Evolving VFX and animation with generative AI tools

In the film industry, generative AI is gaining traction as a technology that can assist in reshaping production workflows

Image credit: Sora and Canva

Visual effects (VFX) and animation have long been central to cinematic storytelling, enabling filmmakers to create detailed environments, characters, and sequences that would otherwise be impossible or cost-prohibitive. Today, generative AI is being explored as a new toolset within this domain—one that is beginning to change how these visual elements are produced and refined.

This newsletter looks at some of the ways generative models are being applied in VFX and animation workflows, and how the technology is shaping both creative processes and production strategies.

Recent advances in generative AI for film production

Over the past year, several generative AI platforms—such as Runway, Adobe Firefly, NVIDIA Picasso, and OpenAI’s Sora—have progressed from experimental tools to increasingly practical resources in media creation. These models now offer capabilities that support creative professionals in tasks like:

• Visualizing complex scenes from simple inputs

• Animating characters or backgrounds with minimal manual setup

• Enhancing or editing footage with reduced reliance on traditional post-production tools

Rather than replacing artists, generative AI is being positioned as a complementary tool that can streamline certain tasks and facilitate early-stage experimentation.

Applications in visual effects

Within VFX workflows, generative AI is already being used in various capacities, including:

• Producing digital environments from sketches or text prompts

• Simulating natural elements such as weather or fluid dynamics

• Matching lighting and color across different shots

• Assisting with digital doubles or de-aging effects that were previously more labor-intensive

Platforms like Runway’s Gen-4 model are also enabling users to create concept scenes and story previews using text-to-video prompts—offering new methods for prototyping ideas before investing in full-scale production.

Supporting animation workflows

Generative AI is also being integrated into animation pipelines. Tools such as Flow Studio and Pika allow artists to:

• Animate static illustrations or keyframes

• Auto-generate movement, transitions, and facial syncing

• Rapidly prototype scenes and storyboards

This approach can reduce the time and resources needed for early development stages. For independent creators, it offers a faster route to visual output. For larger studios, it opens the door to more iterative exploration during pre-production.

Ethical and professional considerations

As with any emerging technology, the use of generative AI in creative industries raises important questions. Key areas of discussion include:

• Intellectual property rights for AI-generated content

• Potential workforce impacts, especially in post-production and animation roles

• Consent and control over the digital representation of individuals

While some film studios are cautiously integrating AI tools into their production pipelines, others are evaluating policies to define acceptable use cases. In parallel, industry professionals are engaging with the technology to better understand its implications for originality, authorship, and long-term career viability.

Looking ahead

Though generative AI is still in the early stages of adoption in the film industry, its practical applications are expanding rapidly. Its potential lies in enhancing efficiency, supporting creative ideation, and augmenting existing workflows—not replacing them.

As the technology evolves, its role in visual storytelling is likely to grow—raising both opportunities and challenges that will need to be addressed collaboratively by creators, technologists, and stakeholders across the industry.

📚 Bibliography / Further reading

1. Runway Gen-4 – AI Video Generation Platform
   Runway. (2025). Gen-4: Advanced AI tools for text-to-video and image-to-video generation. https://runwayml.com

2. OpenAI – Sora: Text-to-Video Model
   OpenAI. (2025). Sora: Creating video from text.
   https://openai.com/sora/

3. NVIDIA – Edify: AI Foundry for Generative Models
   NVIDIA. (2025). Edify: AI Foundry for Visual Generative AI Models.
   https://www.nvidia.com/en-us/gpu-cloud/edify/

4. Adobe Firefly – Generative AI for Creatives
   Adobe. (2025). Firefly: Free Generative AI for creatives.
   https://www.adobe.com/products/firefly.html

5. Pika Labs – AI Text-to-Video Generator
   Pika Labs. (2025). Pika Labs AI Text to Video Generator. https://pikalabsai.org/

6. Autodesk Flow Studio – AI-Powered CG Animation
   Autodesk. (2025). Empowering Animators with AI Innovation: Autodesk at Annecy Film Festival 2025.
   https://blogs.autodesk.com/media-and-entertainment/2025/05/16/empowering-animators-with-ai-innovation-autodesk-at-annecy-film-festival-2025/

