跳至主要內容
AI 影片模型
ByteDance
Seedance

Seedance 2.0

Seedance 2.0 / 功能標籤 — text 不接受素材;frames 必須提供首幀,可選填尾幀;reference 必須有提示詞,以及至少一種圖片、影片或音訊參考。多模態參考不能與首尾幀欄位同時送出。 Seedance 2.0 支援文字、首尾幀和多模態參考三種互斥場景。目前可呼叫輸出為 720p、1080p、4K 與 4–15 秒;含參考影片的請求只開放 1080p 或 4K,因為目前只有這兩個組合具備可核驗價格。 每次 164–3,840 點

功能標籤
文字、首尾幀、多模態參考
點數消耗
每次 164–3,840 點
提示詞長度
3–20,000 字元
上傳限制
參考圖 9 張/影片 3 段/音訊 3 段
30 秒快速瞭解
  • Seedance 2.0 / 三種互斥輸入情境 — text 不接受素材;frames 必須提供首幀,可選填尾幀;reference 必須有提示詞,以及至少一種圖片、影片或音訊參考。多模態參考不能與首尾幀欄位同時送出。
  • Seedance 2.0 / 上傳限制 — Kyeo 允許 JPG/JPEG/PNG/WebP/GIF/BMP/TIFF 圖片最大 30 MB、影片最大 50 MB、音訊最大 15 MB;reference 最多 9 張圖、3 段影片、3 段音訊,影片與音訊各自的總長度不得超過 15 秒。圖片寬高須為 300–6000 像素、比例 0.4–2.5;參考影片寬高 300–6000、像素積 409600–927408、幀率 24–60。上傳中繼資料會在生成前複驗。
  • Seedance 2.0 / 解析度 — Seedance 2.0 支援文字、首尾幀和多模態參考三種互斥場景。目前可呼叫輸出為 720p、1080p、4K 與 4–15 秒;含參考影片的請求只開放 1080p 或 4K,因為目前只有這兩個組合具備可核驗價格。
  • Seedance 2.0 / 點數消耗 / 生成情境 — 目前可呼叫組合:沒有參考影片時,720p/1080p/4K 分別為每輸出秒 41/102/208 點;含參考影片時,1080p/4K 分別為輸入影片與輸出合計每秒 62/128 點。請求會預留完整估算額,並按成功任務的實際用量結算。

熱門搜尋: 如何在 Kyeo 使用 Seedance 2.0 標準版 · Seedance 2.0 標準版每次使用多少點數 · Seedance 2.0 可上傳哪些參考素材

示例

Cinematic scene

Create a luxurious 9-second skincare commercial for LUMIÈRE Skin Bloom Essence with a soft, luminous pink aesthetic. Begin with floating translucent pink droplets, then reveal a young woman with glowing, hydrated skin and slicked-back dark hair as water droplets move around her in slow motion. Show a close-up of the frosted glass bottle with water droplets and a silver cap, followed by the model gently touching her smooth jawline. End with her smiling while holding the product beside her face, surrounded by floating rose petals. Use elegant white serif typography, subtle sparkles, cinematic lighting, dreamy slow-motion transitions, and a premium Japanese beauty-ad feel. Voiceover in Japanese: 「肌は…まだ見ぬ光を知る。満ちる透明感、解き放たれる輝き。リュミエール、美しさのその先へ。」

