10 favorite AI video platforms for creative agencies and how they’re using them

Creative agencies are reaching for a growing swath of AI video tools—those developed by tech stalwarts such as Google to those from lesser-known platforms from China—that enable faster and cheaper ways to edit shots, pitch concepts and generate entire ads. The video tools sweeping the ad industry are all being judged for their pros and cons to see where they fit best in the creative process, according to advertisers. And the video-generating models demonstrate unexpected differences that agencies need to understand to use them appropriately, according to ad leaders at VaynerMedia, VML, Highdive and others. Video generator Kling, for example, is a top choice among agencies for its ability to portray realistic motion—an objective that has proved difficult for other models. Meanwhile, OpenAI’s Sora has been relatively underwhelming despite its powerful name recognition, experts told Ad Age. The fact that marketers are experimenting with both well-known and obscure video tools is a testament to how enthusiastic parts of the community are about incorporating the latest AI into their creative toolkits. Also, new generative video platforms crop up all the time, providing agencies with more opportunities to update their strategies. “This year we will achieve high-quality, high-resolution, lifelike video,” said Jason Zada, founder of production company Secret Level, which is using a mix of AI video tools, including China-based Kling, Luma Labs’ Dream Machine, Google’s Veo 2 and Chinese-owned platform Hailuo.More: Agency leaders share AI prompting lessons for marketers Ad Age spoke with 11 agencies to understand the respective strengths and weaknesses of some of the most high-profile video makers. Here are their 10 favorites and how creatives in advertising are using them. Kling More than half of the agency insiders who spoke with Ad Age referenced Kling, the AI platform developed by Chinese tech company Kuaishou. Kling, which includes both a text-to-video and image-to-video tool, burst onto the scene last summer when it released its first internationally available version. Text- and image-to-video means that advertisers can prompt the AI by chatting with it through text or by uploading an image. Peter Steiner, group creative director of research and development at Code and Theory, said Kling’s pricing is cheaper than well-known platforms such as Sora and Runway. The relatively low cost has allowed creatives at Code and Theory to experiment without having to worry about burning through their allotted credits, Steiner said. The price is crucial because creatives need to keep prompting video generators to get the desired result, and each prompt costs money, Steiner said. “The best technique for generative AI is still getting as many iterations as possible,” Steiner said. “One hundred generations beats a good prompt any day.” Miniature people doing some work. Clips generated with Kling AI. pic.twitter.com/eBIqyGW1Lz— ILikeToasters (@LikeToasters) February 20, 2025 Design studio Addition has used Kling as its go-to platform for turning images into video. “It is ideally suited for tasks that involve bringing motion to existing art or combining video models with fine-tuned image models,” said Paul Aaron, Addition’s CEO. Secret Level has also used Kling for its image-to-video capabilities, especially for depicting motion that is lifelike, as opposed to the cartoonish, perhaps even surreal movement that AI often portrays, Zada said. Kling offers both a subscription-based plan starting at $10 a month, with a capped number of credits, and a credits-based plan at $5 per 330 credits. Credits are the currency by which users pay for AI-generated content; the number of credits necessary to produce content depends on both the platform and the kind of content being produced. For example, text-based content typically costs just a few credits, while short videos can cost multiple dozens of credits. Veo 2 by Google Gemini Veo 2 is the second version of Google’s Gemini-powered text-to-video AI platform, which debuted with limited availability in December. Earlier this month, the tech giant launched Veo 2 on YouTube Shorts, and marketers have been testing the model. “It’s the most reliable tool for generating high-fidelity visuals with excellent prompt adherence and cohesive outputs,” said Tyler Cohen, creative director of experience at agency Tool. Cohen has posted his Veo 2 experiments to social media. Tool has used Veo 2 to craft commercials, including a local ad for a coffee shop in Los Angeles that is set to be released next month, Cohen said. Cohen declined to share the name of the brand. PJ Pereira, founder and creative chairman of Pereira O’Dell, said he was also impressed by Veo 2’s realistic production, such as how it accurately portrayed movement. Addition has been using Veo 2 to help create a bank of videos to deploy for content that can be personalized to consumers, said Aaron. Videos of this kind, as well as other custom content, have been a primary AI use case for marketers, such as Carvana and Honda. While Google has not yet revealed any subscription-based plans for Veo 2, it offers a cost-per-second plan—each second of AI-generated video from Veo 2 costs 50 cents. More from