Yovid

Seedream 5.0 Prompts: 30 Image Prompts to Copy

NoraNora18 min readPublished on Nov 20, 2026
Share on X

Direct answer: Seedream 5.0 prompts work best as one or two plain sentences: name the subject, what it is doing and where, then add the style, the light, the composition and what the image is for. ByteDance's prompt guidance for its Seedream models favors coherent natural language over keyword lists, asks you to put any words that must appear in the image inside double quotation marks, and, for edits, to say exactly what changes and what stays the same. Seedream 5.0 comes in two versions: Pro, built for text-heavy layouts, multilingual text and precise edits at up to 2K, and Lite, which renders at 2K to 4K and returns up to 15 images from one prompt. Below are 30 copy-ready prompts in five groups: realistic portraits, illustration and Chinese-style art, products, scenes and architecture, and stills meant to become the first frame of a video.

Last verified: 2026-11-20

What Seedream 5.0 Pro and Lite are good at

Seedream 5.0 is ByteDance Seed's image model family. Seedream 5.0 Lite came first, in February 2026. Seedream 5.0 Pro followed in July 2026, and ByteDance introduced it as a model that understands design: it plans the layout of an infographic, a menu or a storyboard before it draws, and it can edit an image from a short instruction.

According to ByteDance's announcement, Pro takes prompts and renders text in Chinese, English and more than ten other common languages, including French, German, Russian, Japanese, Korean, Spanish and Arabic. It follows each script's rules, so Arabic runs right to left and Spanish keeps its accents. ByteDance also says fine-grained text rendering and pixel-level editing consistency still have room to improve, so small print needs a proofread.

Seedream 5.0 Pro Seedream 5.0 Lite
Output sizes 1K, 1.5K or 2K 2K, 3K or 4K
Images per prompt 1 1 to 15
Reference images Up to 10 Up to 14
Aspect ratios 11 presets from 21:9 to 9:21, auto, or exact pixels Same presets, auto, or exact pixels
Strongest at Text-heavy layouts, multilingual text, precise edits Large output and batches of variations
  • Best for: posters, menus and infographics with readable headings, multi-panel comics and storyboards, product shots, and edits where most of the picture must stay put.
  • Not for: long paragraphs of small print you can't proofread, exact charts or engineering figures, and logos or faces you haven't supplied as a reference.

The Seedream 5.0 prompt formula

Subject + action + setting + style + light + composition + purpose

  • Subject: who or what, with one or two concrete details. "An elderly jazz trumpeter in a gray suit" beats "a musician".
  • Action: what the subject is doing. Even a still image reads better with a verb.
  • Setting: where it happens and what sits behind the subject.
  • Style: a photo, an ink painting, a flat vector illustration, a comic page. Name the medium the way you would brief an artist.
  • Light: the direction, the softness and the time of day.
  • Composition: the camera angle, where the subject sits in the frame and the aspect ratio.
  • Purpose: what the image is for, such as a product poster, a profile photo or a picture book page. BytePlus's Seedream prompt guide, written for Seedream 4.0 and 4.5 and still the vendor's general advice for the family, recommends stating the image's purpose and type whenever you have one in mind.

The same guide contrasts a full sentence, a girl in a lavish dress walking under a parasol along a tree-lined path in the style of a Monet painting, with a list of the same nouns, and recommends the sentence. It also notes that newer Seedream models understand prompts well enough that concise, precise wording usually works better than piling on ornate adjectives.

Every prompt below ends with an aspect ratio. BytePlus's Seedream 5.0 Pro tutorial explains that when you pick a resolution tier instead of exact pixels, the model reads the shape or purpose described in the prompt to set the final width and height. Naming the ratio keeps the prompt working whether your ratio setting is on auto or a fixed preset.

Which language should you write Seedream 5.0 prompts in?

Both versions read English and Chinese. Pro also accepts prompts in the other languages ByteDance lists, and ByteDance says it matches the cultural context of the language you write in. ByteDance's Lite announcement doesn't list prompt languages, so if you want to write in French, Russian or Arabic, use Pro. All 30 prompts here are in English so they run on either version.

How do you get readable text in a Seedream 5.0 image?