Cinematic scene

Create a 15-second ultra-luxury jewelry commercial featuring a beautiful female model wearing an elegant diamond necklace. Warm golden luxury lighting, premium boutique atmosphere, soft cinematic bokeh, realistic skin texture, natural makeup, champagne satin dress, photorealistic 8K quality, shallow depth of field, high-end fashion editorial style. Scene 1 (0–3s): Extreme close-up of the diamond necklace. Soft fade-in as the diamonds sparkle with realistic light reflections. Slow cinematic push-in. Scene 2 (3–6s): Medium portrait of the model looking confidently into the camera. Gentle smile, graceful posture, subtle hair movement, warm golden lighting. Scene 3 (6–9s): Close-up as she delicately touches the necklace with her fingertips while looking down. Smooth camera movement, elegant expression. Scene 4 (9–12s): Macro shot of the necklace pendant. Brilliant diamond fire, crisp reflections, luxurious sparkle, rotating camera with cinematic focus pull. Scene 5 (12–15s): Final centered portrait of the model facing the camera with the necklace perfectly framed. Premium glow, elegant background bokeh, luxury brand aesthetic. End with the text: “Elegance you can wear. Shine without saying a word.” Style: Hyper-realistic, cinematic, luxury fashion advertisement, premium jewelry campaign, HDR, soft golden tones, volumetric lighting, flawless composition, smooth transitions, no watermark, no distortion, no extra jewelry, no text until final scene.

Cinematic scene

CAMERA / LOOK: Propped iPhone/Mini DV camcorder footage on a sleek vanity mirror stand. Soft diffuse lighting, delicate lens flare, gentle zoom adjustments, subtle tape grain, polished high-end aesthetic. STYLE: High-maintenance luxury beauty/skincare routine with sassy, condescending ASMR elements. Paws delicately patting bottles, slow blinks, arrogant pauses. SUBJECT: Fluffy white Persian cat with a flat smushed face, emerald green eyes, wearing a tiny silk hair wrap. Smug, unimpressed expression, regal demeanor. SETTING: Marble bathroom vanity with expensive serum glass bottles, jade roller, plush velvet towel, warm vanity light bulbs. STORYBOARD: → (2s, medium shot) Adjusts camera angle with one claw. "Good morning to everyone except my human who woke me up 3 minutes early." → (1.5s, detail shot) Taps a glass serum bottle with front paw. Crisp clinking glass ASMR sound. → (1.5s, close-up) Slowly pushes serum bottle off the counter while maintaining eye contact with camera. Loud shatter/thud sound below. "Oops. Gravity works. Moving on." → (2s, detail macro shot) Softly licks back of paw and wipes ear in slow motion. Soft grooming ASMR sound. → (1.5s, close-up) Gazes into mirror, tilting head. "Skin is glowing, pores are non-existent. You're welcome." → (1.5s, medium handheld shot) Uses jade roller on forehead using tiny paw. "Reducing puffiness from sleeping 18 hours instead of 19." → (1.5s, macro detail shot) Sniffs a tub of catnip cream, letting out a delicate sneezed 'achoo'. → (1.5s, medium shot) Sits tall, smoothing chest fur with arrogant posture. "If you can't afford this routine, just say that." → (1s, ending shot) Slow blinks condescendingly into lens. "Don't forget to subscribe. Or don't. I don't care." → (1s, final shot) Swats camera violently off vanity counter. Video cuts to static grid. AUDIO NOTES: Glass bottle clinking, soft licking/grooming sounds, heavy purring sub-bass, arrogant posh voice with refined lip movements. REALISM & COMEDY NOTES: Photorealistic white Persian fur physics, subtle whisker twitches, natural cat slow-blinks, hysterical condescending tone paired with classic cat chaotic behavior (knocking things off tables).

