This is Hamamoto from TIMEWELL.
AI Is Reshaping the Creative Industries
Artificial intelligence is advancing rapidly, and its impact on the creative industries is growing by the day. At the forefront of this shift is Sora, OpenAI's model that generates high-quality video from text. What once required specialized knowledge, expensive equipment, and considerable time is now within reach — but getting professional results from Sora takes more than entering a prompt. It requires a strategic approach and a structured workflow.
This article explores the practical methodology developed by acclaimed DP (director of photography) and director David Sheldrick, based on his experience producing music videos with Sora. Sheldrick adapted a format he developed before COVID — designed to complete shooting efficiently in a single day — and applied it to systematic high-quality video production with AI. His process covers everything from the initial creative concept through rendering, music selection, and detailed editing. For business professionals interested in AI-powered visual communication, this is essential reading.
Contents:
- Unlocking creativity: From concept to world-building
- Using the Sora Explore page for research and prompt learning
- Defining your world: From time period to visual elements
- ChatGPT integration: Refining prompts and building style presets
- Systematizing multiple creative elements: The Marie Antoinette example
- Execution phase: Sora rendering and music selection
- Applying style presets and efficient rendering strategy
- Adding dynamism: Inserting dance sequence prompts
- The importance of music: Using Artlist.io
- Assembly and finishing: Building the timeline from raw clips
- The "Sausage" method: Organizing material and building structure
- Music sync: Cutting to the beat
- Speed adjustment and transitions
- Real editing time: The road to a finished assembly
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
Step 1: Building a Strong Creative Foundation
Every Sora video production journey begins with establishing a solid creative foundation. This is not casual brainstorming — it is the phase that determines the quality and direction of the finished work. As Sheldrick emphasizes, before starting any assembly work you should invest at least half a day, ideally a full day, exploring different creative directions and experimenting. The trial and error at this early stage is what smooths out the later steps.
Using the Sora Explore Page
One of the best ways to find creative inspiration and learn effective prompting is the "Explore" page on Sora's official website. This page shows videos generated and shared by users worldwide, along with the prompts used to create them. It is not just a gallery — it is a practical school for prompt engineering. By observing how other creators use specific language to achieve particular visual effects and atmospheres, you can dramatically improve your own prompting skills. Phrases like "cinematic shot," "detailed skin texture," and "dynamic camera movement" all have measurable effects on the generated output, and seeing them in context accelerates learning.
World-Building: Define the Universe First
Alongside research, work on defining your "world" — the unified atmosphere, style, and time period that will run through the entire piece. Before filling in the details of individual scenes, establish the overarching tone. Is this a historical period piece? A futuristic sci-fi setting? A pastoral landscape? An ornate palace? These foundational choices determine your costume, production design, props, lighting, and color palette. Sheldrick's example was "18th-century Marie Antoinette" — a clear, evocative anchor for all subsequent creative decisions. World-building is not just visual; it deepens the narrative and thematic identity of the work.
ChatGPT Integration: Turning Concepts into Prompts
To translate a defined world into concrete Sora-ready instructions, Sheldrick recommends using a large language model like ChatGPT. Enter your keywords and concept with an instruction like: "Expand this prompt and make it more detailed for use in Sora video rendering as a preset." From a simple idea like "18th century, Marie Antoinette style," you can generate a rich prompt that specifies fabric textures (silk, lace, brocade), color palette (pastels, gold accents), lighting atmosphere (soft natural light, chandelier shimmer), camera movement (elegant dolly shots, close-ups), and mood (decadent, romantic, whimsical). The resulting detailed prompt becomes a Sora "style preset" — a powerful tool for maintaining visual consistency throughout the project.
Step 2: Systematizing Your Creative Elements
With your overall style established, map out the specific scenes and creative elements that will live within it. Sheldrick's Marie Antoinette example included:
- Creative 1: Hair and makeup close-up — an enormous 18th-century wig, white powder makeup, focus on the model's expression
- Creative 2: Grand palace interior — wide corridors and ballrooms, capturing the architecture and lavish decor
- Creative 3: Hunting scene — the noble pursuit, horses, costumes, natural landscape
- Creative 4: Garden — perfectly manicured hedge mazes and geometric gardens, like Hampton Court Palace
- Creative 5: Horses — as aristocratic motifs, elegant movement and harness detail
- Creative 6: Kintsugi model — a conceptually interesting visual referencing the Japanese art of repairing ceramics with gold, functioning as a metaphor for finding new value in what is broken
Placing multiple specific creative elements under a single overarching style ensures variety while maintaining coherence. Each creative element has its own prompt but is anchored to the master style preset. This structure makes the rendering workflow efficient and simplifies asset management during editing.
Step 3: Rendering and Music Selection
Once the creative framework is defined, it is time to render with Sora. From the "Manage Presets" menu, paste your detailed style prompt to register it as a preset. From that point on, individual creative prompts (hair and makeup, palace, garden, etc.) can be entered without repeating all the style details — the preset handles consistency.
Sheldrick's core rendering principle: render each creative element many times. Because AI generation is probabilistic, the same prompt produces slightly different results each time. By running multiple generations per scene, you collect a pool of clips that includes both close matches to your vision and unexpected happy accidents. This "volume approach" is an efficient strategy for AI content — you select the best from your collected clips or combine several to achieve a richer result. Presets can be adjusted iteratively as you review results.
