Source footage for executive AI avatars, recorded to the platform’s specification by one cinematographer.
Why the footage decides the avatar
An AI avatar inherits every flaw in the footage it is built from. A flicker in the light, a slow drift in the frame, a hum under the voice or a glance away from the lens all end up in the avatar, and none of it can be fixed in post. The platform does the generating. What I deliver is the recording that makes the result usable: one continuous take per person, eye contact held with the lens, even light that does not change across the take, a steady frame and clean, uninterrupted sound.
That is a different discipline from a normal interview, where a cut, a cutaway or a grade can rescue a moment. Here the capture has to be right on the day, in one pass.
​
Recent avatar work: ODDO BHF, Zurich
In June 2026 ODDO BHF, the Franco-German private bank and asset manager, commissioned me to record the source footage for an executive AI-avatar project at its Zurich offices on Gartenstrasse. Two members of the leadership team were recorded to drive avatars generated on the HeyGen platform, for the bank's internal and client-facing communications.
I scouted the room in advance for light, background, power and acoustics, so the setup was solved before the participants arrived. Each executive read a prepared script from a teleprompter at the lens, in one take, and the footage was delivered in the format HeyGen ingests, ready for the bank to generate its avatars. The full case study is on my Zurich page.
​
How an avatar session works
It starts with the platform's own capture guidance, which sets the rules for framing, movement, light and audio, and with a short call about who is being recorded, in which languages and how the avatars will be used. I then check the room, either in person or from photos and a floor plan, because light from windows, air-conditioning noise and a busy background are the usual problems and all of them are easier to solve before the day.
On the day I light for an image that holds perfectly still for the length of the take, frame for the platform's crop and record dialogue on a Sennheiser MKH-416 or a Sanken COS-11 lavalier, with a backup recorder running. The participant reads from a teleprompter at the lens, or speaks to me through an Interotron rig when the script is loose, so eye contact never breaks. I review each take on a monitor before moving on, and deliver the files in the format and frame rate the platform asks for, on your own drives, keeping a backup for 14 days after delivery.
​
Who this is for
Avatar sessions suit communications and learning teams that want an executive to appear in more videos, and in more languages, than their diary allows, for internal updates, training or client communications. If your platform or vendor has sent you a capture spec, send it to me with your enquiry and I will tell you whether the room and schedule you have in mind can meet it.
​
Who you're hiring
I'm Theo Solnik, founder of Haikai Media and the cinematographer on every avatar session I deliver. I trained in directing and cinematography at the German Film and Television Academy Berlin (dffb) under Fred Kelemen, Béla Tarr's cinematographer on The Turin Horse, and I have been shooting corporate, documentary and cinema work for 30 years. My first feature, Anna Pavlova Lives in Berlin, won the inaugural Prize for Young Film Art (Preis der Nationalgalerie für junge Filmkunst). Corporate clients include Google, Pfizer, AWS, Merck and Novo Nordisk.
I work in English, Portuguese, Spanish, German, Italian, French and Norwegian, which helps when a participant records in more than one language. My own kit covers a RED Komodo 6K with Zeiss Standard Speed primes, a full lighting package, Sennheiser and Sanken microphones and Zoom recorders. Full biography and credentials here.
​
Pricing
Every avatar session is quoted individually based on the number of participants, languages, recording days and any finished edits you need alongside the source footage. Payment terms are 50% on signing, 50% within five business days of final delivery. I invoice under Italy's regime forfettario (Art. 1, c. 54–89, L. 190/2014), so my invoices carry no VAT.
​
Frequently asked questions
What footage does an AI avatar need?
Usually one continuous take per person with steady eye contact, minimal movement, even lighting, a quiet room and a locked-off frame. Each platform publishes its own capture guidance, and I record to that specification.
Which avatar platforms have you filmed for?
HeyGen, for ODDO BHF's executive avatar project in Zurich in June 2026. For Synthesia or any other platform I work from that platform's current capture guidance, so send it with your enquiry.
Can you record several people in one session?
Yes. For ODDO BHF I recorded two executives in one session. I set the room up once and keep the light and sound identical for each participant.
Do you make the avatars?
No. The avatars are generated by you or your vendor inside the platform. I deliver the source footage the platform needs, in the format it ingests.
​
Start your avatar project
Tell me how many people you need recorded, in which languages, where and by when. I read every enquiry myself and reply within one business day.
Client feedback
