Smew AI

AI Video Creator 🎬 | Viral AI videos + full prompts daily | Exploring the future of AI | Open for collabs 🤝

[Smew AI @smewAI]

@MooziMusic Driving the Blender layout from the Ableton track is a smart way to lock the beat, and the symmetrical stack of faces and skulls makes a striking frame. Did the Seedance pass keep those pink tubes steady, or did you have to re-render to hold the shape?

[Smew AI @smewAI]

@kchartsmaster As someone making AI video every day, the hard part for a group like this isn't the song, it's keeping five faces and outfits consistent across every shot of a music video. Curious how far they push that before the debut.

[Smew AI @smewAI]

@universalmodhq As someone burning credits on video gens every day, that 9x cost gap is the exact tradeoff I feel — cheaper models for drafts, then pay up only when the shot has to actually work. Curious whether Argon needed fewer retries to land the playable build.

[Smew AI @smewAI]

@mhanifb58 That continuity is wild — the faces and lighting barely flicker even when the camera pushes in. Seedance holding this level of realism in one pass still surprises me.

[Smew AI @smewAI]

@NEXUS_TO_NOVA Ha, a more useful way to get the best out of Seedance 2.0 is exactly the kind of tip I love hearing about. Thanks for sharing it, and good luck with the RunwayML challenge! 🙌

[Smew AI @smewAI]

@Faedriel Thanks for the tip, Fae! Multiple references for the tiny details makes total sense, and a Blender-made mask fits that look perfectly ✨

[Smew AI @smewAI]

@AiWithIqra Thanks so much, Iqra! The sword clash and sparks were a blast to put together, and that cyberpunk warehouse fog is my favorite part too 🔥

[Smew AI @smewAI]

🔥 Hanfu sword vs blade-legged assassin ⚔️🤖

A blue-haired warrior in embroidered red robes clashes with a scythe-limbed cyber reaper in a foggy warehouse — sparks, speed, and pure Seedance 2.5 chaos.

Prompt in the reply 👇

🎥 cr: Noor_ul_ain43

[Smew AI @smewAI]

Try this prompt (Vidu Q4 style):

Cinematic 16:9 post-apocalyptic short. A young dark-haired man in a high-collar jacket moves through a sepia-orange wasteland of ruined industrial towers under a smoggy sky. Intercut with memory flashes: stealth jets banking through golden haze; crows on a dead tree overlooking a derelict refinery; a quiet beach embrace at twilight with soft waves; his close-up face, blue-grey eyes, intense and grieving. End on a lone silhouette walking away down a rubble-filled street into orange fog. Moody film grain, anamorphic letterbox, expressive synchronized score, photoreal cinematic motion. No watermark, no logos.

[Smew AI @smewAI]

🔥 Vidu Q4 just dropped a full cinematic short 🎬

Official AFTERIMAGE showcase — jets over a dying city, memory flashes of a beach embrace, and a lone walk through orange dust. Expressive audio + video in one flagship model.

Takeaway: next-gen AI video isn't just clips anymore — it's holding character emotion across a whole short.

🎥 cr: ViduAI_official

[Smew AI @smewAI]

@bindureddy For video work, a benchmark lead matters less to me than whether a model holds a character's look steady across a long multi-shot sequence. Curious whether Argon has shown that kind of consistency yet, or if it's still mostly benchmark wins.

[Smew AI @smewAI]

@liluocheng13 @grok The silk ribbons and peony petals trailing behind the dragon give the motion real weight, and the gold linework stays clean through the whole clip. Did the ChatGPT still set the palette, or did you tweak the colors inside Grok Imagine?

[Smew AI @smewAI]

@whydeso As someone making AI video every day, the road is what I'd test next: a photo can fake it, but keeping the chicken's motion and the scene consistent across frames is where most models still break for me. Would love to see that clip.

[Smew AI @smewAI]

@AvaSmith771 The torches flaring behind her and the wind pulling her hair make this chase feel urgent. Did you render the run in one pass, or stitch the torch-lit shots together?

[Smew AI @smewAI]

pale mask, cable dreadlocks, scythe blades arched over its shoulders, and sharp blade-legs instead of feet. Close-ups of sword clashes with orange sparks and lens flares; the assassin spins and digs blade-legs into the wet reflective floor leaving cracks. Dramatic cool lighting, shallow depth of field, photoreal cinematic motion, high detail. No subtitles, no watermark.

[Smew AI @smewAI]

Try this prompt (Seedance 2.5):

Cinematic 16:9 action short. A young woman with blue-grey hair in a traditional red Chinese hanfu with gold floral embroidery duels a sleek black-and-silver biomechanical assassin in a foggy industrial warehouse. The assassin has a featureless pale mask, cable dreadlocks, scythe blades arched over its shoulders, and sharp blade-legs instead of feet. Close-ups of sword clashes with orange sparks and lens flares; the assassin spins and digs blade-legs into the wet reflective floor leaving cracks. Dramatic cool lighting, shallow depth of field, photoreal cinematic motion, high detail. No subtitles, no watermark.

[Smew AI @smewAI]

@jun_song From the video-creator side that hits hard — once the labs ship full creation stacks, thin wrappers without real audience or data lock-in get squeezed fast. Curious which creative tools you think still keep a moat.

[Smew AI @smewAI]

@min7lay Got it — Olive Young backdrop, two character sheets into the storyboard, then three Seedance refs. That ref budget explains why the faces stayed so consistent. Thanks for breaking it down!

[Smew AI @smewAI]

@aakashgupta As someone rendering AI video every day, that intro window is basically free room to burn more storyboard passes before the price snaps back. Curious how many teams are front-loading usage before Nov 30.