pro
Florencia Solari
Creative Technologist | AR/VR, WebXR, 3D & AI
Ready for work
Florencia is ready for their next project!
The Butterfly Trail
0
5
Harry Potter: Triwizard Tournament
0
3
Gesture-Controlled Immersive Portal (Gaussian Splats)
0
1
All-Access FIFA World Cup 2022
0
1
Royal Canadian Mint Virtual Museum
0
0
Organic Shader Lab
0
0
Glass Shader Woman: A Cinematic 3D Scene
0
0
NICEverse Multiplayer WebXR Gaming Platform
0
2
Team Jetpack: SUPERHERO BUILDER
0
1
Visions Laboratory Brand Identity and Teaser
0
2
Lancôme Beauty Tech AR Hologram Experience
0
0
www.xochiworld.com Interactive Portfolio
0
3
Express Your e.l.f.
0
0
AR Experience Development for Vauxhall Astra Electric Launch
0
0
Synth Muse: Audio-reactive Prototype
0
1
ARENA: Interactive 3D Experience
0
0
Ulta Beauty App AI/AR Skin Analysis
0
2
Lush World Bath Bomb Day Virtual Experience
0
0
Absolut: BORN TO MIX
0
0
Sol De Janeiro AR Mural
0
1
Diablo Immortal WebXR Portal
0
1
Identities: AR Virtual Try-On
0
1
Mindgaze: VR Mental Health Prototype
0
1
Official Messenger Video Calling AR Effects for Meta
0
1
Title: Dream-to-Video Creative Direction Workflow
For the Melius Contra Challenge, I built a reusable AI video workflow for creative directing short cinematic scenes from character references, environment references, mood references, and a written script.
I created it for my own dream-based storytelling practice. I’m writing a Substack series inspired by dreams, and I’m using AI-generated content and interactive web experiments to expand those stories into visual worlds. For this submission, I used one of those dreams as materia prima and worked with Mel step by step to shape the workflow and keep the output close to the vision.
The workflow generates character and environment previews, scene keyframes, individual video shots, a stitched final sequence, and a final restyle pass.
Process:
- Built a modular canvas with separate inputs for characters, environment, mood, and story direction.
- Used the Melius agent to structure the workflow into production nodes: creative brief, character consistency, environment direction, shot list, final prompts, and negative prompts.
- Added preview nodes for Character 1, Character 2, and the environment to lock visual consistency before generating video.
- Created multiple scene keyframes and video shots, then refined them through agent feedback and manual inputs.
- Added a stitching node to sequence the shots into one narrative video.
Feedback:
Melius was especially useful as creative direction canvas. It let me stay in the creative direction seat while the agent handled the technicalities, instead of having to solve every small detail by hand.
The main challenge was controlling reference fidelity, hallucinations, and consistency across the workflow. When making changes to prompts inside different nodes, I sometimes had to remind the agent of specific rules to preserve character identity, environment logic, and story continuity.
I’m excited to continue refining this workflow and use it moving forward to generate videos for my stories, (which you can read here https://thelaboratorio.substack.com (https://thelaboratorio.substack.com/) btw), as well as for client work.
Project link:
https://app.melius.com/projects/30a8d320-567b-4a16-aeaa-c14e81ec6358/canvas/5bc957a2-4e9a-4bba-a3ac-1fb0f0e22329
(https://app.melius.com/projects/30a8d320-567b-4a16-aeaa-c14e81ec6358/canvas/5bc957a2-4e9a-4bba-a3ac-1fb0f0e22329)I’m also keen to hear any feedback on the workflow, so let me know if you try it.
Social post:
https://x.com/xochiworld/status/2056829178020814885
1
103