Put the exact words in double quotation marks, keep them short and say where they go: "the headline "SUMMER SIP" in bold white letters across the top". BytePlus's guide gives the same advice, with a poster titled in quotes as its recommended example. State the kind of image too, because "a poster" or "a menu" tells the model to lay the text out as a design rather than scatter it. Proofread the result before you use it, especially numbers and small labels.

How do you write a Seedream 5.0 edit prompt?

Name the exact element, say what should change and list what must not change. "Dress the tallest panda in a pink Peking Opera costume, keeping its pose unchanged" is the kind of instruction BytePlus recommends; "put that one in pink clothes" is the kind it warns against. With several uploads, refer to them as Image 1, Image 2 and so on, and say what to take from each. ByteDance adds that boxes, arrows and simple doodles drawn on the uploaded image help point the model at the right area when words alone aren't precise enough.

Prompt

Replace the coffee cup on the table in Image 1 with a glass of iced matcha latte, keeping the table, the hands, the window light and the background exactly the same.

Seedream 5.0 prompts for realistic portraits

A ceramicist in her sixties with silver hair and a clay-dusted linen apron holds a freshly thrown stoneware bowl in a sunlit pottery studio

Image prompt: Photorealistic environmental portrait of a ceramicist in her sixties standing in a sunlit pottery studio, silver hair tied back, a linen apron dusted with dry clay, holding a freshly thrown stoneware bowl at chest height and looking calmly into the lens. Soft window light from the left, shelves of unglazed pots softly blurred behind her, 85mm lens look, shallow depth of field, natural skin texture with fine lines, warm muted earth tones, 16:9 frame.

Portraits look real when the prompt gives the person a job, a place and a light source. "Natural skin texture" and a lens description steer away from the airbrushed look.

Prompt

Close-up portrait of a young street food cook in Bangkok at night, wiping sweat from her brow beside a smoking wok, neon reflections on her face, handheld 35mm documentary look, shallow depth of field, natural skin texture, 4:5 vertical frame.

Prompt

Studio headshot of a man in his forties with a salt and pepper beard and a navy knit sweater against a warm gray backdrop, soft large key light from the front left, gentle fill, crisp focus on the eyes, calm confident expression, professional profile photo, 1:1 square frame.

Prompt

Black and white portrait of an elderly jazz trumpeter resting his instrument on his knee in a dim backstage corridor, a single bare bulb above him, deep shadows, visible film grain, quiet reflective mood, 3:4 vertical frame.

Prompt

Golden hour portrait of a teenage girl laughing on a path through a wheat field, wind lifting her hair, sun flare behind her shoulder, warm backlight with soft fill on her face, 85mm lens look, candid lifestyle photography, 3:2 horizontal frame.

Prompt

Environmental portrait of a firefighter in his thirties sitting on the back step of a red fire engine after a night shift, soot on his cheeks, helmet in his lap, early morning blue light, honest editorial photography, 16:9 frame.

Prompt

Family portrait of three generations, a grandmother, a mother and a small boy, sitting close together on a green sofa in a sunlit living room, everyone looking at the camera and smiling, soft window light, natural colors, 4:3 frame.

Seedream 5.0 prompts for illustration and Chinese-style art

Seedream comes from ByteDance and handles Chinese painting styles and Chinese lettering well. Name the technique precisely: ink wash, gongbi, flat vector, watercolor picture book or black-and-white comic all give very different results.

Prompt

Chinese ink wash painting of a lone scholar in a pavilion on a misty mountain lake, pine branches in the foreground, generous empty space in the upper half, soft gray gradients on rice paper texture, traditional guofeng style, 9:16 vertical frame.

Prompt

Gongbi style illustration of a red crowned crane standing among white lotus flowers, fine outlines and delicate mineral colors, gold leaf accents on a pale silk background, elegant traditional Chinese composition, 3:4 vertical frame.

Prompt

Spring Festival poster in modern guofeng illustration style, a girl in a red hanfu holding a lantern under blossoming plum branches, flat colors with subtle paper grain, the large title "新春快乐" in gold calligraphy at the top, 2:3 vertical frame.