Cinematic scene

“PINKY RARE 10s | 9:16 Vertical | Luxury Fashion Commercial 1. 0:00–0:01 — WIDE ESTABLISHING SHOT Camera: 28mm locked, slow dolly forward Notes: Pool water ripples gently, palm leaves sway naturally. 2. 0:01–0:02 — SIDE POOL VIEW Camera: 35mm, slow left-to-right slide Notes: Realistic parallax, sparkling turquoise water and striped umbrellas. 3. 0:02–0:03 — MACRO POOL WATER Camera: Macro, tilt downward Notes: Tilt naturally until shimmering reflections fill the frame. 4. 0:03–0:04 — FEET WALKING Camera: Low angle-level tracking, 85mm Notes: Follow her bare feet walking beside the pool. Sunglasses hang naturally from her hand. 5. 0:04–0:05 — SUN KISS PORTRAIT Camera: Low-angle, tilt up Notes: She raises her hand to shield the sun. Warm lens flare between her fingers. 6. 0:05–0:06 — BEAUTY CLOSE-UP Camera: 85mm, slow push-in Notes: She lowers her hand, gently adjusts her sunglasses and smiles confidently. 7. 0:06–0:07 — OVER-SHOULDER PORTRAIT Camera: 50mm, slow arc Notes: Camera arcs around her shoulder as she turns back toward the pool with relaxed elegance. 8. 0:07–0:08 — SWIMSUIT MACRO Camera: 100mm macro, slow push-in Notes: Reveal realistic stitching, folds, woven texture and sparkling water droplets. 9. 0:08–0:10 — FINAL HERO SHOT Camera: Extreme macro, slow push Notes: Hold on final frame. Logo remains clean and static. Tiny water droplets shimmer in the sunlight. Audio Direction: Dreamy luxury summer music, subtle pool ambience, gentle breeze, soft footsteps, elegant fashion commercial sound design. Notes: Smooth cinematic transitions. Consistent lighting, color grading and environment. No extra people, no outfit changes, no AI morphing.”

Cinematic scene

Create a 30 second cinematic streetwear fashion film in 16:9, naturally recomposed for a wide frame. One young adult man, short beard, black distressed military style cap with metal eyelets, slim tinted glasses, silver hoop earring, oversized mustard yellow hoodie with a green collegiate crest, extremely wide black trousers, rugged black platform boots, silver watch, chain bracelet and rings. Keep face, clothing, accessories and body proportions consistent in every shot. Shoot at 24 fps with natural 180 degree shutter. Use 35 mm for full body, 50 mm for medium portraits and 85 mm for details. Real optical depth, gentle focus breathing, believable motion blur, natural skin texture, fine beard detail, realistic hands, correct jewelry reflections, authentic cloth weight and solid boot contact. No plastic skin, warped fingers, floating objects, morphing garments or duplicated limbs. Set everything in a small fashion studio with a cool blue gray mottled muslin seamless backdrop flowing onto the floor. The mustard hoodie is the dominant warm color against black trousers and cool teal shadows. Use one large diffused key high camera left, weak frontal bounce, soft top separation and a circular cool pool of light behind the model. Keep deep shadow under the cap brim with occasional eye detail. Add subtle film grain, soft highlight rolloff, mild halation and restrained vignette. 0 to 3 seconds, six fast editorial cuts about every half second. Extreme side closeup as he adjusts glasses and cap, hoodie torso, low boot closeup, crossed hands and watch, then centered upper body while he pulls the hood into place. Leave clean central space for optional typography later. Do not generate text. 3 to 5 seconds, centered waist shot, then the back of the hoodie with sculptural fabric details. Camera mostly locked with a slow physical push. 5 to 8 seconds, profile portrait touching the cap brim, then macro ring, bracelet and watch shots with real metal highlights. 8 to 11 seconds, reveal the huge trouser silhouette, a three quarter lower body pose with one boot lifted, then a low close shot as he bends and adjusts the laces. 11 to 14 seconds, layered editorial depth with his shoulder and face large in the near foreground while the same model stands farther back adjusting his cap. Make it feel like an intentional in camera composition, not a clone effect. 14 to 17 seconds, tight hood and shoulder texture, then a boot lifts and lands with weight, followed by a close profile portrait and waist level accessory detail. 17 to 21 seconds, seated pose on a dark stool, knees wide, hands loosely clasped, then close views of watch, bracelet, relaxed hand and hoodie cuff. Use shallow focus and a subtle rack focus between jewelry and fabric. 21 to 24 seconds, rear shoulder closeup as he pulls the hood outward, then intimate face portraits under the cap as he turns from profile toward camera, ending on a quick boot detail. 24 to 29.3 seconds, slower full body hero sequence. He rotates front to side to back and front again, briefly lifts one leg, settles with hands at the hips, then relaxes with one hand in a pocket. Keep pose changes human and unhurried. 29.3 to 30 seconds, hard cut to the opening extreme closeup as his fingers touch the glasses near his temple, creating a seamless visual loop. Edit like a premium fashion campaign with sharp hard cuts on musical beats and no artificial transitions. Music is a dark stylish instrumental around 123 BPM with deep controlled sub bass, dry punchy kick, crisp hats, sparse percussion, muted synth texture and subtle impact accents on major cuts. No vocals. The final result must feel photographed in a real studio, tactile, physically believable and imperfect.

