@junwatu

AI Engineer 🤝 AI Creator | Tech Writer.

Remote
Joined July 2011
Pinned Tweet
Another AI slop for you! 😆 Made with Wan 3.0 × @Flovaai No sugarcoating, their AI Agent is actually really easy to use. Just tell it the story you want, and make sure to create references to keep everything consistent. The skill I used here: Human Interest Documentary Short Feel free to use my referral link if you want to try the platform: flova.ai/?refCode=6HE38KAA #flovacpp #flovaai @Flovaai_Japan
🤖 Made with AI
If AI slop is like this, please keep making it! Never thought that Wan 3.0 would be this good at slice-of-life scenes. FlovaAI x Wan 3.0 Limited-time offer: Wan 3.0 480P starts at $0.013/second.
12
1
21
4,827
This is only possible by using an existing video. Below is the source video, Russian Roulette by @KotteAnimation. The bigger picture here is how easy it has become with AI workflows to take an existing scene and completely reskin it while keeping the original motion and choreography. Yes, bypass all the complex prompting.
肝了一夜!Dario大战Altman! Seedance2.5的参考视频没法去掉音效,所以有点瑕疵 无论如何,Claude封了我两个号,我选Altman赢!
66
709
54
9,306
558,057
Consider helping this account by clicking on this post. Thanks! nitter.cf/junwatu/status/2100029…
Another AI slop for you! 😆 Made with Wan 3.0 × @Flovaai No sugarcoating, their AI Agent is actually really easy to use. Just tell it the story you want, and make sure to create references to keep everything consistent. The skill I used here: Human Interest Documentary Short Feel free to use my referral link if you want to try the platform: flova.ai/?refCode=6HE38KAA #flovacpp #flovaai @Flovaai_Japan
1
1,627
Ngl, this one's easy to make. nitter.cf/junwatu/status/2100604…
Creating an animation scene with P-Video-2-Pro from @PrunaAI The model supports different generation modes and prompt enhancement settings, which really help with getting better results. This one was generated using Quality mode with Max prompt enhancement. Prompt and assets ⬇️
385
Start frame generated with Hy Image 3.5 Preview, then animated with Wan 3 Prime.
🤖 Made with AI
151
Made this with p-video-2-pro and p-video-edit from @PrunaAI. The workflow is pretty simple: Generate the scene → swap the outfit with p-video-edit using a fashion reference → feed the last frame back in → repeat. And the whole process can be automated with any AI agent + the Pruna skill.
🤖 Made with AI
5
1
10
349
Hy Image 3.5 is really good for creating images like this.
3
196
Been here forever, thanks for 5K!!
13
235
Hy Image 3.5 Preview 2K, 16:9 created on @gmi_cloud > A cinematic, wide-angle shot of a group of joyful children playing soccer in a recently harvested rice field during golden hour. Low-angle perspective capturing splashes of muddy water and dust as a child kicks a worn leather ball. The field is mostly bare, covered with short golden-brown rice stubble, dry straw, patches of exposed wet soil, and shallow muddy areas left after harvesting. Warm golden sunlight filters through distant palm trees, casting long shadows across the expansive post-harvest rice fields. Rich textures of dry straw, uneven earth, and muddy ground, soft atmospheric haze, highly detailed, photorealistic, 8K resolution.
🤖 Made with AI
1
6
204
DΞV retweeted
I've got access to PixVerse R2 and honestly, I hate it when I enjoy myself this much 😂 PixVerse R2 is a real-time AI world model. If you’re into games and interactive storytelling, you’re gonna love this. Watch me explore this world and you’ll get a glimpse of how powerful this can be. @PixVerse
🤖 Made with AI
3
1
1
11
744
MiMo-V2.6-Pro is now live on GMI Cloud Xiaomi's new flagship open-weights model: - 1T+ params (42B active MoE) - 1M token context - Native omnimodal input (text, image, video, audio) - Built for long-horizon agentic work and coding Available with GMI API key and on @OpenRouter 👇
13
5
2
55
2,279
Effects are an important part of video editing, and invideo takes this a step further by bringing them into its agentic AI workflow. Kudos to the team! 👏🏼👏🏼👏🏼 @invideoOfficial
🤝 Paid partnership
Build any effect you want on your timeline by describing it in chat, or providing a reference video for it. With each effect, you also get a custom controls panel to adjust things like intensity, direction, length, mix, etc. so that the effect feels like your own. And to top it off, 90+ inbuilt effects that you can drag-drop for free.
