AI Creative Director - Campaigns, Content & Custom Builds
SEEDANCE 2.5 | 30-SECOND ONE-SHOT CINEMATIC TEST
So I wanted to see how far I could push Seedance 2.5 with one continuous 30-second camera movement.
The challenge was simple:
Take one complex storyline, keep the same heroine, maintain her wardrobe and appearance, move continuously through the environment, introduce a gravity anomaly, bring in the crater, transfer the power to the heroine, reveal the creature, and end with incoming meteors.
All inside one generation.
I pushed it.
The result was surprisingly cinematic.
The character consistency held throughout the sequence. Her outfit stayed intact. The continuous tracking movement worked. The gravity effect came through. The crater interaction worked. The power-transfer moment worked. The monster appeared. The incoming meteors created the final cliffhanger.
The biggest issue was pacing.
I gave the model too many major story beats for 30 seconds, so it compressed several actions together. The pool transition is where the sequence loses some continuity. Instead of giving that moment enough time to develop, the model rushed her into the pool and back toward the crater.
Still, I decided to keep the test.
Why?
Because this was never intended to be the final version.
I wanted to understand how Seedance 2.5 interprets a complex one-shot prompt, and this test gave me a clear answer.
The model followed the overall direction.
The next step is better scene planning.
Instead of forcing an entire story into one 30-second generation, I would break the major action beats into shorter sequences and give each moment enough room to breathe.
My biggest takeaway:
More story inside a single generation does not automatically create a better video.
Sometimes better AI video direction means giving the model less to execute at once.
This was a test.
This was the result.
And I’m keeping it because the experiment itself was worth documenting.
Tool tested: Seedance 2.5 @Renoise
Format: 30-second continuous one-shot
Goal: Character consistency + camera continuity + complex action
Result: Strong visual execution, rushed narrative pacing
Next experiment: Shorter controlled sequences with fewer major actions per generation
2
5
344
So I wanted to see how far I could push a simple idea inside Renoise.
The story starts with my AI character on an ordinary train ride. A football match catches her attention, the train makes its final stop, and suddenly she's running onto a live stadium pitch to take the shot herself. I loved playing with that unexpected transition between two completely different worlds.
To build it, I used Face Pass to keep character consistent, added an outfit reference for continuity, then used Renoise Canvas to generate and connect the images that become the entire scene. After a few refinements, the AI created the visual shots, I focused on directing the story and editing the final sequence.
I'm also sharing a sped-up Canvas walkthrough so you can see exactly how I built it.
#RenoiseChallenge
2
4
504
Title: First Memory
What would your first memory feel like if you had just been created?
First Memory is a short cinematic film clip about an AI-generated character experiencing her very first memory and slowly realizing that her world is being generated in real time.
What begins as a bright, dreamlike lifestyle film gradually transforms into a story about identity, memory, and existence. As subtle visual glitches emerge, the character's perfect memories begin to unravel, leading her toward an unsettling self-awareness.
I started with one of my previously created AI character images and imported it into Renoise Canvas, where I built the film scene by scene. Using Renoise's video generation tools and FaceLock for character consistency, I was able to maintain her identity while intentionally evolving her environment and emotional journey.
My goal was to explore a simple but compelling question:
If an AI could remember the moment it first existed, what kind of story would it tell?
#RenoiseChallenge
AI didn’t make me creative. It made me faster. Craft wins.
$5k+
Earned
3x
Hired
25
Followers
AI didn’t make me creative. It made me faster. Craft wins.
Satirical B-Roll News Report (AI- Generated)
I am a storyteller at heart with a passion for turning ideas into narratives, leveraging generative AI as an alternative to traditional pipelines with fast delivery and uncompromising production values. I create short and long form content, can work within your budget and turn projects around in as little as 24 hours.
1
289
Satirical B-Roll News Report (AI- Generated)
I am a storyteller at heart with a passion for turning ideas into narratives, leveraging generative AI as an alternative to traditional pipelines with fast delivery and uncompromising production values. I create short and long form content, can work within your budget and turn projects around in as little as 24 hours. This recent project challenged me to push the limits of authentic sounding voice design using elevenlabs.
