Build a Complete AI Video Ad From Scratch | 27 September, Kolkata

Build one finished vertical AI video ad in three hours: face lock, base stills, Veo motion, ElevenLabs voice. 27 September 2026, Salt Lake, Kolkata. ₹799. No filmmaking background.

Build a Complete AI Video Ad From Scratch

Walk in with an idea. Walk out at 2:00 PM with a finished vertical ad: one face-locked character, 4 to 6 production stills, motion clips, and a natural voiceover, built on one repeatable prompt system. Three hours, offline, no filmmaking background required.

Sunday, 27 September 2026, 11:00 AM – 2:00 PM. Salt Lake, Kolkata. 50 seats. ₹799.

What you build

• The locked actor: one continuity-safe face reference with real skin texture, so the same person survives every scene
• Four to six production base stills: vertical 9:16 scenes planned through explicit camera, lens, and lighting instructions
• The motion clips: your approved stills turned into controlled video in Google Flow and Veo, with handheld movement instead of random morphing
• A human voiceover: natural cadence built with ElevenLabs bracketed delivery cues, or your own recorded take run through voice changing
• The recipe: the master prompt stack saved on your laptop, so the next ad starts at step one instead of a blank chat window

What to bring

A laptop and charger. This is three hours of building, not a talk. Bring headphones, your phone, and one product, service, or offer you actually sell. If you would rather practise on something else, we hand you a brief at the table.

Our first workshop

Our first workshop in June 2026 was about making websites and hosting them. Each workshop stands alone; no earlier session is required.

Visit us in Kolkata

Salt Lake is also known as Bidhannagar. The exact workshop venue is shared after reservation. Ask about travel or accessibility before booking.

+91 98307 15557 · theaiworkshop.in@gmail.com

The three-hour plan

1. 11:00–11:40: The brief and the face lock

   Write the brand brief in plain words, lock who the ad is for and what it must make them believe, then choose one format: UGC or cinematic. Generate the master character anchor — one headshot, one person, no collage — and save the face-lock block you will reuse all day.

2. 11:40–12:40: Scene breakdown and the base image pipeline

   Turn the script into scene-by-scene visual beats, then convert each beat into a standalone 9:16 image prompt with the skin-enhancer protocol attached. Generate, inspect, and approve one still per scene before anything moves.

3. 12:40–1:30: Motion and voice injection

   Feed approved stills into Veo with prompts that describe movement only — camera behaviour, subject movement, continuity notes, negative motion rules. In parallel, build the ElevenLabs prompt with bracketed cues and test one line before generating the whole script.

4. 1:30–2:00: QC and final packaging

   Run the checklist: nightmare hands, distorted text, broken continuity, background strangers, robotic delivery. Fix only what matters, stitch the edit, then save the named folder structure and the prompt stack you will open again on Tuesday.

Questions and answers

Need a filmmaking background?

No. The system replaces the film-school parts with written instructions: what the camera does, how the light falls, what stays the same between shots. If you can type and follow steps in order, you can produce the ad. We start from an empty chat window in the room.

What is finished by 2:00 PM?

One vertical ad for one product or offer: a face-locked character, four to six base stills, the motion clips generated from them, a voiceover, and the stitched edit on your laptop. You also leave with the prompt stack saved, so the second ad does not start from scratch.

Do the tools cost money?

Some do, and we will not pretend otherwise. Strategy and prompt writing run on the free tiers of ChatGPT or Claude. Image generation, Veo video generation, and ElevenLabs voice all work on credits, and free allowances are small and change often. Budget roughly ₹500 to ₹1,500 of your own credits if you want to render a full ad on the day. We will show you where credits get wasted so you spend fewer of them.

Why do my AI videos look plastic?

Usually because the prompt gave the tool an idea instead of a production plan, and because a weak base image was pushed straight into video. A person melting between frames is almost always a base image problem, not a video problem. The session fixes the order of operations: lock the face, approve the still, then add motion.

Can I use my own product?

Bring your own. D2C founders, agency owners, freelancers, and local business owners get the most out of the day because the output is an ad they can actually run. If you do not have one ready, we hand you a brief at the table and you build with that.

Workshop details · Reserve your seat · The June workshop · About