3
184
“Bangkit” means “rise” in English. This video was created using @Flovaai agent. Prompt ⬇️ #Flovacpp #Flovaai @Flovaai_Japan
🤖 Made with AI
6
20
700
Use this prompt: Seedance 2.5 for video, 720p. 30 second cinematic action sequence, 16:9 widescreen, Unreal Engine 5, MetaHuman quality realistic CG, hyper realistic cinematic fantasy action, physically based materials, realistic porcelain skin, detailed fabric, individual hair strands, realistic water simulation, volumetric mist, dramatic storm lighting, AAA game cinematic quality, 24fps. REFERENCE ANCHORS: Use exactly two attached visual references for the entire sequence. @[Baldy] Baldy identity anchor defines Baldy's identity, face, bald head, facial structure, body proportions, muscular anatomy, normal outfit, forearm wraps, sandals, and katana. @[bankai] Bankai design anchor defines his transformed Bankai appearance, including the porcelain white skin, white torn robes, dark inner garments, luminous markings, pale eyes, and transformed silhouette. Lock both references as persistent anchors for the entire video. Do not reproduce their backgrounds, studio setup, framing, or character sheet layout. Baldy remains the same person throughout the entire sequence. His facial identity, anatomy, body proportions, and katana remain consistent while his appearance transitions from the normal state into the Bankai state. No character morphing, identity drift, face replacement, anatomy changes, extra limbs, or disappearing/reappearing objects. CHARACTER TRANSFORMATION: The transformation should feel like a simple but profound evolution of the same warrior. Normal state: dark black and gold samurai clothing, natural human skin, restrained appearance. Bankai state: smooth porcelain white skin, predominantly white torn robes, dark inner garments, subtle metallic gold details, luminous white markings, pale supernatural eyes, concentrated white energy. The transformed state should feel elegant, unnatural, powerful, and frightening without becoming monstrous. No dragon features, horns, scales, wings, armor, summoned creatures, giant accessories, or complicated mutations. The katana remains the same physical weapon throughout the entire sequence. ENVIRONMENT: A vast black ocean beneath an enormous stormy sky. Baldo stands alone ankle deep in perfectly still shallow seawater, far from civilization. The ocean stretches toward a huge dark horizon. Distant jagged black rock formations and small volcanic islands are barely visible through atmospheric mist. The water is extremely reflective and dark blue black, creating a mirror like reflection of Baldy and the storm clouds. Massive layered storm clouds cover the sky. Occasional distant lightning illuminates the horizon without becoming the primary light source. The environment feels empty, ancient, and enormous. No battlefield, ruins, buildings, army, debris field, or random objects. The simplicity of the environment keeps Baldy and the transformation visually dominant. The ocean reacts physically to his power throughout the sequence. Ripples begin subtly and become increasingly violent as the Bankai activates. Wind pushes mist and surface water horizontally. Energy, wind, spray, and water must interact with believable physical force. BANKAI ENERGY: The Bankai produces a concentrated white supernatural aura closely connected to Baldy and his katana. The energy behaves like dense white plasma combined with silk like flowing energy, electrical filaments, atmospheric distortion, and luminous vapor. It must feel dense and powerful rather than like generic magical smoke. Keep Baldy's face, body, clothing, and sword clearly visible at all times. Do not hide him behind excessive particles. Energy intensity produces corresponding physical reactions in the surrounding water and atmosphere. TIMELINE. 