Between improvisation and rigour: making an audiovisual work with AI
Workshop with Jacek Nagłowski, new media visual artist and researcher focusing on virtual reality and experimentation with AI systems. He has produced numerous documentary and fiction films, frequently as international co-productions. His films and VR works have screened at Cannes, Venice, Sundance, IDFA, and New Horizons.
Short Description
A five-day workshop for filmmakers making one short AV poem, up to three minutes, with AI systems for text, image, sound, and video. Two modes of working are practised: improvisation, for finding what these systems can do that the author could not specify in advance, and rigour, for holding an authorial voice against AI’s statistical defaults. The aim is work that uses AI without being shaped by it.
Most discussions of AI in audiovisual work frame these systems as tools for speed, cost reduction, and the replacement of human labour. Here they are treated instead as material and as collaborators: capable of opening territory the author could not specify in advance, but also pulling the work toward the average of their training data when no authorial pressure is applied. That average has a name: AI slop.
Who is it for?
The workshop is intended for filmmakers at any career stage who are beginners or moderately advanced in working with AI. It is not aimed at practitioners arriving with settled AI production pipelines. It suits people who are still open to changing how they work with these systems.
Each participant arrives with a short prose statement, around half a page, describing the subject and tone of the poem they intend to make. This is not a draft of the poem; the poetic form is developed on Day 1 through an improvisation exercise with language models.
Across the five days, each participant works through the production chain of an AV poem: textual development with language models, image generation and reference development, sound and voice work, video generation, and editing. Theory and discussion sit mostly in the first two days; the later days are practical.
Outcomes
Participants will leave the workshop with:
- a finished short AV work (up to three minutes) made individually with AI systems
- working knowledge of current text, image, sound, and video generation systems and how they differ
- an individually developed workflow for working with AI
- practical experience of moving between improvisation and rigour
- an ability to recognise and resist AI slop in their own work
- references, prompts, and methods to take forward
- a clear sense of these systems’ current capabilities and limits, and of how quickly both shift
Participants will receive a certificate of completion.
Workshop Info
Workshop dates: 24/08/2026-28/08/2026
Registration deadline: 24/07/2026
Notification of acceptance : 26/07/2026
Fee: 800 €
Participants: 6-8
Language: English
Hours: approx. 10-18
Place: Lodz Film School
Participants bring their own laptops and their own access to language models of choice. Personal LLM accounts are required because conversations about personal texts often become intimate, and this material should stay with the participant rather than on shared school infrastructure.
Access to image, sound, and video generation platforms (including HiggsField, subject to confirmation) is provided by the school.
Participants must arrive with basic editing literacy and basic familiarity with video-editing software, and must have their own editing tools available throughout the workshop.
A note on environmental footprint: generative AI work has measurable energy and water costs. The workshop addresses these directly as part of the practice.
An online introductory meeting will be organised before the workshop to align expectations and confirm what each participant is bringing.
Contact: workshops@filmschool.lodz.pl
Application Requirements
Applicants should submit:
- a half-page prose statement describing the subject and tone of the AV poem they intend to develop during the workshop (this is not a draft of the poem)
- a short CV or bio
- a link to one previous work in any medium (film, writing, visual, sound) that gives a sense of authorial voice
- a short paragraph on why they want to take this workshop
Communicative English is required.
Workshop Structure
Day 1 – Foundations and the Poem
Introduction to AI as material and as collaborator. How current generative systems work in practice, including their internal mechanisms, what kinds of output they reward, and their systematic blind spots. Introduction to authorial voice as a working principle in relation to AI. Afternoon: developing the poem from the participant’s prose seed through improvisational dialogue with language models.
Day 2 – Image, Reference, and Visual Direction
Working with text and image generation systems. Comparison of current models and their characteristic aesthetics. Methods for developing a consistent visual language for the for the audiovisual work instead of inheriting the model’s default tendencies and biases. Practical session: producing visual references, base images, and early production materials.
Day 3 – Sound, Voice, and the Audio Layer
Sound and music generation systems.Voice generation, including speech synthesis and voice cloning, with attention to the ethics of voice work, Music and SFX generation. Building the audio layer of the poem in parallel with continued visual work. Practical session: generating the soundtrack and voice elements; continued video generation.
Day 4 – Production and Iteration
Fully practical day. Video generation across multiple platforms. Iterative refinement, problem-solving, and editing of the assembled materials. Individual consultations throughout. The tutor’s role today is troubleshooting and discussion of authorial decisions.
Day 5 – Finishing, Screening, and Reflection
Final editing and refinement. Internal screening of the completed AV poems. Collective discussion of the process. Reflection on what each participant has learned about their own authorial voice through this work as well as surprises and limitations of current generative AI systems.
About the Tutor
Jacek Nagłowski is a filmmaker, new media artist, and researcher. He has produced or co-produced more than thirty feature films, documentaries and VR experiences, and has written and directed work across cinema, VR, and AI-based audiovisual practice. From 2018 to 2023, together with Paulina Borkiewicz, he led the VR/AR Studio at the Visual Narratives Lab of the Lodz Film School (vnlab.org). His works as both, producer and director, were screened and awarded at multiple international festivals, including Venice, Cannes, Sundance, IDFA, DOK Leipzig or Krakow.
Since 2020 his work has centred on artificial intelligence as a practice and a research subject. He teaches AI at the Lodz Film School and runs AI workshops for cultural institutions and NGOs (Film Spring Open, Mozilla Foundation, and others). His recent AI work includes the audiovisual essay series, the VR experience co-created with AI systems across text, image, sound, and Unreal Engine programming and an autonomous parallel reality AI radio system. He is currently building AI Social Lab, a non-profit dedicated to developing AI systems for the public good.