Adding Dynamism with Dance Prompts
One characteristic of Sheldrick's method is frequently inserting a "second prompt" focused on dance sequences. Even for a hair and makeup scene, he adds descriptors like "bold camera shot of ethnically diverse K-pop couture fashion while dancing in unison, dancing in a Queen's bedroom, crunk dancing, street dance, dancing with attitude, dynamic dance, movement, dynamic music video camera work." Specifying specific dance styles (K-pop, crunk, street) alongside abstract cues like "attitude" and "dynamic movement" encourages Sora to interpret the scene with unexpected energy. The goal is to keep even potentially static scenes from feeling flat.
Music Selection: Using Artlist.io
Music is not background noise — it defines the atmosphere, rhythm, and emotional impact of the entire piece. Sheldrick recommends high-quality stock music platforms rather than current AI-generated audio, which he considers still too inconsistent. His personal choice is Artlist.io, valued for its track quality.
Music selection typically happens when you have a reasonable collection of rendered clips and are beginning the assembly. The chosen track becomes the blueprint for editing: its structure (intro, verse, chorus, bridge, outro), rhythm, and energy peaks drive all decisions about when to cut, how long to hold a shot, and where to place transitions. On platforms like Artlist.io you can filter by genre, mood, tempo, and instruments, making it practical to find music that matches the specific atmosphere of your world.
Step 4: Assembly — Bringing It All to Life
With creative direction set, rendered clips assembled, and music selected, the final editing phase begins. Sheldrick calls this "Assembly" — not a mechanical arrangement of clips, but a series of creative decisions about which shots to use, in what order, at what length, and synchronized to the music.
The "Sausage" Method
The first step is what Sheldrick calls the "sausage" — laying all generated clips end-to-end in a single unbroken chain on the timeline. At this stage, don't worry about sync or timing; the goal is simply to get a visual overview of everything you have. From there, rough-group the clips by creative element (all hair and makeup clips together, all palace shots together, etc.) and begin mapping them against the music structure. Which creative element belongs in the intro? Which climactic images hit at the chorus drop?
Cutting to the Beat
Once the music track is on the timeline, the real editing work begins: matching cut points to musical moments. Sheldrick pays particular attention to bass hits and drops — moments where the rhythm asserts itself most strongly. If the bass enters and you want a model's eyes to open at exactly that moment, you trim the clip's start point to hit the beat precisely. When a clip contains an unintended internal cut (which Sora-generated footage sometimes does), split it at that point and remove the problematic section. This granular work — listening carefully to the music and responding with visual cuts — is what separates a finished video from a collection of AI-generated clips.
Speed Adjustment and Transitions
Speed adjustment is a valuable tool beyond simply matching length to music. For a three-second musical phrase with a five-second clip, accelerating to roughly 167% hits the timing exactly. Slow motion can emphasize a specific movement or create dramatic effect. Combining fast cuts with slow motion creates dynamic sequences with genuine pacing variety. Transitions — from simple cuts to fades, dissolves, and more creative effects — further shape the flow and emotional register of the piece.
How Long Does It Take?
Sheldrick notes that assembly takes real time. The demonstration showed only the first few scenes being built, and he described the full assembly from "sausage" to a rough cut as taking "an hour or two." His complete assembly reportedly took about four hours. This varies with the volume of rendered material, the complexity of the music, and the editor's quality standards — but it confirms that even with AI generating the source footage, the editorial work of shaping it into something meaningful still requires human creativity and time.
What David Sheldrick's Method Teaches Us
Sheldrick's Sora workflow is a masterful fusion of cutting-edge AI technology with the systematic approach developed in traditional film production. His method illuminates four universal principles of high-quality video production:
- Clear vision and world-building
- Structured creative development
- Tight integration with music
- Creative trial and error in editing
The front half of the process — research, prompt refinement with ChatGPT, preset building, creative decomposition, and high-volume rendering — is a refined strategy for turning ideas into visuals efficiently. The defining move is establishing an overarching style first, then populating it with specific creative elements, which controls the inherent variability of AI generation while maintaining a coherent aesthetic.
The back half — the sausage method, beat-matched cutting, speed adjustment — is how human sensibility and rhythm get injected into AI-generated material, elevating a collection of clips into a story that moves an audience. The four-plus hours Sheldrick spent on assembly underscores that AI automation handles part of the production process, but the human creativity required to make that material meaningful has not gone away.
Beyond music videos, this approach applies directly to corporate promotions, product introductions, brand content, short-form social video, and training materials. High-quality visual expression that once required significant budgets and specialized crews is now much more accessible. With Sora and a structured workflow, focusing time on the upstream work of concept development and world-building turns AI into a powerful visualization tool that can substantially expand what's possible in marketing and communications.
AI video generation technology is still developing, but the pace is extraordinary. More intuitive controls, higher fidelity rendering, better AI-assisted music and editing — all of these advances are coming. Learning from pioneering practitioners like David Sheldrick is how we stay ahead of that wave. The key is not to be driven by the tool, but to use it with clear purpose and strategic intent, releasing human creativity rather than replacing it.
Reference: https://www.youtube.com/watch?v=0dhX84UkwFs