Dreamina 3.1
Polished, print-ready stills when the brief is a finished image rather than a sketch.
All modelsOverview
Dreamina 3.1 is an image generation model developed by ByteDance, the parent company of TikTok and CapCut. It is built to produce high-resolution, 4-megapixel outputs with a focus on cinematic lighting, nuanced reflections, and accurate text rendering. As part of ByteDance's generative media ecosystem alongside models like Seedream 4.0, it is useful for commercial design, poster creation, and generating photorealistic scenes that require precise style control.
Specifications · Text to image
- Input mode
- Text to image
- Aspect ratios
- 16:9, 9:16, 1:1, 4:3, 3:4, 16:10, 10:16, 3:2, 2:3
- Resolutions
- 540p, 720p, 1080p
Pricing
Build with this model: the Dreamina 3.1 API on the Hedra Developer Platform.
Dreamina 3.1 API →Real output · generated with Dreamina 3.1
Best of Dreamina 3.1
Prompting
Prompt tips
- Quote your text: For accurate typography, place the exact words you want rendered inside quotation marks.
- Specify lighting and architecture: Use terms like "natural window lighting," "clean geometric details," or "balanced composition" to leverage its strengths in architectural and fantasy scenes.
- Define the medium: Explicitly state the desired format (e.g., "commercial photography," "cinematic film frame," or "3D-style graphics") to lock in the aesthetic.
- Bilingual prompting: Take advantage of its native understanding of both English and Chinese prompts for nuanced cultural or regional details.
About the model
Questions, answered
What is Dreamina 3.1 best for?
Dreamina 3.1 by ByteDance is a cinematic image generation model that excels at producing high-resolution, photorealistic visuals. It is particularly strong at rendering nuanced lighting, sharp text, and complex textures like glossy reflections and fabrics. Creators frequently use it for luxury product photography, high-fidelity cinematic scenes, and polished social media assets.
When was Dreamina 3.1 released and what is its lineage?
ByteDance released Dreamina 3.1 in July 2025 as a major update to its Dreamina (known as Jimeng AI in China) and CapCut creative suites. It succeeded earlier models like Seedream 3.0 and was later superseded by the highly advanced Seedream 4.0 in September 2025, followed by Seedream 4.5 and Seedream 5.0 Lite.
How can I get the best results from Dreamina 3.1?
To get the best results from Dreamina 3.1, use detailed prompts that specify material properties, directional lighting, and camera framing. If you encounter issues like plastic-looking reflections or unnatural fabric drapes, follow the community's iterative image-to-image workflow: generate a base 2K image, upload it as a reference, and adjust your prompt to focus strictly on texture correction (e.g., "enhance visible linen weave texture"). For more techniques, consult official Dreamina tutorials.