Patrick

Cinematic Exploration #2 This time, I wanted to explore a medieval world. I called this short sequence “MEDIEVAL” A small story about two friends, with one trying his best to handle the other, a drunk troublemaker, until the situation spirals out of control way too fast. Below, you can read the synchronized script used for the sequence. I asked Codex to create a clean sync between the written script and what appears on screen, and it handled it surprisingly well. P.S. This was not made with Seedance 2.5 😅

@lena_z01 @BytePlusGlobal Can I get it please?

Stop wasting credits testing Seedance 2.5. @BytePlusGlobal just released the official prompt guide, and it reveals how to get much better results from the model. Highlights: → Extend videos with continuous creation → Generate long-form videos up to 180s → Control timing with text prompts → Remove unwanted subtitles & BGM → Improve reference consistency with multimodal inputs Comment below and I’ll send you the full guide 👇

@steftranquillin Hard to pick a winner here 😅 But I'd probably give it to Flux 3. It's the most consistent overall both in terms of acting and voice

We're so spoiled with these models releases ! I wanted to do a quick comparison between the three latest models on an old french movie scene, with the same prompt and the people talking in french. 1- Minimax H3 2- Seedance 2.5 3- Flux 3 Quite happy with the result !

@brumaombra @dreamina_ai looks real right?

@patrickassale @dreamina_ai Wow 😳

Same prompt with Seedance 2.5 @dreamina_ai #dreminacpp

@steftranquillin @dreamina_ai Super, hâte de voir ça 👀 et je viens justement de partager le prompt de celui ci en commentaire comme promis

@patrickassale @dreamina_ai Super merci pour le tip. Dans Seedance 2 même en écrivant il faisait des fautes de français. Je vais poster un test là de français entre Seedance, Flux et Hailuo. Ça a bien progressé le français heureusement !

Un autre test en français avec Seedance 2.5, cette fois dans un style moins cinématique. Je partage le prompt en commentaire sous peu. @dreamina_ai #dreaminaCPP

T2V prompt: A 30 second ultra realistic handheld POV video set in France at around 3 am, outside a busy bar and nightlife street. The camera is the point of view of a young woman filming one handed on an iPhone. We never see the phone itself at any point. The footage must feel like raw, authentic amateur nightlife footage with natural iPhone aesthetics. It is very late. People are leaving bars, smoking outside, waiting for Ubers, talking too loudly, eating kebabs, laughing in tired, sloppy groups. The mood in the street is a mix of drunken energy and visible exhaustion. Faces look tired, makeup is slightly worn, outfits are a bit messy after a long night. The pavement is slightly wet and reflects neon signs, headlights, and warm bar lights. The air feels cold and heavy. The whole scene must feel intensely real, socially awkward, and uncomfortable. The camera operator is a woman, and her friend is another young woman walking a few steps ahead of her, visibly angry. The camera follows her in a shaky, one-handed handheld way. The tension is already there from the first second. The argument is about the friend disappearing with a guy inside the bar and leaving the camera operator alone for nearly an hour, which made her miss the last train and feel abandoned. This is the core of the argument. The friend suddenly turns around while still walking and says in natural French from France: “Tu vas me suivre encore longtemps comme ça ?” The woman behind the camera replies, irritated and hurt: “Bah ouais, parce que tu m’as laissée toute seule comme une conne pendant une heure.” The friend rolls her eyes, tired and defensive: “Oh ça va, j’étais pas ‘avec un mec’, on parlait, c’est bon.” The camera gets a little closer as the woman filming answers immediately: “Mais arrête, tu m’as complètement lâchée. J’te cherchais partout.” The friend stops walking and turns fully toward the camera now, angry but exhausted: “J’allais pas te tenir la main toute la soirée non plus.” A few people nearby start glancing at them. Someone exits a bar laughing, a smoker looks over, a taxi passes in the background. The world keeps moving around them, but the argument becomes the emotional center of the scene. The woman filming replies, more intense now: “C’est pas ça le problème, le problème c’est que t’as disparu sans rien dire. J’ai raté le dernier train à cause de toi.” The friend laughs bitterly in disbelief: “À cause de moi ? Sérieux ? T’es incapable de te débrouiller toute seule ?” The camera jolts slightly from the emotional reaction. The breathing of the woman filming becomes more noticeable. Her voice gets sharper: “Mais t’abuses en fait. Tu fais toujours ça. Tu penses qu’à toi.” The friend steps closer, now visibly losing patience: “Et toi tu fais quoi là ? Tu me filmes en pleine rue comme une malade.” The woman behind the camera replies immediately: “Parce que tu racontes n’importe quoi !” The friend reaches toward the camera, not to grab a visible phone, but clearly trying to push it away from her face. The image jerks sideways hard from the contact. “Enlève ça de ma tête !” The camera operator snaps back: “Me touche pas !” Now the tension breaks completely. The friend, furious and emotionally overwhelmed, shouts: “Mais ferme-la un peu !” The woman filming fires back: “Sinon quoi ?!” There is a half-second pause. Heavy breathing. Eye contact directly into the lens. Then the friend lunges. She pushes into the camera operator. The footage becomes chaotic and violent, shaking hard as the woman filming loses balance. She lets out a shocked cry: “Putain !” The camera stumbles backward and drops toward the ground. As it falls, we catch blurred fragments: legs, jackets, streetlights, pavement, people turning around, someone shouting: “Eh oh ! Ça va pas ou quoi ?!” The friend jumps onto her as both fall out of balance. The camera lands badly on the ground, still recording for a split second, partly blocked by clothing or the ground. We hear struggling, overlapping voices, panic, and nearby people reacting. Then the video abruptly cuts. The entire performance must feel spontaneous, messy, and completely believable. Natural interruptions, overlapping dialogue, breath, hesitation, frustration, and tiny realistic body movements are essential. The acting must feel like a real late night argument between two exhausted friends, not a scripted drama. No cinematic polish, no stylized fight choreography, no slow motion, no music, no visible phone, no artificial lighting setup, and no glamorous performance. It must feel like a real one handed iPhone recording captured in the heat of the moment.