7. Google Veo 3 – AI Video Generation Tool
   Times of India. (2025). Will Google's Veo 3 AI technology revolutionize filmmaking or threaten creative jobs? Netizens react.
   https://timesofindia.indiatimes.com/entertainment/english/hollywood/news/will-googles-veo-3-ai-technology-revolutionize-filmmaking-or-threaten-creative-jobs-netizens-react/articleshow/121420132.cms

8. SIGGRAPH 2024 – Creativity and Innovation at the Intersection of AI, Computer Graphics, and Design
   SIGGRAPH. (2024). Creativity and Innovation at the Intersection of AI, Computer Graphics, and Design.
   https://s2024.siggraph.org/creativity-and-innovation-at-the-intersection-of-ai-computer-graphics-and-design/

9. Deloitte Insights – Generative AI and Hollywood
   Deloitte. (2025). Generative AI and Hollywood: Studios cautious of GenAI adoption.
   https://www2.deloitte.com/us/en/insights/industry/technology/technology-media-and-telecom-predictions/2025/tmt-predictions-hollywood-cautious-of-genai-adoption.html

10. Forbes – AI and Hollywood's Next Golden Age
    Forbes. (2025). AI And Hollywood's Next Golden Age.
    https://www.forbes.com/sites/charliefink/2025/04/28/ai-and-hollywoods-next-golden-age/

More news!

Understanding the rise of "AI slop"

In May 2025, a Reddit post titled "[Am I the asshole] for telling my husband's affair partner's fiancé about their relationship?" gained significant attention, earning over 6,200 upvotes and 900 comments. However, the post was later removed by moderators amid suspicions it was written by artificial intelligence. Indicators included repetitive phrasing and a scenario seemingly crafted to provoke reactions rather than reflect a genuine situation.

This post reflects a broader trend: the rise of AI-generated content across social media. Often referred to as "AI slop," this type of content includes low-effort posts, comments, and images designed to attract engagement. It appears on platforms like Reddit, Facebook, and LinkedIn, where estimates suggest a significant portion of longer English-language posts are now written by AI.

While some AI-generated content may be harmless or entertaining, concerns are growing about its potential to mislead or manipulate. Reports have highlighted how such content can be used to boost visibility artificially, spread misinformation, or promote products and services through low-quality or deceptive tactics. Researchers also warn that AI-generated content could affect political discourse, with influence campaigns emerging around major elections.

Attempts to address the issue include efforts by social media platforms to detect and limit low-quality or spam-like posts. However, detection remains difficult—especially for short-form content or increasingly realistic images and text. Some researchers and platforms are exploring labeling tools and improved AI detection methods, though these approaches are still developing.

Meanwhile, some users are leaving mainstream platforms, frustrated by the volume of AI-generated material. This shift may lead to more fragmented online communities and greater polarization, as people seek spaces aligned with their views.

Overall, the growing presence of AI-generated content raises questions about authenticity, trust, and the future of online interactions. As both humans and algorithms are influenced by this evolving content landscape, finding effective ways to manage and understand its impact remains an ongoing challenge.

Study explores how language models handle social interactions through game theory

Researchers from Helmholtz Munich, the Max Planck Institute, and the University of Tübingen examined how large language models (LLMs), including GPT-4, Claude 2, and Llama 2, behave in cooperative and competitive interactions. Using classic two-player games like the Prisoner’s Dilemma and Battle of the Sexes, they found that LLMs tend to perform well in competitive settings but struggle with tasks that require coordination and compromise.

The study, published in Nature Human Behaviour, applied behavioral game theory to assess whether LLMs can adapt to social dynamics over repeated interactions. While the models often acted in their own interest, their performance improved with simple interventions—such as prompting them to predict a partner's move before responding.

These findings suggest that current LLMs lack robust social intelligence but can be guided toward more human-like collaboration. The researchers hope this work will inform the development of AI systems that are not only accurate but also capable of empathy and cooperation in fields like healthcare, education, and social support.