Storyboard to animatic, with a scratch voiceover, in an afternoon

Why now

An animatic used to be a storyboard artist, a sound booth and two days in an edit suite. Three recent releases make a convincing rough version an afternoon's work:

• Midjourney V8.2 became the default on Jul 24, 2026. It brings a new Edit Model that replaces Omni Reference and Character Reference, and 2K HD images without upscaling. Both help when the same character has to appear in six frames. (Version docs)
• Runway's unified pricing (Sep 18, 2026) puts Gen-4.5, Aleph 2.0, Veo 3.1 and Kling 3.0 on one credit balance, so you can pick the best model per shot. (runway.com/pricing)
• Eleven v4 (Sep 28, 2026) is ElevenLabs' most expressive text-to-speech model, aimed at voiceover and dialogue. (ElevenLabs blog)

Who it's for

Anyone pitching a film, ad or brand video who knows a static deck undersells it: agency creatives, in-house brand teams, directors applying for a grant, founders storyboarding a launch video.

What you get

• A 6-shot, ~30-second animatic (16:9, 1080p) with a scratch voiceover, captions and temp sound.
• A shot list with the exact prompt behind every frame and clip, so the real production team can see intent.
• A rough you can show as a rough. This is a pitch tool, not a final.

You'll need

Setup

1. Write the shot list (any doc, 20 min)

Six shots of about 5 seconds each. One row per shot:

#  | Shot (what we see)                         | Camera            | VO line
1  | Wide: lighthouse keeper climbs stairs, dusk | slow push-in      | "Every night for forty years..."
2  | ...                                        | ...               | ...
Style line (for every prompt): <e.g. "35mm film still, muted teal and amber, soft grain">
Character line (for every prompt): <e.g. "a woman in her 60s, cropped grey hair, yellow oilskin">

The style and character lines are the trick. They go into every prompt verbatim.

2. Lock the look with one hero frame (Midjourney, 30 min)

On the Midjourney website, generate shot 1 with:

<shot 1 description>, <character line>, <style line> --v 8.2 --ar 16:9

Iterate until one image is right. Then use it as the reference image for every other shot, editing with the Edit Model (or add a Style Reference code), so the character and palette stay consistent. Turn on HD for the final picks. Download all six.

3. Animate each shot (Runway, 60 min)

In Runway, start an image-to-video generation for each frame. Gen-4.5 is a sensible default, but choose whichever model in the picker accepts an image input and suits the shot. Set the length to 5 seconds and use a motion-only prompt:

Slow push-in. The woman climbs two steps; her coat moves in the wind. Camera steady.
No new characters. No text.

Describe only motion and camera. The image already carries the look. Generate 2 takes per shot and keep the better one. If a take is nearly right, edit it with Aleph ("remove the second figure", "make it dusk") rather than regenerating.

4. Record the scratch voiceover (ElevenLabs, 20 min)

In ElevenLabs' text-to-speech, choose Eleven v4. Pick a library voice (don't clone anyone without written consent) and paste the VO lines as one script, with a blank line between shots. Generate, listen, regenerate lines that land wrong, and download the MP3.

5. Assemble (Descript, 40 min)

Create a project and drop in the six clips in order, then the VO. Ask Underlord:

Align each video clip to its VO line in order (one clip per paragraph). Add
burned-in captions from the VO, a 0.5s crossfade between shots, and a 2-second
title card at the end: "<project> — animatic, not final".

Add temp music if you have a license for it (Eleven Music works; see Drop 001). Export 1080p.

6. Package the pitch (5 min)

Share the MP4 with the shot list as a one-page PDF beside it, prompts included. The "not final" card and the prompts make it obvious what's AI scratch and what the real shoot will improve.

Time to set up

About 3 hours for a 30-second animatic (estimate, not timed). Runway generation time is the most variable part.

Cost

• Minimum: Midjourney Basic $10 + Runway Standard $15 + ElevenLabs Starter $6 = $31/mo, with Descript Free (watermarked export).
• Clean export: + Descript Hobbyist $16/mo annual.
• Client work under NDA: Midjourney Pro ($60/mo, $48 annual) for Stealth Mode.

Runway credit math: Gen-4.5 is listed at 12 credits per second, so 6 shots × 2 takes × 5 s = 720 credits. That's more than Standard's 625. Either take one try on easy shots or step up to Pro (2,250 credits, $35/mo). Aleph edits cost more per second than plain generation. Check the in-app price before you run them.

Prices are list prices as of Oct 14, 2026.

Gotchas

• Midjourney images are public by default. Stealth Mode is Pro and Mega only. Anything you generate on Basic or Standard is visible in Midjourney's gallery. Don't put an unannounced client brand in a Basic prompt.
• Companies over $1M revenue need Pro or Mega under Midjourney's terms.
• Consistency drifts by shot 5. Re-use the hero frame as the reference every time, not the previous shot, or the character slowly mutates.
• Runway credits don't roll over on Standard or Pro (Max has one month of rollover). Don't buy on the 28th and start on the 1st.
• Voice consent. Use library voices, or clone only a voice you have written permission for. ElevenLabs Starter includes a commercial license, but that covers the output, not someone else's likeness.
• Label it. Clients and festival juries increasingly ask what's AI-generated. Say so up front.

Remix

• Vertical cut: generate at 9:16 in steps 2–3 and you have a social teaser alongside the pitch.
• Live-action reference: shoot phone footage of the blocking, and use Runway's video-to-video tools to restyle it instead of animating stills.
• Figma-first teams: do the frames in Figma Weave and the cut in Figma Motion (Drop 001), and use this drop's ElevenLabs and Descript steps for VO.

Freshness log

• Oct 14, 2026: First published. Desk-checked against Midjourney's version and plan comparison docs, Runway's pricing page and model credit costs, ElevenLabs' Eleven v4 announcement and pricing page, and Descript's pricing page.

Last verified: Oct 2026 (desk-checked, not run end-to-end). We checked every step, setting and price against the vendors' current docs. We have not yet run this chain live. If a step doesn't match what you see, reply and tell us. It'll be fixed in the next refresh.