Updated
Seedream 5.0 Pro: photoreal images, and people you can then put in a video
Seedream is ByteDance’s image model, served through BytePlus, and 5.0 Pro is its current flagship. It is picked for three things: photorealism, a true 21:9 frame, and people who can go on to move in Seedance video, which refuses ordinary photos of people.
Below is the model itself, not a screenshot. Describe an image or upload your own: the answer arrives on this page, and the first attempt needs no sign-up.
What you are trying here
This is not a demo mode: the request goes to the same model the main chat uses. The only difference is that the conversation ends after the first answer — you can continue in chat, and the image travels with you.
A person who can then move in video
ByteDance’s video model, Seedance, refuses pictures of real-looking people at the door. There is one exception: faces that Seedream 5.0 drew from text within the last 30 days through the same BytePlus account — and on Twelver both models run through one. So a hero for a mini-series, an actor for an ad or a presenter is drawn here first, then handed to the video as it is.
Unedited, or it is refused
The exception covers the original file only. Once the portrait has been edited — here, in another model, or by blending it with a second picture — Seedance refuses it like any other photo. To change the person’s clothes or age, draw a new portrait from text. Your own photo of a real person does not become acceptable by editing it here either.
Eight ratios, including a genuine ultrawide
The range of shapes is the widest of the connected image models, and one of them is not available anywhere else: 21:9, a site header, a banner, or a frame for a video edit without stretching or filling in the edges.
| Shape | 1.5K (default) | 2K |
|---|---|---|
| 1:1 — square | 1536×1536 | 2048×2048 |
| 16:9 — wide | 2048×1152 | 2816×1584 |
| 9:16 — vertical | 1152×2048 | 1584×2816 |
| 21:9 — ultrawide | 2352×1008 | 3136×1344 |
| 4:3 and 3:4 | 1792×1344 | 2368×1776 |
| 3:2 and 2:3 | 1872×1248 | 2496×1664 |
Stay at 1.5K unless it is going to print
1.5K is the default for a reason: BytePlus charges it at the 1K rate, and it is enough for the web, a post or a video frame. Ask for 2K when the image is going on paper or a large screen.
Write the prompt in your own language
Version 5 reads prompts in Spanish, Russian, French, German, Portuguese and nine more languages besides English and Chinese. Describe the shot the way you would to a photographer; there is no need to translate it first.
Blending several sources
The blend mode takes two to four of your images and assembles one frame from them: the character from here, the background from there, the clothing from a third. The classic uses are redressing a character, transferring a style, and placing a product in a different setting.
Each request returns one image. When you want options, ask for several: each one is a separate request with its own variation of the prompt, and each is billed.
How Seedream differs from its neighbours
| Job | Which model |
|---|---|
| A person who will appear in a Seedance video, an ultrawide frame | Seedream |
| Holding a face or a product across photo edits | Nano Banana |
| Legible text inside the image, a transparent background | Ideogram |
| An image you will animate with the same family afterwards | Grok Imagine |
What Seedream will not do
- It does not guarantee lettering. For a caption, sign or poster that must read correctly, use Ideogram.
- It does not hold a person across edits. For that there is Nano Banana — but an edited face will not get into Seedance video.
- It does not go above 2K and returns one image per request.
- It will not take more than four sources in blend mode.
Frequently asked questions
What is Seedream and whose model is it?
Is Seedream 5.0 available?
Can I make a video with a person drawn in Seedream?
What resolution does Seedream 5.0 Pro give?
Which frame shapes does Seedream support?
Can I write the prompt in Spanish or Russian?
Can I blend several of my own images into one?
Do I need to select Seedream in the chat?
Examples and prices are on the Seedream page and in image generation. What we connected and when is in the release post. Other models are in the book’s table of contents; images in general are in the image generation book.
Try it yourself
Everything in this guide runs inside Twelver
One chat for text, images, video, music and voice — no separate services or subscriptions.
Open Twelver chat