Prompt

Children's picture book illustration of a small fox and a hedgehog sharing an umbrella on a rainy forest path, soft watercolor washes, warm muted palette, gentle expressions, plenty of leaf detail, 4:3 frame.

Prompt

Six panel comic page about a cat who tries to catch a goldfish and ends up falling asleep beside the bowl, clean black ink lines with light gray tones, clear panel borders, simple expressive faces, 3:4 vertical frame.

Prompt

Flat vector illustration of a cozy reading corner with a yellow armchair, a floor lamp, a cat asleep on a stack of books and a window showing a city at night, limited palette of mustard, teal and cream, clean geometric shapes, 1:1 square frame.

Seedream 5.0 product prompts

An unbranded sage green ceramic pour-over dripper on a glass carafe, coffee streaming into the carafe, on a walnut board with scattered coffee beans in morning light

Image prompt: Studio product photograph of an unbranded matte sage green ceramic pour-over coffee dripper on a glass carafe, a thin stream of coffee falling into the carafe, standing on a dark walnut board with a few roasted beans scattered in front. Soft morning window light from the right, a wisp of steam rising from the dripper, warm cream background, gentle shadow to the left, sharp focus on the glaze texture, blank surfaces with no logos, 16:9 frame.

Product prompts need the material, the surface and the light more than adjectives. Keep products unbranded in templates, and put real packaging copy in quotes only when you want the model to render it.

Prompt

Studio product photograph of an unbranded clear glass perfume bottle with a faceted stopper on a slab of pale pink marble, a single peony beside it, soft diffused light from the left, crisp reflections, 4:5 vertical frame, blank bottle with no label.

Prompt

Flat lay of an unbranded brown leather wallet, a brass key ring, a fountain pen and a pocket notebook on dark slate, top down angle, soft even light, rich brown and gold tones, 1:1 square frame, no logos.

Prompt

Hero shot of an unbranded white running shoe floating at a slight angle above cracked desert ground, a small burst of sand beneath the sole, hard sunlight with a strong shadow, bright sky blue background, 16:9 frame, no logos.

Prompt

Ecommerce main image of an unbranded stainless steel insulated water bottle in matte forest green on a seamless white background, condensation droplets on the surface, soft studio lighting, centered composition, 1:1 square frame, plain surface with no text.

Prompt

Product poster for a lemon iced tea can on a bright yellow background with lemon slices and ice cubes splashing around it, the headline "SUMMER SIP" in bold white letters across the top, energetic advertising style, 4:5 vertical frame.

Prompt

Lifestyle photograph of an unbranded wireless speaker in light oak and gray fabric on a kitchen shelf beside a potted herb and a stack of cookbooks, warm morning light, shallow depth of field, 3:2 frame, no logos or text.

Seedream 5.0 prompts for scenes and architecture

Wide scenes need a clear viewpoint. Say whether it's eye level, aerial or isometric, and give the light a time of day.

Prompt

Architectural photograph of a modern concrete house with floor to ceiling windows on a forested hillside at blue hour, warm interior lights glowing, a reflecting pool in front, two point perspective with straight verticals, 16:9 frame.

Prompt

Interior design render of a Japandi living room with a low linen sofa, a round paper lamp, an oak coffee table and a large window onto a bamboo garden, soft afternoon light, calm neutral palette, eye level wide view, 3:2 frame.

Prompt

Rainy night street in an old European town, wet cobblestones reflecting shop windows and a single red tram passing in the distance, people with umbrellas, cinematic teal and amber color grade, 21:9 ultrawide frame.

Prompt

Aerial view of terraced rice fields filled with water at sunrise, curving green and silver bands across the hills, a small village with tiled roofs at the bottom edge, light morning mist, photorealistic drone photography, 16:9 frame.

Prompt

Cutaway isometric illustration of a three story bakery, showing the kitchen with ovens on the ground floor, a cafe in the middle and the baker's apartment on top, warm pastel colors, tiny detailed people, 1:1 square frame.

Prompt

Concept art of a floating market city built on canals at dusk, wooden boats loaded with fruit, layered stilt houses with paper lanterns, glowing water, painterly fantasy style with rich detail, 16:9 frame.