Cinematic scene

Hyper-realistic 10s high-fashion commercial, vertical 9:16. Fast cuts synced to drums, handheld camera, high contrast, natural lighting, shallow depth of field, energetic rock beat with heavy drums and electric guitar. Whip pans, snap zooms, white strobe flash transition frames, subtle glitch effects. REAL photorealistic human woman, natural skin texture and visible pores, NOT anime, NOT cartoon, NOT CG. Use Burry as the ONLY face and outfit reference; keep her facial likeness, hairstyle and exact outfit consistent in every shot. 0-2s HOOK: medium-close shot, she looks confidently into the lens, slightly shaky handheld, a small confident smile. She says in English: "Hey, look here." Cut hard on a guitar drop. 2-4s BURST: rapid cuts on the drums, full-body confident stance, smash cut to a low-angle hero shot, white strobe flash, quick crouch as the camera drops with her, snap zoom to an outfit detail, hair moving, confident smirk. 4-7s PLAYFUL: erratic handheld, she leans toward the lens with slight wide-angle distortion, a quick spin with frame skips and glitch, a playful confident shrug. She says in English: "Keep up with me." Cut sharp on the beat. 7-10s ENDING: rhythm slows but stays intense, close-up of her hand adjusting a collar/sleeve detail, side-profile with a soft smile, hair brushing across her face in slow motion, she looks into the camera: "This is my vibe." End on a white flash frame and a snap-zoom freeze-frame pose. Audio: confident female English voice, energetic rock, punchy drums, electric guitar. Style: high-fashion editorial, tasteful and confident, focus on outfit styling and attitude, NOT a body-part closeup, no sexualized framing. No text, subtitles, logos or watermark.

一句話瞭解

這個模型表現如何?

執行方式 — text 不接受素材;frames 必須提供首幀,可選填尾幀;reference 必須有提示詞,以及至少一種圖片、影片或音訊參考。多模態參考不能與首尾幀欄位同時送出。

它擅長什麼
Seedance 2.0 / 執行方式 — text 不接受素材;frames 必須提供首幀,可選填尾幀;reference 必須有提示詞,以及至少一種圖片、影片或音訊參考。多模態參考不能與首尾幀欄位同時送出。
適合哪些使用者
Seedance 2.0 — 可用來對照測試文字分鏡、兩端畫面轉場或多模態參考任務;是否適用人物、商品、廣告或社群內容,仍須以真實素材驗證。
為什麼要在 Kyeo AI 使用
Seedance 2.0 / 點數消耗 — 目前可呼叫組合:沒有參考影片時,720p/1080p/4K 分別為每輸出秒 41/102/208 點;含參考影片時,1080p/4K 分別為輸入影片與輸出合計每秒 62/128 點。請求會預留完整估算額,並按成功任務的實際用量結算。

關鍵參數總覽

協助你快速評估此模型是否符合你的業務情境。

模型類別
AI 影片模型
廠商
ByteDance
模型家族
Seedance
執行方式
非同步任務
提示詞長度
3–20,000 字元
上傳限制
reference 最多 9 張圖、3 段影片、3 段音訊;frames 使用 1–2 張圖

