Amir Kerr

Here is my entry for the #ProjectOdyssey Turn the sound on please Tools used: @midjourney @Hailuo_AI @Kling_ai @Viduforhuman @udiomusic and special thanks to @hkchengrex for MMAudio.

@DrDreamsMusic Thank you, mate! ✨

@AmirKerr This is amazing. Blown away

Sharing a new set of Wan 2.1 tests with nature forces. The tests showing how mask gives an impact to the trajectory, velocity of the "subject". No need to guide the mask till the very end of the scene, if you just want to "subject" appear and be guided to the needed way. Meanwhile, I continue experimenting with Wan 2.1

@MeanOrangeCat @AIWarper Love your series! 😼

@AmirKerr @AIWarper Really helpful to see, thanks for sharing your process!

Any Kling 3.0 multi shot experts reading this? Curious of best practices and are you just using start and end frame only? Any using “elements” or “reference images”?

@AIWarper Yes, as a first frame.

@AmirKerr You’re feeding this as the first frame? I feel like I tried this on release and it didn’t quite adhere how I wanted

@AIWarper And, of course, I use Custom Multi-shot to describe every shot.

@FugazyFilmsAI @Framer_X Huh...I see...well, at least, they ask if this is YOUR original character, not blocking blatantly.

@AmirKerr @Framer_X Welp, you’d think. I know Utopai’s tool was warning me about Dark Doctor Strange over and over haha.

Seedance 2.0 situation is so absurd. Imagine if Sora 3 or Veo 4 dropped right now and killed all the SD2 hype. Would go from biggest launch of the year → biggest flop overnight.

@FugazyFilmsAI @Framer_X Shang Tsung met Dr. Strange 💪 I mean…with all the VLM CV capabilities - I think it’s possible to differentiate between your original character and Dr. Strange.

That second one is where it gets a little hairy though, for example, I have an original IP character, Orasmas, but admittedly (and unintentionally) he looks a lot like “Dark Doctor Strange”. I wouldn’t get past a censor doing a cursory check, even though my character is not 1-to-1 wears different clothing and different facial features and hair style. He’s still too close.

@FugazyFilmsAI @Framer_X There are a lot of sophisticated ways for IP violation moderation. Ban all the keywords, that trigger licensed IPs. Double checking - first one checks the prompt, Second one checks the visuals for possible IP infringement.

Personally, I think this is an easy solve for them. Whenever they detect something that could be IP infringement- a pop up comes up and says, “You confirm you have the appropriate rights and permissions to generate this image and assume all responsibility if copyright is being infringed.” It puts it on the user to confirm they have the rights or they assume the legal responsibility and absolves the company of any wrongdoing. Simple. Quaint. Point the Disney lawyers at the people actually doing the dirty deeds without killing the functionality of the model.

#ViduClaw is now live. Meet your first AI creative teammate, ready to work 24/7. 🎁 We’re releasing 100 exclusive ClawCodes for early access. Claim yours to get started. How to get started: Visit Vidu.com/vidu-claw Click “Apply for ViduClaw” Enter your ClawCode and you're in. Setup takes just 1–3 minutes 💡 Why ViduClaw? • Always on. From ideas to execution, just say the word • Equipped with tens of thousands of capabilities, from copywriting to video creation • No complex workflows. Start with “Hi ViduClaw” and it handles the rest 🚀 Built to remove friction so you can just create. ‼️ First 100 users get free access. Don’t miss it. Also launching today: Vidu Skills Plug into your own creative extensions and go further: Github：GitHub - Saerdna/vidu-skills: vidu skillsClawhub：clawhub.ai/calvinzhao/vid… Follow #Vidu, comment below to get your ClawCode. First come, first served！ #ViduClaw #ViduAI

@higgsfield It's better be SeeDance 2!

We're about to drop something game-changing today... 🤫 Can you guess what it is? 👀

@ViduAI_official OMG! This feature will empower the whole process! ✨

A mysterious creative Partner just landed at Vidu. Reveal soon.🤩🤩

@yapper_so x.com/amirkerr/statu…

What have you made this week? 🚨We're giving away $100 of @yapper_so credits to three accounts who share their AI videos below👇 Winners will be announced in 24hr!

@JSFILMZ0412 In fact, we don’t even need that amount of denoising/enhancing for existing games. Just a slight refining would be enough. But for new IPs - yes. What if we use organoids or players themselves will be a source of neural computing 🤔 Guess I need to make a short about it.

How long till Neural Rendering replace traditional video game rendering? Technically DLSS is the first step. Seedance 2.0 Red Dead Inspired all text to video

No polishing at all. Completely raw output. • No frame interpolation • No deflicker • No color grading • No upscaling Also no heavy supervision. I didn’t curate every shot.

Been seeing lots of transformation videos made with SeeDance 2.0, so I ran my own test. The setup was intentionally simple 👇

Minimal inputs: • Only 3 reference images (front view of each form) • 2–3 generated clips per 15 seconds of sequence • Simple cut + montage

I asked an LLM one thing: “Describe the transformation process of this rat creature from [ref1] to [ref2].” Then I copied the generated shot sequence directly into SeeDance 2 and started generating.

@JSFILMZ0412 @drzubi01 @yapper_so Most likely, it will be the same situation as with TikTok. Chinese have their own TikTok for internal consumption, where they educate people, instead of stealing the time. So nerfed SD2 for the rest, and powerful SD2 for internal market. Do those 50 likes worth it?

@drzubi01 @yapper_so STOP posting copyrighted material its the reason why it got nerfed bro if you cared about the platform you will stop

Seedance 2.0 on @yapper_so just released a cinematic short that looks like it came straight from a major studio. And this is only the beginning. The upgrades from here are going to be insane.

@chrisfirst Would love to try!

Who wants early access to a new AI tool from a major AI company?👀 Drop a comment and I’ll add you to the list.

@freepik Do you have an option of a Video as input for lip-syncing, not image?

Lip sync can be hard. We fixed that 💥 Introducing Speak on Freepik Upload a visual, add your own audio or a script, and get a lip-synced talking video in seconds → Custom voices in 30+ languages → Up to 5 min talking video → A lip sync tool, made simple Available now