large image

AI Video Production (2024)

AI Video production Header Image

AI Video Production (2024)

AI & Video Production updated on 26/04/2024

Artificial Intelligence was the new trend in 2023. Between Chat GPT, and Midjourney, we’ve seen a potential future whereby the written word, and the still image could be created by an AI. Is this the end of human creativity? (spoiler, no) and what’s next for video?

While the internet lost its collective mind over the potential of AI, I was reminded of a project from 2019; Lexus, Driven By Intuition. This car advert was an experiment in AI script-writing. Partnering with IBM to use its Watson technology, Lexus and The&Partnership input 15 years’ of Cannes Lion award-winning ads and 10 years of the best ‘luxury’ ads, as well as additional data relating to ‘intuition’ and how people make decisions. Based on this data, Driven By Intuition was written.

The salient point of this experiment was that while the AI crafted a perfectly passable automotive advert; it left viewers slightly confused, and more importantly, it lacked originality. Driven By Intuition feels like a familiar car advert because it is just that… familiar.

What is the future of AI Video Production in 2024?

So, for a while at least, true originality and creativity can be left to the human mind. Ultimately a lot of the AI technology out there still recycles deep, human-created data resulting in simple variations of what has come before. 

However, that is not to say that Artificial Intelligence doesn’t have a place in the video production industry. In fact, AI has become a hot conversation, and one of the big trends in video marketing. The majority of tools available have an incredible power to elevate the production process, and either embolden creative ideas or at the very least, simplify the process.

Lambda Films is a video production company that has been fully embracing the opportunities afforded by AI, and so in this post we’ll be dissecting the future of AI video production by examining the tools and platforms available at each stage of the production process. 

Contents

How can AI be used in video production?

PRE-PRODUCTION

Ideation: ChatGPT

Scriptwriting: Scriptbook & Sudowrite

Concept Art: Midjourney

PRODUCTION

Text-To-Video: Pika Labs & Sora, Open AI

Video Generation: Synthesia & Microsoft VASA-1

Artificial Voice Over: Lovo.ai & ElevenLabs

POST-PRODUCTION

Adobe Premiere Pro AI Integration 

Colour Correction: Colorlab AI

Automated Captioning: YouTube

Lip Syncing & dubbing: HeyGen

Honourable Mention: Runway & EMO

ISSUES SURROUNDING AI VIDEO PRODUCTION

Ethical Concerns

Quality & Creativity Limitations

Bias & Representation

DRAWBACKS OF USING AI IN VIDEO PRODUCTION

Storyboard Creation

Scriptwriting with emotional depth

Understanding symbolism and thematic depth

WILL AI REPLACE VIDEO PRODUCTION?

How can AI be used in video production?

Pre-Production

Ideation

ChatGPT may prove less effective than the human mind when generating creative ideas. However, the strength of the platform is when it’s used as a sounding board, and a catalyst to kickstart the ideation process. To get the best ideas from ChatGPT, consider employing ‘user personas’. To get ChatGPT to adopt a user persona, first clearly define the persona’s characteristics and communication style, or ask it to base its personality on someone real (e.g Ryan Reynolds). Now, you’ll get more creative responses albeit aligned with a chosen artist or creative. 

Scriptwriting

AI-powered tools such as ScriptBook and Sudowrite can be used to analyse existing scripts, and generate new ones based on patterns and structures found in the source data. This can help streamline the scriptwriting process, generate new ideas or directions, or analyse your work for commercial viability. For corporate video production projects, these tools can analyse a huge amount of company material, and refine the key messages down to something workable. 

Concept Art

Text-to-image tools such as Midjourney are perfect for taking your idea, and through some simple text prompts, translating that idea in to visual artwork. This concept art is effective at selling in ideas to clients at the pitch stage.

AI Generated Concept Art for a Video Production by Lambda Films
Supporting AI concept art for a video project
AI Generated Concept Art for a Video Production by Lambda Films

Production

Text-To-Video

Pika Labs currently leads the way in pioneering the text-to-video capabilities. The online platform is able to produce small video clips as live action or animation, based purely on a descriptive sentence from the user. In the marketing & commercial space, videos made up from stock have been a regular creative approach. However, Pika Labs is opening the possibility of users creating their own home-grown stock footage, specifically tailored to their needs.

However, as of February 2024, OpenAI have announced Sora, their text-to-video platform. You can read our full guide to Sora for Video Production here. 

The results, as seen below, are impressive but OpenAI have been equally transparent about the current limitations of the platform; namely an understanding of physics, cause and effect and spatial difference. 

Like Midjourney, we would also argue that continuity will also remain an issue for anyone trying to use the platform to create a consistent and coherent sequence using characters, props or locations.

Where Sora will really thrive is in producing tailored stock footage for specific use cases within video projects.

Video Generation

Synthesia may be the closest thing we have to a complete, commercially viable AI video generation tool. The platform allows users to easily create e-learning and thought-leadership style videos using only a simple text script. Users can upload their script, and Synthesia will create a full AI human avatar to deliver your words as a full, realistic presentation. The technology and results are impressive, and if users have basic editing skills, then their AI presentation can be augmented with stock video, images and music to produce professional content.