其他名稱

同一個模型在官方文件、社群討論與 API 串接層可能有不同稱呼,這裡幫你統一整理。

Seedance 2.0
Seedance 2 Standard
ByteDance Seedance 2.0

使用者常見問題

從實際任務、輸入條件和結果要求出發,整理選擇模型時最需要確認的問題。

如何在 Kyeo 使用 Seedance 2.0 標準版
Seedance 2.0 標準版每次使用多少點數
Seedance 2.0 可上傳哪些參考素材
seedance scene 的 text frames reference 是什麼
Seedance 2.0 首幀和尾幀怎麼用
Seedance 2.0 參考影片如何計費
Seedance 2.0 支援幾秒影片
Seedance 2.0 與 Fast 目前參數差異
Kyeo Seedance 2.0 如何使用 4K
Seedance 2.0 是否支援生成音訊

選擇指南

還在猶豫該用哪個模型嗎?先查看這幾個關鍵決策點。

1
先依輸入選擇情境 / Seedance 2.0

先依輸入選擇情境 / Seedance 2.0 — 只有文字時選 text;需要約束首幀或首尾幀時選 frames;要引用圖片、影片或音訊時選 reference。三條路徑互斥,建立任務前先確認素材。

2
標準版與 Fast 要用相同題目比較 / Seedance 2.0

標準版與 Fast 要用相同題目比較 / Seedance 2.0 — Seedance 2.0 支援文字、首尾幀和多模態參考三種互斥場景。目前可呼叫輸出為 720p、1080p、4K 與 4–15 秒;含參考影片的請求只開放 1080p 或 4K,因為目前只有這兩個組合具備可核驗價格。

3
從最低合法組合開始 / Seedance 2.0

從最低合法組合開始 / Seedance 2.0 — 第一次可用 720p、4 秒和單一情境測試。reference 先只放一種素材,確認輸入、音訊、尾幀與結果正常後,再逐項增加素材或長度並重新核對點數。

使用者常搜尋的比較

把常見候選放在同一任務下比較,協助你確認輸入方式、控制項和成本差異。

Seedance 2.0 標準版與 Fast 的計費公式有何不同?
標準版與 Fast 在 Kyeo 是否開放相同解析度?
Seedance 2.0 與 1.5 Pro 分別支援哪些輸入?
Seedance 2.0 與 Kling 2.6 目前可調項目有何不同?
比較標準版與 Fast 時應固定哪些條件?

模型比較表

還在猶豫該選哪一個嗎?這張表協助你快速看清目前模型與替代方案的核心差異。

Kyeo 輸入情境
Seedance 2.0
Kyeo 輸入情境 — 文字/首幀或首尾幀/多模態參考
Seedance 2.0 Fast
Seedance 2.0 Fast — 文字/首幀或首尾幀/多模態參考
Seedance 1.5 Pro
文字或 0–2 張圖片
目前計費公式
Seedance 2.0
目前計費公式 — 目前可呼叫組合:沒有參考影片時,720p/1080p/4K 分別為每輸出秒 41/102/208 點;含參考影片時,1080p/4K 分別為輸入影片與輸出合計每秒 62/128 點。請求會預留完整估算額,並按成功任務的實際用量結算。
Seedance 2.0 Fast
Seedance 2.0 Fast:480p 按每輸出秒 11.7 點計費,有參考影片時為 6.8;720p 對應 24.8 或 15。每次合計 47–450 點。
Seedance 1.5 Pro
目前計費公式 — Seedance 1.5 Pro: 目前可用的同類模型,可從詳情頁繼續比較功能與價格。
參考素材上限
Seedance 2.0
參考素材上限 — 9 張圖、3 段影片、3 段音訊;影片/音訊各總計 15 秒
Seedance 2.0 Fast
Seedance 2.0 Fast — 9 張圖、3 段影片、3 段音訊;影片/音訊各總計 15 秒
Seedance 1.5 Pro
最多 2 張圖片
長度與解析度
Seedance 2.0
長度與解析度 — 720p/1080p/4K,4–15 秒
Seedance 2.0 Fast
480p/720p,片長 4–15 秒
Seedance 1.5 Pro
480p/720p/1080p,片長 4/8/12 秒

