AI video generators are revolutionizing the way we create content, offering powerful tools for both professionals and hobbyists. After spending some time on waitlists, I finally gained access to OpenAI’s Sora Video Generator and Adobe’s Firefly Video Generator. Let’s dive into their similarities and differences and put them to the test.

Interface & Features

Both tools offer text-to-video generation of 5 second video clips. While Firefly caps clips at 5 seconds, Sora offer 5, 10, 15, and 20 second clips based on your plan level.

Sora

Sora offers a simple but powerful interface.

• The “+” button allows for user uploaded video or image assets.
• The preset button contains a list of preset styles that can be used to apply to your video, which includes papercut, stop motion, film noir, and even ballon world. Users can create their own presets by describing things like Camera, Lighting, Colors, and Content transformation. Reference images and video can also be uploaded to enhance presets.
• Aspect ratio options include 16:9, 9:16, and 1:1.
• Resolution options include 480p, 720, and 1080p with plan and credit amounts varying depending on selection.
• Clip length option include 5, 10, 15, and 20 seconds depending on plan.
• Variations allows the user to select how many versions of the prompt are created, with the options including 1, 2, and 4.
• The “?” shows the amount of credits the current configuration will cost.

Story Board

The storyboard feature allows the user to fine-tune their composition by using text, image, or video to breakdown their video into frames along the timeline and get more detailed about what is created and at what point of the video.

Post Generation Options

Once your video is created, Sora offers more options to edit and refine your vision.

• Edit Prompt allows you to edit your text prompt and visual settings to create a new video.
• View Story brings up the story board and allows for fine tuning.
• Re-cut brings up the Storyboard and allows for trimming and extending of the clip.
• Remix uses the existing video as reference and allows the user to describe changes that will be used to create a new video.
• Blend creates a transition from the current video and another video from your library.
• Loop creates a looped video from a user-defined segment of the existing video.

Export Options

Once complete, videos can be exported as either .mp4 video with watermark (watermark removed at certain plan levels), or animated GIF.

Firefly

Firefly also offers a simple interface but with more preset options presented in a visual way.

We will be testing Sora with OpenAi’s Team plan and Firefly with Adobe Creative Suite.

General Settings

• Model is the version of Firefly in use and at this point, beta is the only option.
• Aspect ratio allows for 16:9 and 9:16 options.
• Frames per second defaults to 24 FPS and is the only option with more likely to be added in the future.

Camera Settings

• Shot size allows for preset framing such as Close Up or Medium Shot.
• Camera Angle controls the placement of the camera such as Arial, Low angle, or Top-down shots.
• Motion is a collection of preset camera motions that can be applied to the generated video.

Advanced Settings

Seed: From Adobe, a “seed” is a numerical value that acts as a starting point for the AI generation process, essentially controlling the randomness involved, allowing users to generate similar images by using the same seed across multiple requests, effectively fine-tuning the output to achieve a desired aesthetic or style.

Output Options

Once complete, videos can be exported as either .mp4 video.

Feature Comparison

Video Output Test

I wanted to see how the output compared, so I ran a simple test. For the purposes of this trial, the same prompt was used in each app and the default settings were used. No other manipulation occurred.

Prompt

A happy German Sheppard playfully running in a grassy park; towards the camera; Sunny day, bright sky with puffy white clouds.

Sora

Firefly

Takeaways

I was impressed with both videos and especially thought the dog looked pretty good in both. For me, the Sora video came out a little bit better. I liked the motion it added without prompting and I felt the overall composition from the dog to the grass and trees, even the lighting, was more natural than the video from Firefly. While Firefly did a nice job with the dog, the background and composition felt less cohesive to me.

Although this was just a single prompt with no other setting changes, I think this shows what each of these tools is capable of. I’m confident that through further manipulation, both of these tests could be improved past their initial state, which is exciting.

Wrapping Up

Both tools show a lot of promise and will have a place in any creative’s toolbox.

Firefly’s simple UI and integration with other Adobe products such as Premiere and Express make it very easy to include in existing workflows for Adobe users.

Even as a big Adobe fan and user, at this time I think I’m leaning towards Sora if I had to pick a winner right now. I really liked the added interface options like Storyboard, the ability to make longer clips, and maybe the most exciting feature for me is the Loop feature, a feature I had hoped to see in Firefly. I also thing Sora had a slightly better video output test.

Text-to-video generators will continue to improve, and as they do, we will continue to explore and test their capabilities. Look for more on these tools in the future.