The next level in video generation comes from Microsoft’s VASA-1. VASA-1, like Synthesia, has the ability to generate realistic human movement and emotion from a written script. However, the real power of VASA-1 comes from its ability to use a single still photo of a real person rather than an AI-generated avatar.

In production & business applications, this could be a huge step in generating thought-leadership, internal comms and training videos using a recognisable personality. However, the flip-side is the very real mis-use of this technology to create believable deepfakes and harmful content. Microsoft themselves stipulate that VASA-1 won’t be openly available until this technology will be used responsibly. So, never then?

 

Artificial Voice Over

It seems we’re slowly moving away from the robotic tones of Alexa & Siri, with tools such as Lovo.ai & Eleven Labs. Both these platforms provide a powerful text-to-speech generator that can create eerily authentic audio narrations for your video or animation.

You can use AI narration in the production process for draft voice overs to produce more accurate editing of drafts, and more realistic animatics for creative pitches.

Post-Production

Generative AI in Editing

In April 2024, Adobe announced that future iterations of Premiere Pro will have integrated support for generative AI tools such as Pika, Runway & Sora. This is a huge step in creating a seamless, AI-powered post-production workflow. The strength of this integration lies in the swift creation of AI generated b-roll & stock footage, image replacement and clean-up.  

Color Correction

AI-powered tools such as Colorlab AI use machine learning algorithms to analyze raw footage and automatically adjust color grading, making it faster and easier to achieve a desired look.

Automated Captioning

AI-powered speech recognition technology within platforms such as YouTube, can be used to automatically generate captions for videos, making them more accessible to a wider audience.

Lip-syncing & dubbing

Tools such as HeyGen are revolutionising the localisation process in video & commercial production. Previously, localisation would require dubbing or subtitle integration. With HeyGen, you can retain your original talent and the tool will automatically translate, dub and lip-sync a localised script in the talent’s original voice. 

Honourable Mentions

I’m including a mention of Runway, who are spearheading a whole suite of AI post-production tools from in-painting and out-painting, to video upres, automatic rotoscoping & seamless, interpolated slow motion.

EMO is a tool that is just so damn cool. Much like Microsoft’s VASA-1, EMO can take a still image of a person and generate realistic human emotion & movement, coupled with dubbing and lip-syncing. Where it differs however, is that EMO has a strength in generating humanism in response to song.

Issues surrounding AI in video production

Ethical Concerns

AI can create realistic images and videos, including deepfakes, which pose ethical challenges. These deepfakes can be used for misinformation, propaganda, or impersonating individuals, raising concerns about consent and the potential for misuse.

Quality and Creativity Limitations

While AI can efficiently produce content, there are concerns about its ability to match the creative nuances and quality that experienced human professionals bring. AI-generated videos might lack the emotional depth, artistic flair, and storytelling prowess that come from human creativity.

Bias and Representation

AI systems can inadvertently perpetuate biases present in their training data. In video production, this could lead to the creation of content that lacks diversity or reinforces stereotypes, impacting representation and fairness.

Drawbacks of using AI in video production

Storyboard Creation

Storyboarding is a crucial step where the visual narrative of a video is sketched out, scene by scene. It requires a deep understanding of the narrative, artistic vision, and the emotional flow of the story. AI currently struggles with these creative and interpretive aspects. It may assist in generating basic layouts or suggesting compositions based on existing templates, but it lacks the nuanced understanding of storytelling and visual symbolism that a human artist brings to a storyboard.

 

Scriptwriting with Emotional Depth

AI can generate scripts based on predefined parameters and data inputs, but it often struggles to infuse scripts with nuanced emotional depth and character development. Human scriptwriters draw from personal experiences, cultural nuances, and a deep understanding of human emotions, which are areas where AI lacks proficiency. AI-generated scripts may end up being formulaic or lacking in the subtleties that make stories compelling and relatable.

 

Understanding Symbolism and Thematic Depth

In video production, symbolism and thematic elements are often used to add depth and layers of meaning to the narrative. AI has difficulty understanding and implementing these subtle and abstract concepts. While it can identify and replicate patterns, the ability to creatively use symbolism and weave complex themes into a narrative is a distinctly human skill that AI has yet to master.

Will AI replace video production?

Overall, AI is transforming the world of video production in a variety of ways, from streamlining the pre-production process to optimizing production and enhancing post-production. While there are still many areas where human creativity and expertise are indispensable, the increasing use of AI is making video production faster, more efficient, and more engaging than ever before.

As the technology continues to evolve, we can expect to see even more exciting applications of AI in video production in the future. As for the answer to the big question…? We envisage a video production industry that is far more weighted to the post-production process where editors will cross with prompt engineers.

Platforms such as Sora & Pika Labs will provide a greater resource for filmmakers to produce small clips and visuals that push the creative production values of their work. However, until these platforms can create coherent sequences (as well as overcome other limitations) they will occupy the same space as traditional stock footage sites.