât always about giving the model more information. Sometimes itâs about giving it less â but making every piece of information intentional. This opens up some interesting workflows: â Transfer choreography onto completely original characters â Create motion-capture-style results without a mocap suit â Preserve specific movements more accurately â Build character showcases from existing performance footage â Separate identity, motion and environment instead of letting them bleed into each other The character sheet answers: âWho is dancing?â The depth map answers: âHow are they moving?â And keeping those two questions separate may be one of the most useful tricks for getting more reliable motion transfer out of AI video models. The model doesnât necessarily need to become smarter. Sometimes you just need to stop giving it the wrong problem.
Playing with my sonâs toy cars and magnetic tiles for some Seedance 2.5
Playing with my sonâs toy cars and magnetic tiles for some Seedance 2.5 testing. The idea was simple: use real toy movement as a motion reference, then see how Seedance translates it into a cinematic real-world scene. A traffic jam made from toy cars became a full-scale road scene. A toy excavator clearing the jam became a massive construction vehicle pushing through real cars. A magnetic-tile house became an actual building attempt. Whatâs interesting is that Seedance doesnât just copy the action. It picks up on camera angle, object dominance, timing, contact points, weight, direction, and even accidental scale cues. In one clip, the toy excavator looked huge because it was close to the camera and cropped at the edge of frame. Seedance translated that almost literally, making the real excavator tower over the cars. Thatâs the fascinating bit. The reference video becomes more than choreography. It becomes a kind of physical prompt: perspective, momentum, gravity, friction, hierarchy, and staging all baked into a few seconds of footage. Basically: tabletop previs -> real-world cinematic scene. Still messy. Still unpredictable. But honestly, thatâs what makes it fun. Feels less like prompting now, and more like directing motion with whatever is lying around the house. Share this with someone who still thinks AI video is just about writing better prompts.
I tried generating a long dance video with scaiL-2: 30 FPS, 301 frames,
I tried generating a long dance video with scaiL-2: 30 FPS, 301 frames, around 10 seconds, 896Ă1600 resolution â basically a 9:16 vertical video. I also used Uni3Câs camera-motion module to get the automatic camera movement from the reference video. But honestly⌠this workflow https://t.co/VH1ad9nAbO
20 SEC SCENES - BETRAYAL đŹ Created with @thankyouai_hq . Every shot was
20 SEC SCENES - BETRAYAL đŹ Created with @thankyouai_hq . Every shot was planned with intentionâfrom claustrophobic handheld close-ups to ultra-wide cinematic reveals, using fast-cut pacing, match-on-action transitions, and grounded choreography to build tension before the first https://t.co/SN9BDhXjlG
Image to video (I2V) Start from a frame you've already designed, then de
Image to video (I2V) Start from a frame you've already designed, then describe what happens next. Useful when the look is locked and you only want motion, camera movement, and timing from the model. Source frame â animated output: https://t.co/DwS6xQhHQi
Luxury skincare, entirely imagined with AI. â¨đ§´ Built this skincare comme
Luxury skincare, entirely imagined with AI. â¨đ§´ Built this skincare commercial with MidJourney, Nano Banana Pro, and Seedance. I used MidJourney and Nano Banana Pro to create a consistent visual identity for every environment, product shot, and beauty scene, then brought https://t.co/lxF2d7fyGd
@Kling_ai CHRONICLES OF CLARA. Full 2m35s Sci-Fi Short Film generated en
@Kling_ai CHRONICLES OF CLARA. Full 2m35s Sci-Fi Short Film generated entirely with Kling 4.0 .A masterclass in narrative continuity, seamless action transitions, multi-character CQC, Directed by Velion Gen Studio. #VisualSpectacles #KlingAI #ShortFilm #Cinematic #SciFi https://t.co/f3aISRRaHh
Same moves. Same camera. My face. đş Day 1/100 exploring AI: putting your
Same moves. Same camera. My face. đş Day 1/100 exploring AI: putting yourself into the viral motion-transfer trend with Seedance 2.5. The prompt + full breakdown đ https://t.co/BUNm2VTxO0 https://t.co/DNUCtssKuk
Alibaba has open-sourced Wan Animate 2, a new video-to-video AI model de
Alibaba has open-sourced Wan Animate 2, a new video-to-video AI model designed for controlling character animation via reference driving video. Generating realistic character movements from reference video has historically struggled with non-standard body rigs and multi-subject consistency. Wan Animate 2 tackles these edge cases directly. Key capabilities: - High-fidelity expression transfer from reference video - Simultaneous tracking and control for multiple characters - Stable handling of non-standard and irregular body proportions Making weights open source allows developers to integrate controllable video animation directly into custom pipelines without relying on proprietary APIs. The bottleneck in generative video is rapidly shifting from raw video synthesis to precise local control and continuity.