Ad Age: AI marketing glossary—key terms to know Runway Runway has been a leading AI video platform for marketers for several years now, and was one of the first to strike a content partnership deal with a Hollywood studio. Code and Theory has relied heavily on Runway to create projections for experiential installations because it excels at generating short, dramatic and surreal clips, Steiner said. “Their enterprise tools are really well developed too, so it makes it easy for a large team to collaborate,” Steiner said. Highdive has also leveraged Runway’s strength in abstraction to produce insert shots—shots filmed on a subject from a different angle for the purpose of focusing the viewer’s attention—said Larissa Berringer, managing director of 1986 Studios, Highdive’s in-house production studio. Runway’s image-to-video tool has a strong ability to maintain visual consistency while transitioning between frames, said Brian Yamada, VML’s chief innovation officer. Consistency is a key factor because advertisers need their products, brands and logos to maintain the same look throughout a commercial. Pereira O’Dell has used Runway’s out-painting feature, which expands images by filling them in with AI-generated content. For instance, Runway can change the aspect ratio of videos, turning a vertical video into a horizontal one, by adding the content in the wider frame, Pereira said. Runway offers a subscription-based plan at $15 per month with a capped number of credits. Playing with AI is the best way to gain a creative edge (from 2023) Sora Sora is perhaps the most well-known generative AI platform on this list by virtue of its developer, OpenAI, which also runs ChatGPT. Sora, its text-to-video tool, launched a year ago to widespread acclaim from the ad industry, and it became publicly available in December. Marketers including Toys R Us have already deployed Sora to create entire ads, although these efforts garnered mixed reviews from the creative community. According to Pereira, Sora is particularly helpful for creating multiple variations of a scene. The feature responsible for this is called Remix, which allows users to easily replace elements within a scene without having to re-prompt the system from scratch. “In general, it’s a very good tool to just sit there and let ideas flow,” Pereira said. Loving the abstraction that's possible with sora remix and blend features. pic.twitter.com/6YJqHxMIXH— Justin Maynard (@JADM823) January 17, 2025 However, Sora also has large expectations to meet, and some creatives have been disappointed with its work. For instance, Callum Gill, head of strategy at agency Rehab, said that Sora has often deviated too far from his original prompts, taking creative liberties and displaying videos that were less precise as a result. It’s unclear why some platforms may do this over others, but differences in training likely play a role. Also, Sora is not ideal for fast-paced campaigns that require quick turnaround because its rendering time is relatively slow, said Kristian Bottini, CEO of 270B, the agency that helped Avocados From Mexico design its AI activation timed to the Super Bowl. Runway also renders videos slowly, Bottini said. Secret Level’s Zada said Sora’s videos “have always looked good, but not great.” The videos have not felt cinematic, which is what filmmakers and the ad industry need, Zada said. Sora offers two subscription-based plans that are integrated with ChatGPT; one costs $20 a month and the other $200 a month. Both plans have a capped number of credits, although the latter has a far higher cap. Recent news: How X’s AI-generated ads for marketers will work Hailuo Hailuo is a model offering both text-to-video and image-to-video capabilities. It was launched last March by MiniMax, an AI company based out of China. Pereira O’Dell has found that Hailuo offers high-quality camera controls and is ideal for depicting real people, Pereira said. The agency has used Hailuo to create shots of actual athletes for a campaign it is creating for a sports-related client. Hailuo is also adept at adding complex movements in image-to-video productions, said VML’s Yamada. VML is using Hailuo as a tool for storyboarding and post-production, Yamada said. Made with Hailuo I2V Director🎥Images: Midjourney // --sref 3081481455📸Music: Udio🔊 pic.twitter.com/P8hQzIzH4L— saltxdiamond (@saltxdiamond) February 24, 2025 Hailuo offers a subscription-based plan starting at $15 per month with a capped number of credits. Dream Machine Dream Machine is an AI platform developed by Luma Labs and launched to the public last summer. Its video generator is called Ray 2, which offers both text-to-video and image-to-video capabilities. Pereira O’Dell has found value in Dream Machine’s Brainstorm feature, which offers AI-generated creative suggestions based on the visuals the user is crafting, Pereira said. The agency also used Dream Machine to generate background scenes for the work it is doing with its sports-related client, Pereira said. Luma Photon and Ray2 are fantastic with anime style scenes!This is an image-to-video process.The images came out so good, I didn't have to even include a prompt for the video. pic.twitter.com/Phk6wEg5rb— Jerrod Lew (@jerrod_lew) February 14, 2025 VML and Rehab are also experimenting with Dream Machine’s video capabilities, they told Ad Age. Dream Machine offers a subscription-based plan starting at $10 per month with a capped number of credits. HeyGen HeyGen is a specialized AI video generator that creates digital avatars, as well as offers capabilities in speech translation. 