@Douzyycuus @dreamina_ai Ah oui, ce pnj 😂

@patrickassale @dreamina_ai Le sandwich doit être vraiment délicieux hein

@BlackEyedHatMan @dreamina_ai 780 credits for this one

@patrickassale @dreamina_ai How many credits did she take?

@steftranquillin @dreamina_ai Oui chaque dialogue est écrit. Quand tu lui laisses de la liberté, il n’est même pas capable d’aligner une phrase correctement sinon 😂

@patrickassale @dreamina_ai Super ! Est-ce que tu écris chaque dialogue? Parce que pour l'instant dans mes tests, je donne seulement le contexte de la conversation et Seedance 2.5 n'est pas terrible en français comparé à Hailuo H3 et Flux 3

@brownigweco @dreamina_ai It feels more realistic across every format, with better dialogue and much more flexibility in terms of duration, especially with the new 30 second one output generation from a single prompt. But, It is a bit more expensive to generate videos now though

@patrickassale @dreamina_ai Seedance 2.5 is on a whole new level aside from the Speed and good color grade what is have you observed in this new upgrade?

Motion design test with Seedance 2.5 @dreamina_ai #DreaminaCPP

@lansenai @dreamina_ai Thank you

@patrickassale @dreamina_ai 🔥场景设计和人物设计很棒

Seedance 2.5 at full power. 30 seconds of pure intensity. @dreamina_ai #dreaminaCPP

@second_bassoon @dreamina_ai Yesss it looks even more realistic now

@patrickassale @dreamina_ai Massive jump in quality!

@AiMeowing @dreamina_ai thank you Nisho

@patrickassale @dreamina_ai wild, nice shot btw

NEW WORK 🔥 FULLY AI-GENERATED 4K MUSIC VIDEO Full video link in the comments 👇🏾 Created in collaboration with @Dreamina_ai, using Seedance 2.0, shortly before the release of Seedance 2.5 With “Ma Story,” @SpycoJ and I wanted to show what becomes possible when an independent artist keeps full control of his vision and uses AI as a real production tool. We developed the concept together, then used DreaminaAI to turn that vision into a fully AI-generated 4K music video. Questions about the process? Ask me in the comments. #dreaminacpp #seedance2 #DreaminaAIMV #aimusicvideo #aifilmmaking

@Martinolli Merci Christophe

@patrickassale Propre, vraiment.

Deuxième test de Seedance 2.5 en français

@Cryptws Merci 🙏🏾

@patrickassale oh là c'est super biennnnnnnnnnnnnnnnnnnnn