00:00 to 00:05, the calm before: Begin with an enormous wide cinematic shot of the black ocean. The sea is almost perfectly still. Baldy stands alone ankle deep in shallow water, extremely small against the vast ocean and storm filled sky. His reflection stretches across the dark surface. Distant black rock formations emerge through the mist. The camera slowly approaches him from behind. His dark robes move gently in the wind. The water around his feet remains almost completely still. Only distant thunder, wind, and soft waves are heard. Baldy slowly turns his head toward the horizon. He remains completely calm. 00:05 to 00:10, Bankai concentration: Cut to a low angle medium shot facing Baldy. He slowly raises his katana vertically in front of his body. His hands tighten around the hilt. His breathing becomes slower and more controlled. Perfectly circular ripples begin spreading from his feet. The wind gradually changes direction and begins flowing toward him. His dark clothing remains clearly readable. Small droplets of seawater lift from the surface and briefly hover around him. The atmospheric pressure visibly increases. Slowly push toward his face. His expression remains calm. 00:10 to 00:14, Bankai declaration: Perform an extreme cinematic push toward Baldy's face. He looks toward the horizon. The transformation begins. His natural skin gradually becomes smooth porcelain white while retaining realistic human skin detail, pores, muscles, and believable specular highlights. Luminous white markings emerge beneath the skin across his face and neck. His dark robes rapidly transform into layered, torn white fabric while dark inner garments remain visible underneath. White energy begins flowing around his shoulders and arms. The ocean reacts violently, water rises around his feet and circular waves expand outward. Baldy forcefully declares, "BANGKIT!" Immediately after the declaration, the transformation completes. His exposed skin is now porcelain white. His robes are predominantly white and torn. The luminous markings intensify. His eyes become pale and supernatural while retaining his recognizable facial identity. White energy erupts outward from his body. A massive circular pressure wave spreads across the ocean. Water is pushed away from him in every direction. The transformation happens immediately after the declaration. 00:14 to 00:19, full Bankai reveal: Reveal Baldy in a dramatic three quarter low angle shot. His Bankai form is now fully established. Porcelain white skin, torn white robes, dark inner garments, subtle metallic gold details, luminous markings, pale glowing eyes, and concentrated white energy. The white fabric streams violently behind him in the storm wind. A dense white aura coils around his body. The surrounding ocean forms a huge circular depression as water is pushed away by the pressure. The exposed seabed becomes briefly visible around his feet. Water rises into slow walls around the outer edge of the pressure circle. Baldy remains perfectly still. The camera slowly circles around him. His reflection remains visible in the disturbed water. The white energy illuminates the storm clouds and creates bright reflections across the ocean. He slowly lowers the katana into a horizontal stance. 00:19 to 00:22, Bankai charge: Baldy grips the katana with both hands. His body lowers into a powerful horizontal stance. His right foot presses firmly into the submerged ground. The physical blade remains clearly visible. Dense white energy concentrates along its edge, forming a thin supernatural extension around the blade while leaving the original steel visible. The luminous markings intensify. Water begins rotating in a huge circular pattern around him. Mist and spray are pulled toward the sword. The ocean directly in front of him becomes unnaturally flat. A long horizontal path of disturbed water stretches toward the horizon. Move the camera behind his shoulder, keeping Baldy and the distant black rock formations in the same composition. He exhales once. Everything becomes almost completely silent. 00:22 to 00:23, one Bankai slash: Baldy performs exactly one enormous horizontal sword slash from right to left. Extremely fast, clean, and decisive. No spinning. No combo. No second slash. No unnecessary movement. At the exact moment of the slash, the concentrated white energy erupts directly from the katana. A gigantic horizontal wave of compressed white energy races across the ocean. It behaves like an enormous blade of compressed supernatural pressure, not a generic magical beam. The ocean is violently displaced beneath it. Massive walls of water rise on both sides of its path. The camera briefly tracks beside the katana, then immediately follows behind the energy wave. Use intense directional motion blur only during the release. 