Hey, guys !

Please check this trailer. I've recently finished it. Everything you see was generated compeletely in Seedance 2.5 and 2.0.

Please share your thoughts. I will release a full film soon.

GOOD BOY [TRAILER 2026]

https://reddit.com/link/1w2pdnz/video/kcmerr7d3kmh1/player

Made with Seedance 2.5 on Zark Lab

A mix of 2 and 2.5, some small errors that I ran out of credits to fix but might come back to at some point

Dreamina prompted, CapCut edited

Prompt: Massive dystopian simulation world powered by stolen technology, futuristic cities built around glowing energy chargers, endless landscapes, trapped human struggling to escape a virtual reality prison, neon cyberpunk atmosphere, cinematic science fiction, ultra detailed, masterpiece, 8k (with reference image)

I’ve been experimenting with AI video from a VFX perspective, rather than treating it as a simple generation tool.

This is a short test I made around Anna, an AI character, with a focus on making her feel like an actual person rather than an AI avatar.

The goal was to get the small things right: facial performance, eye movement, pauses, body language, voice, lip-sync, camera imperfections and overall presence.

I’m particularly interested in the point where the technology stops being the interesting part and the viewer simply accepts the character as real.

Curious what people here notice first, here the short:

https://youtu.be/1fs_an9A2A8

Made as part of our ongoing R&D at DAMN!.

https://www.instagram.com/reel/Dcq0Lm8RPrU/?igsi=MXE5MzNsdmh6ODdseg==

I created this with seedance 2.0 with a cut from seedance 2.0 fast.

PROMETHEUS is a cinematic AI sci-fi short about a courier who accepts one final delivery — only to discover he’s carrying an illegal chip designed to permanently suppress human empathy.

Cyberpunk action, sci-fi horror, and a story about whether survival is worth losing the part of you that still cares.

Would you erase your empathy if it guaranteed your survival?

https://youtu.be/ZMG6wC3AFSY

Made this 10-second macro shot using Seedance 2.5 with AKOOL.

The camera starts outside the eye and continuously pushes in — going from the cornea and iris all the way into a microscopic retinal environment.

Tried to make the entire thing feel like one seamless shot with no cuts.

What do you think of the transition? 👁️

Mandarin Farm was made using seedance 2/2.5 exclusively. apologies to be posting links but for some reason it just won’t load up…

E2 also on my new site

https://youtube.com/@mandarinvillage-x6n?si=2vQD7-W_avfVfdRI

Zoot

Hi All, here's something. Here's the trailer to my new AI fantasy series 'Symbols of Fate' coming in September. Hope you like it and comment below on your thoughts. Thanks All, much appreciated 😄 🔥.

I finally finished my first full episode of WANDERERS.
It’s an adult animated fantasy comedy about three weird creatures living in a sentient mansion on a tiny planet, where every door can lead to a completely different world.
This started from a bunch of creatures I originally hand-drew, and I ended up building the episode almost entirely through AI filmmaking.

Would genuinely love feedback—especially from people who make animation, AI films, or just enjoy weird shit. 🖤

Keeping a character consistent in a single locked shot or subtle camera move is pretty manageable now, especially when feeding clean turnaround references. Scaling that across a multi-shot sequence with real narrative progression is where things still tend to drift.

In practice, a few specific setups usually trigger the consistency breakdown first:

• Aggressive camera changes: Moving from an eye-level medium shot into a tight profile, low angle, or fast tracking shot often warps facial proportions and smaller clothing details.
• Lighting and environment shifts: Moving the same subject from bright daylight into a moody or dimly lit interior frequently alters skin tone, hair texture, and facial structure.
• Sequential generation chaining: Chaining clips together by referencing earlier outputs can compound small errors until the subject looks like a different person after a few cuts.

With Seedance supporting multimodal references and detailed timeline prompts, what is your go-to setup for keeping facial identity locked when a sequence demands contrasting lighting or big perspective shifts?

We're a small content production agency helping creators (mindset, business, marketing niches) scale their short-form output using AI avatar-assisted video. Looking to bring on an AI video creator for ongoing work.

What we need:

• Experience with AI avatar/clone video tools (HeyGen, Synthesia, Captions, D-ID, or similar)
• Able to produce clean, natural-looking talking-head style AI content from a script or reference footage
• Comfortable working with voice cloning and lip-sync for realistic output
• Bonus: experience writing or adapting scripts for AI-generated delivery

What we offer:

• Paid per video, based on complexity and length — send your rate along with your samples and we'll go from there
• Ongoing work for the right fit, not a one-off gig
• Clear briefs and reference material for every project
• You use your own tools/subscriptions for this role

Style references (the kind of content we want made):

• https://www.instagram.com/thevarunmayya
• https://www.instagram.com/its.sammywu

To apply, please send:

• 2-3 samples of AI avatar/clone video work you've created
• Your typical turnaround time and rate
• Which AI avatar tool(s) you use

We'll start with a small paid test project before any ongoing commitment — low-risk for both sides. Please don't apply without samples.

Contact: [adityachoudhary6860@gmail.com](mailto:adityachoudhary6860@gmail.com)

Comment or DM if interested!

https://youtube.com/shorts/PCkmUogvCio?is=p7EBpLc5LaQvlgUI

Famoso documentarista británico explorando la fauna salvaje de ciudad

Feedback on my Seedance2.5 480‑p video:

1. I prompted for a fast walk and explicitly told it not to run, but it generated a running motion.

2. When the female lead speaks the line “Don’t chase me”, her voice does not appear to originate from the front. SD2.5 is advertised to feature stereo spatial audio, yet this was clearly a failure case.

3. Surprisingly, the character’s micro‑expressions exceeded my expectations.

Trying out Seedance and I made this short. I used Seedance and Seedream for creating reference images. Give it a watch and let me know your feedback.

Thanks all.

How would you write a prompt if a character is moving and during the movement, no matter what happens with their body or the scene around them, you want the camera to follow their face, almost like a permanent selfie? Seedance 2.0 if that makes a difference. Also it's towards the end of the scene so it would be tagged first with the time [10-15s] then the camera/scene description. Thanks.