Every Tarantella video starts the same way: with a conversation. Whether you're a business owner who needs a product promo, a family wanting to preserve a 50th birthday, or a wellness creator building a meditation library — the process of turning your idea into a finished video follows a clear, collaborative path.

Here's exactly how we work, step by step.

Step 1: Brief & Discovery

Everything starts with understanding what you actually need. We send you a short brief form — or simply have a call — where we ask about your goal, your audience, your tone, and any references or examples you like. The more context you give us, the better. But if you just have a rough idea, that's fine too.

Key questions we ask: Who is this video for? What do you want them to feel or do after watching? What's the ideal length? Do you have existing brand assets — logo, colors, fonts?

Step 2: Script & Narrative

Before any visuals are created, we write the script. This is the backbone of the video — it determines the pacing, the emotional arc, the voiceover, and the structure of every scene. We share the script for your approval before moving forward. Nothing is produced until you sign off on the words.

Step 3: Storyboard & Visual Direction

Once the script is approved, we create a storyboard — a visual plan showing how each scene will look. For AI video, this means selecting the right tools, styles, and reference images that will guide the generation process. We define the visual tone here: cinematic, warm, dynamic, serene, illustrated, realistic — whatever matches your brand and message.

Step 4: AI Production

This is where the magic happens. Using a combination of tools — Runway ML, Kling AI, Midjourney, HeyGen, ElevenLabs, and others — we generate the visuals, create the voiceover, and compose the original music score. Every element is produced to specification and assembled into a first cut.

The AI generates at speed — but a human creative director reviews, refines, and directs every frame. We are not prompt-and-ship. We iterate until the output matches the vision.

Step 5: Review & Revisions

You receive the first cut and give us feedback. We revise until you're happy — there's no arbitrary revision cap. Common revision requests: adjust the pacing, change the music feel, update a line in the voiceover, tweak the color grade. All normal, all included.

Step 6: Final Delivery

Once approved, we export the final video in whatever formats you need — MP4 for web and social, vertical cuts for Reels/Shorts, square for Instagram, and broadcast-quality ProRes if required. You receive full ownership and usage rights.

How long does the whole process take?

For a standard 60–90 second video: 5–10 business days. For complex productions with custom animation or multiple versions: 2–3 weeks. Rush timelines can often be accommodated — just ask.

Ready to start? Tell us about your project →

The most common mistake people make when creating a memory film is sending every photo they have. We understand the impulse — every photo feels important. But the best memory videos are built on carefully selected images that together tell a coherent, emotional story.

Here's how to choose and prepare your photos for a video that will genuinely move people.

1. Start with the Story, Not the Photos

Before opening your photo library, ask yourself: what is this video actually about? Is it a celebration of a person's life journey? A family reunion across decades? A tribute to someone who has passed? The answer determines which photos belong in the video and which don't — even if the ones that don't belong are technically good photos.

2. Aim for 20–40 Photos for a 2-Minute Video

A good rule of thumb: 10–15 photos per minute of video. For a 2-minute tribute, 20–30 photos is ideal. For a 3-minute film, 35–45. This gives each image enough screen time to be felt — not just seen.

If you have 200 photos, your job is to edit ruthlessly. Choose the ones with the most emotional charge, the most clarity, and the most variety.

3. Prioritise Variety of Format and Era

The best memory videos mix old and new — black and white with colour, formal portraits with candid moments, close-ups with wide group shots. Variety creates visual rhythm. If every photo is the same format and era, the video will feel flat.

• Mix close-up portraits with group shots
• Include photos from different decades if available
• Balance indoor and outdoor settings
• Include both posed and candid moments

4. Quality Matters — But Don't Discard Old Photos

Modern AI tools can significantly enhance older, lower-resolution photos — removing grain, sharpening faces, restoring faded colours. So don't exclude a photo just because it's from a 1980s disposable camera. The emotional content of an image matters far more than its technical quality.

That said, if you have a choice between a blurry and a sharp version of the same moment, always send the sharper one.

5. Organise by Chapter

If your video has a narrative arc (childhood → adulthood → family, for example), sort your photos into folders by chapter before sending them. This saves time in production and helps us understand the intended flow of the story.

6. Include a Few Words About Each Section

Even a sentence or two about the person, the event, or the context of each photo group is enormously helpful. "These are from Dad's years in the army, 1975–1978" gives us context that transforms how we treat those images in the edit.

Ready to create your memory video?

Once you've curated your photos, share them with us and we'll handle the rest — script, visuals, music, and final delivery.

The AI video landscape has changed dramatically in the past two years. Tools that seemed like science fiction in 2022 are now part of our daily production workflow. But there's also a lot of hype — and some genuine misunderstanding about what AI video can actually deliver today.

As practitioners who use these tools every day, here's our honest breakdown.

What AI Video Can Do Exceptionally Well in 2025

Generate photorealistic video from text or images

Tools like Runway ML Gen-3, Kling AI, and Sora can generate short video clips (5–15 seconds) from text prompts or reference images with increasingly impressive realism. Landscapes, product shots, abstract visuals, and stylised sequences are all achievable at a quality that would have cost tens of thousands in traditional production.

Create realistic AI avatars and spokespersons

HeyGen and similar tools can generate a realistic talking-head avatar from a single photo — or create a fully synthetic spokesperson in any language. Lip-sync quality in 2025 is good enough for corporate presentations, training videos, and social content.

Compose original music on demand

AI music tools like Udio and Suno can generate original, royalty-free music to match any mood, tempo, or genre — in seconds. The quality has reached a point where it's genuinely difficult to distinguish AI-composed music from human-composed equivalents in many categories.

Produce high-quality voiceovers in any language

ElevenLabs and similar platforms can generate natural-sounding voiceovers with emotional range, in virtually any language and accent. This has essentially eliminated the need for studio voice recording in most video production scenarios.

What AI Video Still Struggles With

Consistent characters across shots

One of the biggest challenges in AI video is maintaining a consistent character appearance across multiple generated clips. If you generate 20 shots of "a woman in a red dress at a beach", you may get 20 slightly different women. This is an active area of development — tools are improving rapidly — but it's still a limitation that requires human creative workarounds.

Complex narrative storytelling over 2+ minutes

AI excels at short-form content. For longer-form narrative pieces — a 5-minute documentary, a wedding film with emotional arc — the AI generates the raw material, but human editing, pacing, and storytelling judgment remain essential. This is not a weakness of AI video; it's a reminder that filmmaking craft still matters.

Working with proprietary footage you own

AI generation tools create new content — they don't yet seamlessly integrate your existing recorded footage with AI-generated scenes in a fully automated way. Combining real footage with AI-generated content requires skilled post-production work.

Where the Technology Is Heading

The pace of improvement is extraordinary. In the next 12–18 months, we expect consistent character generation across long-form video to become standard. Real-time video generation will follow shortly after. The distinction between "AI video" and "video" will become essentially meaningless.

For creators and businesses, the opportunity is now: AI video offers professional-quality production at a democratised price point. The organisations that adopt it early — and pair it with human creative intelligence — will have a significant advantage.

Questions about what AI video can do for your project?

Talk to us → We're happy to discuss what's realistic for your specific goals and budget.