00:23 to 00:27, ocean destruction: The energy wave travels across the ocean toward the distant black rock formations. It strikes the largest formation with overwhelming force. Do not split the rock neatly in half. The impact pulverizes the formation. Black rock fragments explode outward. Water erupts vertically around the impact. A gigantic circular shockwave expands across the ocean. Huge waves radiate from the impact point. Mist and seawater fill the atmosphere. White energy flashes through the expanding spray. The shockwave travels back toward Baldy. The entire ocean surface rises and falls in massive concentric waves. Water, rock, pressure, debris, mist, and energy must interact with convincing physical weight. Baldy remains grounded in the foreground. His white robes and fabric strips whip violently in the returning pressure wave. He does not move from his position. 00:27 to 00:30, final Bankai image: Transition into an enormous aerial shot. The black ocean is covered with massive concentric waves. The distant rock formation has been devastated into scattered black fragments surrounded by violent white surf and mist. The storm clouds begin opening slightly. Soft light breaks through and reflects across the ocean. Residual white energy slowly fades through the atmosphere. The camera rapidly descends toward Baldy. End on a powerful three quarter low angle shot. Baldy stands alone in the shallow black water. His porcelain white skin contrasts strongly against the dark ocean. His torn white robes move in the returning wind. Dark inner garments provide contrast. Luminous markings remain visible across his face, neck, and arms. His pale eyes glow intensely. His katana hangs naturally beside his body. Residual white energy coils around his shoulders and blade. The ocean slowly settles. His reflection becomes visible again. Baldy remains completely calm and in control. Hold on his face for the final moment. FINAL VISUAL RULES: Maintain the established character identity and transformation design throughout. Prioritize consistent facial identity, realistic human anatomy, porcelain white skin, torn white robes, dark inner clothing, luminous white markings, pale supernatural eyes, unchanged katana, dark reflective ocean, physically believable water and energy interaction, strong visual contrast between Baldy and the environment, elegant supernatural power rather than monstrous mutation, and clear cinematic action. Exactly one horizontal slash. The energy originates directly from the katana. No dragon features. No horns. No scales. No wings. No summoned creature. No armor. No generic magical beam. No excessive particles obscuring Baldy. No random objects. No repetitive attack loops. No random camera cuts. No subtitles. No UI. No text on screen. AUDIO DESIGN: No background music. Begin with distant ocean waves, deep wind, distant thunder, subtle water movement, and controlled breathing. As concentration begins, introduce a low frequency atmospheric pressure tone and subtle metallic katana resonance. During transformation, add deep sub bass pressure, rising energy hum, electrical crackling, fabric movement, water displacement, and violent atmospheric wind. Baldy's declaration, "BANGKIT!", must be clear, powerful, and isolated. Immediately afterward, the sound explodes into a dense white energy surge, massive water displacement, and expanding pressure wave. During the slash, use one extremely sharp metallic sword release followed by a massive compressed air explosion and deep energy impact. The destruction contains realistic water impacts, rock fragmentation, crashing waves, distant thunder, and expanding shockwaves. End with falling water, distant rock impacts, fading white energy, gentle wind, and the ocean gradually returning to a calmer state.
458
My experience with Agent Two has been pretty good so far, especially now that invideo has added powerful video editing features and agentic sound design. I’d say the 50% off is a rare chance to try the whole workflow in one place and see what you can actually create with it. @invideoOfficial
🤝 Paid partnership
Limited time offer: you now get up to 50% off on purchase of any invideo annual plan. If you're a serious creator or editor, this one's for you. Go crazy.