Seedream 5.0 prompts for video first frames

A cyclist in a yellow raincoat pauses at the start of an empty coastal road at dawn, the road curving toward a distant white lighthouse above a misty sea

Image prompt: Wide cinematic film still of a lone cyclist in a yellow raincoat paused at the start of an empty coastal road at dawn, one foot on the ground, the road curving away toward a distant white lighthouse, low mist over a calm sea. The rider sits small in the left third of the frame with open road and sky ahead, soft pink and pale blue light, fine film grain, photorealistic, 16:9 frame.

A good first frame is the moment just before something happens, with room in the frame for that movement. Pick the video's shape now, usually 16:9 or 9:16, and leave text out of the picture, because lettering tends to wobble once it moves.

Prompt

Wide establishing shot of a small fishing boat leaving a harbor at sunrise, gulls above the mast, calm water with long reflections, the boat in the right third and open sea ahead, photorealistic film still, 16:9 frame.

Prompt

Close-up of a barista's hands holding a milk pitcher just above a cup of espresso, the first drop of milk about to touch the crema, warm cafe light, shallow depth of field, cinematic still, 16:9 frame.

Prompt

Vertical film still of a dancer in a flowing red dress caught mid spin on an empty rooftop at dusk, city lights blurred behind her, the fabric fanned out around her, open space above her head, 9:16 frame.

Prompt

Low angle still of a vintage red convertible parked on a desert highway at golden hour, heat haze on the road ahead, a long empty horizon, the car in the left third facing right, cinematic color grade, 21:9 frame.

Prompt

Overhead shot of a chef's hands about to sprinkle sea salt over a seared steak in a cast iron pan, light smoke rising, dark moody kitchen, crisp detail on the crust, food commercial still, 16:9 frame.

Prompt

Medium shot of a child in a yellow raincoat standing at the edge of a puddle on a gray city sidewalk, one boot lifted mid step, soft overcast light, shallow depth of field, gentle cinematic still, 9:16 frame.

How Seedream 5.0 prompts differ from GPT Image and Nano Banana prompts

The same sentence-style prompt runs on all three families, but each rewards a different habit. The limits below are the ones each model has in Yovid's image generator.

Seedream 5.0 Pro Seedream 5.0 Lite GPT Image 2.5 Nano Banana 2.1
Prompt length Up to 2,600 characters Up to 2,600 characters Up to 32,000 characters Up to 2,000 characters
Images per prompt 1 Up to 15 Up to 10 1
Reference images 10 14 16 14
Transparent background No No Yes No
  • Seedream 5.0: short, precise sentences, the image's purpose stated up front and every word of on-image text in quotes. It's the one to pick for Chinese lettering and for layouts with several text blocks.
  • GPT Image 2.5: takes long, detailed briefs and transparent cutouts. Our GPT Image 2.5 prompt examples show how far a detailed brief can go.
  • Nano Banana 2.1: Google recommends describing the clean result you want instead of listing what to avoid. The Nano Banana product shot prompts use that approach throughout.

Seedream also has a newer member, Seedream 5.5, released on November 12, 2026. Our Seedream 5.5 review runs both models through the same tests if you are choosing between them.

Common Seedream 5.0 prompt mistakes and fixes

Keyword lists. "Girl, umbrella, street, oil painting" leaves the model guessing how the pieces relate. Write one sentence that connects them.

Unquoted text. "A poster that says summer sale" may come back with different words. Write the headline "SUMMER SALE" in quotes, keep it short and say where it sits.

Vague edits. "Make it better" or "change that one" gives the model no target. Name the element, the change and what must stay the same.

Stacked adjectives. "Ultra detailed, masterpiece, stunning, 8K, award winning" adds noise, not quality. Output size comes from the size setting, so spend the words on light, materials and composition instead.

The wrong version. Pro stops at 2K and makes one image per prompt. For 3K or 4K, or for a dozen variations of one idea, switch to Lite and set the image count.

