How to generate videos with Symphony Creative Studio
Symphony Creative Studio offers three powerful video generation capabilities that turn simple inputs into high-quality, TikTok-ready video clips in minutes:
Text to video: Generate videos entirely from text descriptions
Image to video: Animate static images into dynamic video clips
Reference to video: Create videos using multiple reference inputs (images, videos, audio)
Learn more about Symphony Creative Studio.
Before you begin
Log in to Symphony Creative Studio with your TikTok for Business account. If you do not have an account, create your TikTok for Business account. Learn more about TikTok for Business accounts.
If it's your first time accessing Symphony Creative Studio, you'll need to:
Complete your profile.
Accept the Creative GenAI Terms.
How to generate videos with Symphony Creative Studio
Access video generation
Log in to Symphony Creative Studio on your desktop.
Hover over the left panel and click Create.
Select your generation mode
Choose from the following options in the dropdown:
Reference to video
Upload multiple images and/or videos as reference material.
Add a text prompt as additional instruction for video generation.
Text to video
Type a text prompt describing the scene you want to create
Click Enhance to automatically make your prompt more detailed and optimized for better results
Optional: Upload a reference image or short video (less than 10 seconds) to extract a descriptive prompt. Note: Visuals from the reference will not be included in the final video.
Image to video
Click [+] to Upload image or Select from asset library.
Choose an image featuring your product, model, or desired scene.
Enter a text prompt to describe the specific movement or animation you want added to the image.
Configure generation settings
Select your preferred AI model from the dropdown.
You may choose the Dreamina Seedance 2.0 AI model. Learn more about TikTok Symphony and Dreamina Seedance 2.0. Note: Dreamina Seedance 2.0 restricts certain capabilities in this initial rollout, including the ability to use images or videos containing real faces as references for video generation. Please select a different model for this instead.
Select your preferred video duration.
Generate your video
Click the up arrow button to start generation.
Typical generation time is a few minutes depending on mode and settings.
Preview and manage generated assets
Click the Library icon located in the left panel.
Hover over generated videos to preview.
For individual clips:
Click [...] in the top-right corner of the video to Open in editor, Delete, or view more options.
Click Download to save to your computer.
Click Sync to Ads Manager to export directly to your TikTok Ads Manager account.
For batch operations:
Click Select multiple in the top-right corner of the Library page and tick your desired clips' checkboxes.
Click Export at the top to batch download or sync to Ads Manager.
Click Save project at the top to save your work for future editing.
Edit your video (optional)
After selecting Edit, you can access the following features:
Assets: Choose assets from your uploads, existing generations, or stock library.
Audio: Choose from trending music, recommended music, playlists, sound effects, favorites, or click Pick music for me.
Voiceover: Edit scripts, translate to different languages, select voices, generate captions, and add lip sync.
Captions: Edit, add, or remove captions.
Avatars: Select and add avatars to your video.
Text: Choose from default text styles or templates.
Effects: Add effects to specific segments of your video.
Stickers: Add stickers to specific segments of your video.
Use the timeline editor to adjust the position and style of all assets
Export your final video
Once you're finished editing, click Export.
Choose to Download to your computer or Sync to Ads Manager.
Note: An AI-generated label is added to all exported videos per TikTok ad policy. All generated content will undergo further moderation and policy reviews before being cleared for publishing on TikTok
Tips for prompting video generation
During the video generation step, you can enter text in the Enter a prompt textbox to customize your creative and motion preferences. Here are some tips to help optimize your results.
Keep it simple. Use descriptive and straightforward language. Complex or overly wordy prompts can lead to less accurate results. Clear and concise prompts are more likely to generate high-quality videos.
Do: "A person uses a new fitness app on their phone."
Don't do: "In a rather elaborate and convoluted way, there should be a person who is kind of doing something with a thing that might be a product but it's not really clear and it's all happening in a place that is sort of like an environment."
Avoid conversational or system-style prompting. For best results, avoid using dialogue prompts in a back-and-forth manner. You should directly state what you want to see in the video without assigning roles or referring to previous generations.
Do: "A person runs on a treadmill in a gym."
Don't: “Make the logo bigger instead of smaller.” (referring to a prior generation)
Stick to a single scene. Avoid scene transitions or multi-scene instructions. If you need multiple scenes, generate them separately and then combine them using a video editor.
Do: "A person runs on a treadmill in a gym."
Don't: "First, a person is at home, then they go to the gym, and then they are at a park."
Do not reference celebrities. Prompts that include the name or likeness of public figures are not supported and will be rejected.
Avoid split-screen or picture-in-picture formats.
Do: "A person uses a new product in a modern office."
Don't: "Show a split-screen with a person using a product on one side and a graph of sales on the other side."
If a generation does not work for you, try regeneration. Repeat the same prompt before changing your prompt. This can sometimes resolve issues without having to come up with a new, more complex prompt.
Do: If the first generated video of "A person uses a new fitness app on their phone" looks blurry or doesn't capture the right mood, try regenerating it with the same prompt before adding more details.