5
266
DeepSeek V4.1 Flash 25% on GMI Cloud 7 days, unlimited use it via GMI API key on openrouter, open code, cline, hermes agent, openclaw, and more
5
6
3
57
2,908
DΞV retweeted
Today, we open-source Pruna-Qwen-Image-2.1, a set of a few-step LoRA adapters that make Qwen-Image-2.1 by @Alibaba_Qwen up to 6.3× faster for image generation and editing. - 𝗛𝗼𝘄 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝗶𝘀 𝗶𝘁? Generate or edit images in just 5 or 8 steps instead of 40. Choose 5 steps for maximum speed or 8 steps for the best balance. - 𝗛𝗼𝘄 𝗴𝗼𝗼𝗱 𝗶𝘀 𝗶𝘁? The 8-step adapter is our recommended default. It supports text-to-image and single- or multi-image editing at 1K resolution, with up to three reference images. - 𝗛𝗼𝘄 𝗱𝗼𝗲𝘀 𝗶𝘁 𝘄𝗼𝗿𝗸?The LoRA adapters load directly on top of Qwen-Image-2.1, while the pipeline remains unchanged. Each adapter uses its own optimized sigma schedule and runs without CFG. 𝘛𝘩𝘪𝘴 𝘪𝘴 𝘫𝘶𝘴𝘵 𝘰𝘶𝘳 𝘧𝘪𝘳𝘴𝘵 𝘳𝘦𝘭𝘦𝘢𝘴𝘦, 𝘸𝘪𝘵𝘩 𝘦𝘷𝘦𝘯 𝘣𝘦𝘵𝘵𝘦𝘳 𝘷𝘦𝘳𝘴𝘪𝘰𝘯𝘴 𝘰𝘯 𝘵𝘩𝘦 𝘸𝘢𝘺. 🚀 Want to try the fast open-source integration? Check it here on Diffusers: buff.ly/1CpiRPp Want to try the fastest image generation and editing endpoint? Check P-Image-Ideogram and P-Image-Edit here on API: buff.ly/gw5tvUV
31
103
11
944
85,955
DΞV retweeted
Comfy Router is live One API for frontier image, video, 3D, and audio models. Same model string. Same arguments. No new SDK, no new key, no redeploy. What Comfy Router gives you: → Explicit routing. You name the provider, we call that provider. It's down? The request fails there. No silent fallback. → Every job returns the provider that ran it. Log it, bill it, debug it. → Async. submit() returns a request ID immediately. The queue retries 429s and transient errors until a slot opens. subscribe() submits and polls to completion. → Batch-friendly. Queue a few hundred jobs, hold the IDs, pull results as they land. Nothing blocking on a 5-min video render. → 24h retention on inputs and outputs, then deleted. → Comfy credits. No sub, no Router fee. Providers at launch: Comfy. Runware, Wavespeed, Fal, Higgsfield. Multi-provider where the model supports it. Get Your API Key with the link below. ⬇️
110
96
115
2,492
7,423,539
Introducing FLUX 3 Action. An open weights 7B World Action Model that achieves first place on the RoboLab benchmark. It outperforms the previous best open model by 6.1 percentage points while using 56% fewer parameters and running up to 3.95x faster.⁠⁠ FLUX 3 Action removes the usual trade-off between world action model performance and VLA speed: it still predicts video and actions together, but plans more than twice as far ahead and runs faster per second of robot motion than the strongest open VLA. Teams can fine-tune FLUX 3 Action on their own demonstrations to create policies for a particular robot and task. Together with @nvidia, we also integrated FLUX 3 Action natively into @huggingface's LeRobot, with fine-tuning recipes included and edge deployment on NVIDIA Jetson. Beyond robotics, we’re also seeing promising results training task-specific policies for acting in simulated environments like gaming, controlling a vehicle, computer use, and wherever else a model needs to understand a visual environment and then choose what to do next. FLUX 3 Action builds on the same image, video, and audio pretraining as FLUX 3, but uses a smaller architecture designed for practical deployment. In midtraining, we trained the model to predict actions and future frames together. We’re releasing the weights, code, fine-tuning recipe, benchmarks, and reproducible examples so researchers and developers can build on the model with their own robots, environments, and tasks (see below).
77
234
52
1,792
166,524
I never disclosed this, but the video I posted back in July 2026 is actually a scene from The Raid. Yes, this was done before any “genjutsu” or whatever you want to call it. 😄 The point was simply to show what AI can already do with an existing scene. Disclaimer: The original footage belongs to its respective copyright holders. This post is shared for educational and technical demonstration purposes.
🤖 Made with AI
Can't wait for the release of Seedance 2.5 to put my fight-scene workflow to the test. Until then, I'm really happy with how this one turned out. ⬇️
4
10
513
Even the best AI video models still struggle with audio quality, which is why most of my creations need manual sound design. Glad to see Invideo introducing Agentic Sound Design directly into their editor. @invideoOfficial
🤝 Paid partnership
The era of manually building sound beds is ending. Launching: Agentic Sound Design on the invideo editor. The online invideo editor + editing agents can access all the music and sound in the world, or generate it from scratch, auto-place it on the timeline, mix + master, and build a entire sound bed that adheres to your vision and direction.
1
4
386