How to turn a Seedream 5.0 image into a Seedance 2.5 video

The first-frame prompts above are written to be animated. On Yovid, it takes three steps:

  1. Generate the still in the AI image generator with Seedream 5.0 Pro or Lite, at the shape you want for the video.
  2. Open the image-to-video generator, choose Seedance 2.5 and set the image as the first frame. Seedance 2.5 follows the shape of the frame you give it, so a 9:16 still becomes a 9:16 clip.
  3. Describe only what moves: the camera, the subject's action and, since Seedance 2.5 generates sound, what you should hear.
Prompt

Slow push in toward the cyclist as she clips into the pedal and rides off along the coastal road, mist drifting over the sea, gulls calling in the distance, soft wind noise.

Seedance 2.5 accepts JPG, PNG or WebP first frames up to 30 MB, with each side at most 6,000 pixels and a shape between 2:5 and 5:2. Every Seedream 5.0 preset ratio fits; if you used custom pixels for a very wide banner or rendered at 4K on Lite, check the size first.

What 30 Seedream 5.0 prompts cost on Yovid

On Yovid, Seedream 5.0 Pro and Lite sit in the image model menu on every plan. Credits depend on the version and size, and attaching reference images doesn't change the price.

Credits per image 1K 1.5K 2K 3K 4K
Seedream 5.0 Pro 2 2 (default) 3 — —
Seedream 5.0 Lite — — 2 (default) 2 2

Lite charges per image, so a batch of 15 is 30 credits. For comparison, Seedream 5.5 starts at 1 credit at 1K or 1.5K and 2 at 2K, but it adds credits for each reference image you attach.

Here is what the 30 prompts in this article cost to run once each, and what a credit pack buys, using the pack prices on the pricing page on 2026-11-20:

Setting All 30 prompts Images from Starter (200 credits, $29.90) Images from Creator (375 credits, $49.90) Cost per image on Creator
Pro at 1K or 1.5K 60 credits 100 187 $0.27
Pro at 2K 90 credits 66 125 $0.40
Lite at 2K, 3K or 4K 60 credits 100 187 $0.27
Lite, 4 variations per prompt 240 credits 100 187 $0.27

Starter credits last 180 days and Creator credits 365 days. Lite at 4K costs the same as Lite at 2K, which makes it the cheaper route to large prints and wide banners.

FAQ

What is the difference between Seedream 5.0 Pro and Lite?

Pro returns one image per prompt at 1K, 1.5K or 2K, takes up to 10 reference images and handles text in many languages, so it's the one for posters, infographics and edits. Lite renders at 2K, 3K or 4K, returns 1 to 15 images per prompt and takes up to 14 references. Both share the same aspect ratios.

What resolution does Seedream 5.0 make?

Pro goes up to 2K, about 2048 × 2048 pixels for a square image, and accepts custom sizes from roughly 0.9 to 4.6 megapixels. Lite goes up to 4K, with custom sizes from roughly 3.7 to 16.8 megapixels. Both offer 11 preset aspect ratios from 21:9 to 9:21, plus auto.

How many credits does a Seedream 5.0 image cost?

On Yovid, Pro costs 2 credits per image at 1K or 1.5K and 3 credits at 2K. Lite costs 2 credits per image at any size. Running all 30 prompts in this article once takes 60 credits on Pro at the default size.

Can I use Seedream 5.0 images commercially?

On Yovid, commercial use is included with Pro and Premium on monthly or yearly billing, with Lite on yearly billing and with every credit pack. Free accounts and monthly Lite don't include it. You remain responsible for the rights to anything you upload, including logos, faces and characters.

Can Seedream 5.0 put text in images?

Yes, and it's one of the model's strengths, especially on Pro. Put the exact words in double quotation marks, keep headlines short and name the kind of design, such as a poster or a menu. ByteDance says fine-grained text still has room to improve, so proofread small labels and numbers before you publish.

Try your first Seedream 5.0 prompt

Copy one prompt from the group that matches your project, open the AI image generator, choose Seedream 5.0 Pro or Lite, set the ratio and generate. More collections are in the Prompt Library.

If you want to see a still in motion, Yovid's free AI video generator makes one watermarked 4-second 480p Seedance 2.0 Mini clip without signing up (not on Android). That trial covers Seedance 2.0 Mini only; Seedream 5.0 images and Seedance 2.5 clips run on credits.