Alibaba 高速中英雙語圖片模型
z image turbo AI 圖片生成器：中英文字，快速到位
用 z image turbo 把英文與中文直接寫進圖片，從霓虹店招、產品標籤到海報標題，每張只要 0.25 點數。
用 Alibaba 的 6B 參數模型 Z Image Turbo，打造雙語招牌街景、印字標籤商品照、字體設計海報、雜誌風人像、美食、室內空間與概念藝術。本頁每張圖片都由我們在 Pixel Dojo 製作，並附上原始提示詞，讓你直接選一張開始創作。Pixel Dojo 一個方案，就能使用所有頂尖模型、六大工作室，以及 AI 導演 Sensei。訂閱期間，未用完的點數可累積保留，所有方案皆含商業使用權。
0.25 credits ($0.02) 所有方案皆包含，每月 US$10 起。
已經是會員？ 登入
作者 Brian Lovett · 更新於
Z Image Turbo 能做出什麼？
這七張圖片都用 Z Image Turbo 的預設 8 步生成，每張 0.25 點數：兩張雙語標籤與招牌、一張字體設計海報、一張人像、一張室內空間、一張美食照，以及一張概念藝術。每組提示詞都像一份拍攝清單，可以原樣套用，也能換成你的主題。點選「使用此提示詞」，就能從任一張開始。
9:16 夜間街拍，搭配中英雙語霓虹招牌。「LUCKY NOODLE」與「好运面馆」都清楚呈現在廚師上方，蒸氣、雨後濕亮的路面與路過的雨傘構成畫面。想做店面與招牌圖片，就從這組提示詞開始。
Candid street photograph at night, 35mm lens at chest height. A noodle cook in a white cap and a splattered canvas apron lifts a tangle of noodles out of a steaming wok at a tiny street stall, above him a glowing red and green neon sign reads "LUCKY NOODLE" over the Chinese characters "好运面馆". Rain slick pavement in the foreground mirrors the neon, clouds of steam rise into the light, two passersby with umbrellas blur past. Mixed tungsten and neon light, slight motion blur on the noodles, high ISO grain, documentary street photography.
4:3 商品照，搭配中英雙語標籤。「OOLONG」與「乌龙茶」清晰印在茶罐上，暖色逆光讓盤旋蒸氣與濕潤石板泛出金光。換上你的商品與標籤文字，就能接著創作。
Low angle studio product photograph, 100mm macro lens. A matte charcoal ceramic tea tin with a cream paper label printed in bold black serif letters "OOLONG" above the red Chinese characters "乌龙茶" stands on a wet slab of dark basalt, beside it a small pale celadon cup of amber tea sends a curl of steam up into the light, a few rolled dry tea leaves scattered on the wet stone. One hard warm backlight from behind turns the steam and the water beads on the stone glowing gold, a soft cool fill on the label, deep charcoal background, crisp reflections in the wet stone, premium tea advertising photography.
3:4 雜誌風人像。柔和的陰天光線下，雀斑、細紋、風吹的銀髮與溫暖笑容自然呈現，灰色海面襯托人物。想做角色與形象廣告人像，就從這組提示詞開始。
Tight editorial portrait at eye level, 85mm lens. A woman in her sixties with a cropped silver pixie cut, round tortoiseshell glasses and a mustard yellow ribbed wool turtleneck laughs as she turns toward the camera, one hand pushing wind blown hair off her forehead, standing at the rail of a grey ferry with a pale sea horizon behind her. Overcast soft daylight, a thin cool rim light along her cheek, real skin texture with freckles and fine lines, shallow depth of field, muted teal and ochre color grade, fine 35mm film grain.
16:9 建築影像。筆直的垂直線條、雕塑般的螺旋樓梯，以及從格狀天窗灑落的陽光，搭配一位穿紅色大衣的讀者作為比例參照。想做室內空間與建築圖片，就從這組提示詞開始。
Wide interior architecture photograph, 24mm tilt shift lens with perfectly straight verticals. A brutalist concrete public library with a five story atrium, tiers of warm oak bookshelves on every level, a sculptural concrete spiral staircase in the center, shafts of late afternoon sun cutting through a grid of skylights and landing on long reading tables, one reader in a red coat sitting alone for scale. Warm sunlight against cool grey board formed concrete, dust floating in the light beams, architecture magazine photography.