實際使用體驗

結合公開資料、目前站內邊界和典型任務,說明模型適合與不適合的情況。

預設組合也依目前費率計算 / Seedance 2.0

預設組合也依目前費率計算 / Seedance 2.0 — 目前可呼叫組合:沒有參考影片時,720p/1080p/4K 分別為每輸出秒 41/102/208 點;含參考影片時,1080p/4K 分別為輸入影片與輸出合計每秒 62/128 點。請求會預留完整估算額,並按成功任務的實際用量結算。

費率較高不代表畫質較好 / Seedance 2.0

費率較高不代表畫質較好 / Seedance 2.0 — Kyeo 標準版的四種費率都高於 Fast,但較高價格不保證畫質、指令遵循、穩定性或成功率更高。

功能詳解

三種互斥輸入情境

三種互斥輸入情境 — text 不接受素材;frames 必須提供首幀,可選填尾幀;reference 必須有提示詞,以及至少一種圖片、影片或音訊參考。多模態參考不能與首尾幀欄位同時送出。

8 項控制與動態點數計算

8 項控制與動態點數計算 — 可選 720p、1080p、4K、包含 adaptive 的 7 種比例、4–15 秒、同步音訊、回傳尾幀、網頁搜尋與內容檢查。含參考影片時只開放 1080p 與 4K。

素材數量、格式、大小與長度限制

素材數量、格式、大小與長度限制 — Kyeo 允許 JPG/JPEG/PNG/WebP/GIF/BMP/TIFF 圖片最大 30 MB、影片最大 50 MB、音訊最大 15 MB;reference 最多 9 張圖、3 段影片、3 段音訊,影片與音訊各自的總長度不得超過 15 秒。圖片寬高須為 300–6000 像素、比例 0.4–2.5;參考影片寬高 300–6000、像素積 409600–927408、幀率 24–60。上傳中繼資料會在生成前複驗。

適合做什麼

純文字鏡頭樣本

在 text 情境只提交提示詞,固定比例、解析度與長度,再檢查鏡頭、動作、音訊和生成文字是否符合要求。

首幀或首尾幀轉場

frames 情境必須提供首幀,也可再提供尾幀。兩端畫面受到約束,但中間人物、物件、文字與動作仍須逐幀檢查。

多模態參考對照

在 reference 情境為每張圖、每段影片或音訊指定用途,再以相同輸入和 Fast 對照;參考素材不代表結果必然一致。

提示詞技巧

用可觀察的鏡頭要求組織提示詞

依序描述取景範圍、運鏡、主體動作、環境變化、對白或聲音。以可逐項檢查的條件取代「電影級、完美穩定」等標籤。

說清楚每份參考素材的角色

指出哪張圖提供人物或構圖、哪段影片提供動作或運鏡、哪段音訊提供節奏。

一次只改一個變因

先完成 720p、4 秒請求,再分別改解析度、長度、素材或模型版本。每次重新計算點數,才能看出差異來自哪個項目。

選擇它的理由

功能標籤 — 同一條工作流程涵蓋文字、首尾幀與多模態參考三種情境。
解析度 — Seedance 2.0 支援文字、首尾幀和多模態參考三種互斥場景。目前可呼叫輸出為 720p、1080p、4K 與 4–15 秒;含參考影片的請求只開放 1080p 或 4K,因為目前只有這兩個組合具備可核驗價格。
上傳限制 — 頁面會核對參考素材數量,以及影片和音訊的總長度。
點數消耗 — 建立任務前可依目前素材組合計算預估點數。