270B has primarily used HeyGen to create digital spokespeople to strengthen pitch ideas, Bottini said. HeyGen played an early role in the agency’s development of an avatar representing Rob Gronkowski for Avocados From Mexico’s recent campaign. The model proved to 270B that an interactive replica of a real person was a feasible idea for a campaign. “It’s a great way to simulate interactive experiences before fully developing them,” Bottini said. AI brings historical figures to life in 2025Audrey Hepburn is a fashion influencer, Albert Einstein is giving a Ted Talk- wait what year is it? Comment down below what you think AI missed or who you want to see next. pic.twitter.com/pOkrhYBBcX— HeyGen (@HeyGen_Official) February 19, 2025 Bottini also praised the model’s ability to easily integrate different languages. HeyGen currently offers AI-generated voice for more than 175 different languages, according to the company’s website. HeyGen offers a subscription-based plan starting at $24 per month. Also read: What the DeepSeek AI frenzy means for marketers ComfyUI While not an AI model itself, ComfyUI is a free interface that agencies are using to access multiple video models in the same place. Rehab, for example, uses ComfyUI to experiment with Sora, Runway, Luma and HeyGen, Rehab’s Gill said. The real benefit of ComfyUI is that it offers the infrastructure to build custom workflows that intertwine multiple AI tools. The alternative is having to constantly leave an AI model’s ecosystem in order to access another model, thus losing one’s work in the process. Through ComfyUI, an agency can create a de facto hub for its AI experimentation. Prompt: A market scene in full swing on a sunny day, with vendors selling fresh produce under colorful canopies. The fruits and vegetables—tomatoes, oranges, and leafy greens—pop in vibrant reds, oranges, and greens. The sunlight shines down on the stalls, creating sharp… pic.twitter.com/NFSV62qxi4— ComfyUI (@ComfyUI) December 28, 2024 270B has used ComfyUI for prototyping content, fine-tuning animations and blending AI-generated material with work from traditional production, Bottini said. Creative agency Supergood (an arm of Supernatural) has also used ComfyUI to run models in a more controlled environment, said senior designer Aarman Roy. DaVinci Resolve DaVinci Resolve is also not a traditional AI video generator—it’s a post-production video editing platform that has existed for over 20 years. Nonetheless, newer versions of Davinci have rolled out with AI capabilities that have impressed Highdive. The platform is a suite of post-production tools developed by Blackmagic Design. Highdive’s Berringer praised the platform’s ability to combine AI video generation with professional editing features. Superscale, for example, uses AI to create new pixels when increasing the resolution of an image. IntelliTrack follows moving subjects and automatically adjusts the sound according to their position in the scene. This is incredible!DaVinci Resolve, the Hollywood color grading software, just released a brand new AI motion tracking tool.With just one click, you can effortlessly track even the fastest moving objects. It’s free to use, link in comments!Here’s how to make it work: pic.twitter.com/wMBGchQOBm— el.cine (@EHuanglu) November 25, 2024 Agencies want more of these kinds of tools that blend traditional aspects of video editing with generative AI. Code and Theory’s Steiner told Ad Age that he wanted to see AI platforms incorporate aspects such as basic motion graphics and compositing tools. “Handcrafted touches like titles go a long way towards making content more human and engaging," Steiner said. DaVinci Resolve has a one-time fee of $295. Adobe Firefly Adobe Firefly is another AI generator well-known to many marketers. Production company Studio Rx used Firefly to design its rebrand last summer, and both Deloitte and IBM have used it for their own public-facing creative, as well. In September, Adobe brought video capabilities to Firefly, and earlier this month rolled out a new video generator inside the platform, dubbed Generate Video. Vayner is using Firefly for numerous creative aspects, including ideation, concept presentations and developing different versions of an asset for different implementations, said Mimia Johnson, executive creative director at Vayner. Pereira O’Dell has also taken advantage of Generative Extend, a Firefly-powered feature that allows users to lengthen the duration of a video scene by generating new frames, Pereira said. “Talk about a killer feature that is super helpful in every single project,” Pereira said. Adobe Firefly offers a subscription-based plan starting at $10 per month, with a capped number of credits. Become an All Access subscriber Get access to exclusive resources including Leading National Advertisers, World’s Largest Advertisers, Agency Report, Agency Review Tracker and TV Pricing Report. Click here

10 favorite AI video platforms for creative agencies and how they’re using them

Featured Stories