1
302
I used Runway AI to produce this about a year ago and like everyone else, I'm so stoked about Seedance 2 that I may need to redo it! AI video is evolving so rapidly by the day when it comes to voiceovers, character consistency and adherence to natural language prompting. Until its release, I'm getting great results with Kling 3, Veo 3.1 and with LTX 2.3 as a local solution. It's an exciting time to be alive!😊
4
520
This project reimagines “Don’t Worry Be Happy” as a modern country anthem — fully developed using AI across music, visuals, and performance.
The goal was not novelty, but cohesion, maintaining consistent character design, controlled cinematography, narrative continuity, and emotional tone across every frame.
The production combined:
• AI music composition and vocal performance
• Script development and narrative structure
• Consistent character modeling and styling
• AI-driven cinematography with controlled lighting and lens logic
• Lip-sync precision and performance continuity
• Structured visual storytelling across sequences
Mental Health Ghostwriter | Content for the 2 AM search
15
Followers
Mental Health Ghostwriter | Content for the 2 AM search
Took my @HeyGen certification video and translated it into Spanish—same voice, same delivery, authentic lip-sync. Not subtitles. Not voiceover. Full AI translation that preserves the original performance while making the content native to Spanish-speaking audiences.
This is what HeyGen's translation tech enables: voice cloning across languages, perfectly synced mouth movements, and culturally adapted delivery. "Meet HeyGen: Revolutionizing Video Creation for Digital Marketers - Spanish" is now accessible to millions more viewers. Translation services coming soon.
1
22
273
Before/After Social Ghostwriting for X
Your CMO asks: "Why isn't our mental health content driving engagement?"
The answer isn't better SEO. It's better empathy architecture.
I help mental health apps, treatment centers, and therapists turn clinical expertise into content that builds trust at 2 AM—when someone's phone is the only light in the room.
This video shows three real transformations using the Digital Therapeutic Voice framework. Watch how the same message shifts from institutional to human without losing clinical integrity.
15
163
An executive director gave me an audio podcast created by Notebook LM about his nonprofit and asked me to create characters and b-roll for an engaging and informative video.
0
108
Before you sign your next contract, read "Don't Sign That," a best-selling contract literacy guide by Leo Mann.
As a creative strategist - using ai in my workflow greatly cuts down on the back and forth.
By integrating what I call an AI "rough-cut" or "pre-visualization" of the concept I am briefing for - I am able to provide a LOT more clarity to either the creator or AI engineer I am handing off to in the next step.
To be able to do this - I need to understand how to generate accurate UGC quality videos myself - so that the AI roughcut is close enough to what i want the intended final deliverable to be.
The role of creative strategist is increasingly becoming AI generalist 👩🏼💻 🤖
0
284
long time no intro!
I'm Julianna - the brain behind @creativesbyjulianna - and a performance creative strategist / director with 4 years of experience leading socials and strategy on accounts ranging from IRL community social clubs to gossipy AI chat apps.
I love diving deep into product, marketing, and consumer research and figuring out how to translate messaging into creative that actually lands — no matter how weird or niche the challenge.
My background is in data analytics, which means I don't just come with vibes. Every concept I pitch is built on a psychological human lever and backed by performance data worth testing. The idea has to feel true plus prove itself on paper. 📈
My workflow is something I genuinely pride myself on. I use tools and automations to reduce friction at every step — because what I'm trying to achieve with creative direction should actually get executed, with minimal back-and-forth. Efficiency is baked into the brief.
I'm a fast integrator. I'll jump into your team, learn your process, and make it better where there are gaps. I take initiative, I test things on my own, and I genuinely think learning is cool and gatekeeping is dumb.
If your creative isn't converting — it's probably a brief problem. Good news is that's fixable!
If you are building something cool that you are passionate about - let's chat.