使用前須知

Seedance 2.0 / NSFW 檢查 — NSFW 檢查預設關閉;無論是否手動開啟,都不能取代人工內容、權利與發布審核。
Seedance 2.0 / 執行方式 — 不能把 Fast 名稱當成速度保證。比較時應使用相同輸入,並記錄排隊、生成、失敗與重試情況。
Seedance 2.0 / 模型類別 — 不能只因影片由模型生成就直接判定。必須檢查人物、聲音、文字、商品、商標、著作權、肖像權與廣告事實;上傳圖片、影片、音訊或開啟連網搜尋,也不會自動取得素材權利或發布許可。

可調整參數

協助你快速評估此模型是否符合你的業務情境。

生成情境
選填
參數類型: select
預設值:text
文字生成影片
首幀 / 首尾幀
多模態參考
解析度
選填
參數類型: select
預設值:720p
720p
1080p
4K
寬高比
選填
參數類型: select
預設值:16:9
1:1
4:3
3:4
16:9
9:16
21:9
Adaptive
7 個選項
影片長度
選填
參數類型: number
預設值:5
4–15 秒
生成音訊
選填
參數類型: select
預設值:true
開啟
關閉
傳回最後一幀
選填
參數類型: select
預設值:false
關閉
開啟
連網搜尋
選填
參數類型: select
預設值:false
關閉
開啟
NSFW 檢查
選填
參數類型: select
預設值:false
關閉
開啟

點數消耗

每次 164–3,840 點

目前可呼叫組合:沒有參考影片時,720p/1080p/4K 分別為每輸出秒 41/102/208 點;含參考影片時,1080p/4K 分別為輸入影片與輸出合計每秒 62/128 點。請求會預留完整估算額,並按成功任務的實際用量結算。

省點數小技巧

批次生成圖片或影片前,建議先用同一組素材,在目前模型與同類候選模型之間進行 A/B 測試;確認效果後再批次生成,避免浪費點數。

常見問題

同類模型推薦

還沒決定該用哪個嗎?順便比較一下這幾個同類候選模型。

Seedance 1.5 Pro

Seedance 1.5 Pro / 功能標籤 — Kyeo 的 Seedance 1.5 Pro 支援純文字或最多 2 張圖片輸入,提供 480p/720p/1080p、4–12 秒、生成音訊、6 種比例與鎖定鏡頭。7–180 點數由參數組合決定;Pro 名稱與較高點數不代表本頁已驗證畫質、穩定性或發布可用性。

AI 影片模型
7–180 點數/次

Kling 2.6

Kling 2.6 在 Kyeo 是一項雙工作流程影片服務:沒有圖片時呼叫文字生成影片,有圖片時呼叫單圖圖片生成影片。兩路都提供 5/10 秒與 `sound` 開關;文字工作流程另有三種長寬比。四個計費組合為 55、110、110、220 點數。

AI 影片模型
每次 55-220 點數

Seedance 2.5

Seedance 2.5 將 30 秒邊界、較高素材數量與影片編輯型參考輸入納入同一套嚴格情境合約。自動片長不會以預設 5 秒低估,而是按最長 30 秒預授權。

AI 影片模型
102–5,700 點

資訊來源

頁面內容依據廠商資料、模型系列說明與 Kyeo AI 平台串接設定整理。

來源說明

Seedance 2.0 支援文字、首尾幀和多模態參考三種互斥場景。目前可呼叫輸出為 720p、1080p、4K 與 4–15 秒;含參考影片的請求只開放 1080p 或 4K,因為目前只有這兩個組合具備可核驗價格。

最後更新: 2026-08-27
ByteDance Seed:Seedance 2.0 Official Launch
官方

ByteDance 官方說明 Seedance 2.0 支援文字、圖片、影片與音訊,最多 9 張圖、3 段影片、3 段音訊及 15 秒輸出。

查看原始來源