A Hitchcock zoom used to require a dolly track, a follow-focus operator,
A Hitchcock zoom used to require a dolly track, a follow-focus operator, and a cinematographer who could coordinate the push-pull perfectly. Kling AI just turned it into a prompt. 20 seconds of the vertigo effect â background stretches, subject holds, zero camera rig. 222 likes in 3 days because every filmmaker watching knows what that move used to cost to shoot. The lesson for AI filmmakers: the most expensive camera moves in cinema history are now free parameters. If you can name the shot, you can generate it. Dolly zoom, tracking shot, whip pan â all of it lives in your prompt now. Take it if you need it. via @Kling_ai
Introducing SCoPE, Tencent ARCâs camera-controlled image-to-video model
Introducing SCoPE, Tencent ARCâs camera-controlled image-to-video model built on Wan2.2-I2V-A14B. đĽ Apache 2.0. đ¤https://t.co/yGlJY7Ur4nđ https://t.co/TGFU6zAczq đ 29% lower rotation error and 43% lower FVD at 14B than the strongest baseline, while leading every closed-loop revisit metric. ⥠Under 0.1% new parameters, with gains at both 5B and 14B while preserving the pretrained RoPE bit-exactly. đŹ Camera rays become positional coordinates inside attention, turning a first frame, prompt, and camera trajectory into precisely controlled, spatially consistent video. đŚ Self-contained inference weightsâno separate Wan2.2 checkpoint required.
@Alibaba_Wan's Spicy Suite is now available on Infron. Spicy models remo
@Alibaba_Wan's Spicy Suite is now available on Infron. Spicy models remove content filters for image-to-video generation. Mature creative content like boudoir, intimate scenes, and artistic nudity generates without refusal. 1. Video Super Resolution - FlashVSR upscales 480P, 720P, 1080P to 2K/4K with single step restoration. https://t.co/ektDZFCdeY 2. Video Animation & Motion Transfer - Wan Animate transfers motion from reference videos onto character images. https://t.co/ONz17no5mm - Wan 2.7 I2V Spicy transforms images into cinematic 720p/1080p clips with mature content support. https://t.co/O7abFixtSx - Wan 2.2 I2V Spicy animates images at lower cost, earlier generation variant. https://t.co/0qrEGRRETX 3. Face & Head Manipulation - Head Swap replaces heads while preserving body and scene. https://t.co/8UGKWwvAYn - Face Swap replaces faces in images seamlessly. https://t.co/JaUv37T17p 4. Image Generation & Editing - Qwen Image Edit Spicy modifies images with text commands. https://t.co/iLpeHjJ5nG - Z Image Spicy generates images from text with mature content support. https://t.co/lng9UgoSmx
@Alibaba_Wan 's Spicy Suite is now available on Infron. Spicy models rem
@Alibaba_Wan 's Spicy Suite is now available on Infron. Spicy models remove content filters for image-to-video generation. Mature creative content like boudoir, intimate scenes, and artistic nudity generates without refusal. - FlashVSR upscales 480P, 720P, 1080P to 2K/4K with single step restoration. https://t.co/ektDZFCdeY - Wan Animate transfers motion from reference videos onto character images. https://t.co/ONz17no5mm - Wan 2.7 I2V Spicy transforms images into cinematic 720p/1080p clips with mature content support. https://t.co/O7abFixtSx - Wan 2.2 I2V Spicy animates images at lower cost, earlier generation variant. https://t.co/0qrEGRRETX - Head Swap replaces heads while preserving body and scene. https://t.co/8UGKWwvAYn - Face Swap replaces faces in images seamlessly. https://t.co/JaUv37T17p - Qwen Image Edit Spicy modifies images with text commands. https://t.co/iLpeHjJ5nG Z Image Spicy generates images from text with mature content support. https://t.co/lng9UgoSmx
I tested the same product video prompt across 4 AI video models. The int
I tested the same product video prompt across 4 AI video models. The interesting part was not which one was âbest.â It was what each model seemed better at: Kling 3.0 Turbo: fast camera/motion drafts Seedance 2 Mini: cheap direction testing Wan 2.2 Fast: quick image-to-video iteration Hailuo 2.3: more realistic human/object motion If you are building a product, the question is not: which model wins? It is: which model gets you to a usable draft fastest for this specific workflow?
When snack cravings hit on a massive scale. đď¸đ Ever been so hungry youâ
When snack cravings hit on a massive scale. đď¸đ Ever been so hungry youâd step over skyscrapers just for a bag of Layâs? Follow the journey of a city-sized giant on a mission for the ultimate snack run. From urban giant to sunset park vibes, sometimes you just need to satisfy that big appetite. #GiantCravings #SnackRun #CinematicAI #ai #Lays #Pringles How it was made This was built using a foundation-first workflow in Google Flow to ensure visual consistency and high-quality scale effects: Character Anchoring: I started with a single uploaded reference photo to establish the character's appearance (black hoodie, short curly hair) across every scene. Scene Foundations: Generated high-detail "First Frame" images for each story beat. This allowed me to precisely control the forced perspective and scale illusions like the giant hand reaching for the 7-Eleven before any motion was added. AI Animation: Using the Omni Flash model, I applied Image-to-Video (I2V) generation to those frames. This brought the scenes to life with realistic physics, from the heavy stride of the giant to the subtle golden hour lighting in the park. Cinematic Direction: Each prompt was layered with specific camera angles (low-angle hero shots, interior office POVs) and lighting transitions to move the story from a bright urban day to a warm, emotional sunset.
I turned my headshot into a marble statue and forced it to dance. By usi
I turned my headshot into a marble statue and forced it to dance. By using my photo for texture transfer and Kling 3 for motion control, Iâm finally seeing real character consistency in AI video. Whatâs the most creative use for motion control youâve seen yet? #AI #KlingAI https://t.co/6X1dY2DgZn