AI Product Photography & Visual Content for Wellness Brands
$•••••
Earned
4x
Hired
5.0
Rating
33
Followers
AI Product Photography & Visual Content for Wellness Brands
My work was featured in Vanity Fair!!
Vanity Fair wanted to feature Mannam, and out of everything they could have run, the founder chose one of my images to represent the brand in the magazine. The whole campaign was AI-generated, built to stand next to traditional editorial photography without looking like AI.
I want to do more of this. Full campaigns for brands with a high visual bar. If that's you, reach out.
The brief was simple: create 4 weeks of GivingTuesday content.
The real challenge was bigger.
Caring Solutions already had stories, photos, community activities, and meaningful work happening every day. What they needed was a clear campaign that connected those moments to why people should care and take action.
I started by reviewing their social media, website, donor journey, existing fundraising content, and similar organizations.
From that research, I developed the “Moments That Matter” campaign and structured the content aroun.
Awareness → Connection → Understanding → Giving → Gratitude
I then turned the strategy into a practical 4-week execution guide their internal team could easily follow — including content direction, assets, CTAs, platforms, and the goal behind each post.
The client later shared:
“He did a wonderful thorough job with helping create a full campaign for us.”
This project reminded me that strong social media strategy isn’t about filling a calendar with posts.
It’s about giving every piece of content a reason to exist.
Here’s a look at the campaign system I created for Caring Solutions.
Social Media Strategy | Content Strategy | Campaign Planning
1
4
395
Instagram Branding & Canva Template Design | Social Media Management
3
210
Home decor catalog Design
I created a a clean modern home decor catalog that highlights each product with balanced layout consistent typography and high quality imagery
I help small businesses get more customers online.
5.0
Rating
14
Followers
I help small businesses get more customers online.
Designed and implemented a custom AI-driven lead capture system where customer conversations (chat or voice) automatically collect contact details, qualify leads, and push them into a single sales funnel.
The system triggers structured email follow-ups with a human-in-the-loop approval step, allowing small teams to review and control messaging before it goes out. This enabled 1–2 people to manage inbound leads, follow-ups, and scheduling efficiently—delivering big results without adding staff.
23
312
Custom AI Content Automation System for Brand Growth
Designed and implemented a custom content automation workflow enabling small teams to create and distribute content at scale with minimal manual effort.
The system integrates tools like OpenAI, Gemini, NotebookLM, HeyGen, ElevenLabs, and trend research into a single streamlined workflow—allowing one or two people to efficiently produce text, visuals, video, and audio while maintaining brand consistency.
What this enables
Faster research-to-content turnaround
Multi-format content generation
Consistent brand voice using templates
Reduced dependency on large marketing teams
Outcome
Significantly reduced content creation time and enabled higher output without increasing headcount. Built as a custom internal system tailored to the client’s workflow.
21
306
Yogaverse - Wellness App for Adults 40+
Built a full-stack wellness app that creates personalized yoga and exercise plans based on smart assessments.
Features:
• Intelligent recommendation engine analyzing flexibility, endurance & lifestyle
• Duolingo-style gamification (streaks, badges, leaderboards)
• Guided yoga sessions with step-by-step instructions
• Progress tracking dashboard
• Stripe subscription payments
• Secure user authentication
Tech: React, TypeScript, Node.js, PostgreSQL, Tailwind CSS, Stripe
Combines the engagement of Duolingo with Strava-style tracking - built for an underserved niche.
12
231
YouTube Podcast Translator — language access without platform limits
We built a custom YouTube podcast translation tool to remove friction caused by YouTube’s native translation limitations.
While some newer videos support auto-translation, millions of YouTube videos remain English-only due to creator settings, regional restrictions, or missing prerequisites.
This solution bypasses those constraints.
Users can paste any YouTube video link — old or new — and receive a translated audio version in their preferred language.
How it works:
Accepts any public YouTube URL
Extracts the original transcript
Translates content using LLMs
Converts the translation into natural-sounding speech
Outputs playable audio
The system works independently of YouTube’s translation rules.
Supported languages: English, Spanish, Hindi
(Designed to expand)