3:4 字體設計海報。橘色與鈷藍色構成平面的瑞士風格網格，標題、場地、日期與開放入場時間，都精準呈現指定文字。換成你的活動與日期，就能直接套用。
A bold Swiss style concert poster in portrait format printed on off white paper. A huge tilted orange circle overlaps a deep cobalt blue rectangle. Giant black condensed sans serif letters read "TIDE SIGNALS" across the top, and below them smaller black type reads "LIVE AT PIER NINE" and then "SATURDAY 14 NOVEMBER". A thin black rule runs along the bottom above the words "DOORS OPEN AT EIGHT". Flat colors, crisp grid layout, subtle paper texture, every word spelled exactly as written.
4:3 美食照。酥脆麵包劃開半熟蛋黃，鑄鐵鍋中撒著菲達起司碎、香菜與辣椒碎，沐浴在窗邊柔和的晨光中。想做菜單與美食貼文，就從這組提示詞開始。
Close food photograph from a 45 degree angle, 50mm lens. A black cast iron skillet of shakshuka with three glossy runny eggs, crumbled white feta, torn cilantro and red chili flakes sits on a worn olive wood table, a hand tearing a crusty piece of sourdough beside it, a rumpled linen napkin and a small dish of flaky salt nearby. Soft morning window light from the left, a wisp of steam rising, deep tomato red and fresh green palette, crisp crust texture, food magazine photography.
16:9 概念藝術。巨龜背著燈籠照亮的村落與虯曲松樹，在日出時分穿過霧海，上空飛鳥與兩艘小船襯托出龐大尺度。想打造遊戲與電影世界，就從這組提示詞開始。
Epic wide concept art from a low camera just above the water. A colossal ancient stone turtle with a moss covered shell carries a small village of lantern lit wooden houses and a crooked pine tree on its back as it wades through a shallow misty sea at dawn, a flock of white birds circles its head and two tiny fishing boats drift nearby for scale. Soft pink and gold sunrise backlight through the mist, volumetric light rays, painterly matte painting detail, cinematic color grade.
Z Image Turbo 一覽：尺寸、設定與價格
一次看懂 Pixel Dojo 上的所有可調設定，以及我們在各種長寬比下實際取得的檔案像素尺寸。
| 功能 | Pixel Dojo 上的 Z Image Turbo |
|---|---|
| 價格 | 每張 0.25 點數，所有尺寸與步數均同價 |
| 圖片文字 | 英文與中文，請以雙引號標示 |
| 正方形 1:1 | 1024 x 1024 |
| 橫式 16:9 | 1344 x 768 |
| 直式 9:16 | 704 x 1280 |
| 標準 4:3 | 1376 x 1024 |
| 人像 3:4 | 1024 x 1392 |
| 推論步數 | 1 至 8，預設 8，細節最豐富 |
| 種子值 | 設定種子值，即可重現同一張圖片 |
| 自訂 LoRA | 每次生成最多 3 個，價格不變 |
| 模型 | Alibaba 通義團隊推出的 6B 參數開放權重模型，採用 Apache 2.0 授權 |
Z Image Turbo vs Flux 2 Klein 4B：招牌與文字實測
我們將相同提示詞交給 Z Image Turbo，以 8 步生成，再讓 Flux 2 Klein 4B 使用 1 MP 設定，以相同長寬比生成。Z Image Turbo 精準呈現了雙語霓虹招牌的兩行文字，以及花店黑板上的指定文字。Klein 4B 每張只要 0.1 點數，Z Image Turbo 則是 0.25 點數，而 Klein 4B 呈現的花藝師人像更溫暖、構圖更完整。只要畫面需要招牌、標籤或標題，不論英文或中文，都選 Z Image Turbo；以參考圖片創作，則是 Klein 4B 的強項。
相同提示詞： Candid street photograph at night, 35mm lens at chest height. A noodle cook in a white cap and a splattered canvas apron lifts a tangle of noodles out of a steaming wok at a tiny street stall, above him a glowing red and green neon sign reads "LUCKY NOODLE" over the Chinese characters "好运面馆". Rain slick pavement in the foreground mirrors the neon, clouds of steam rise into the light, two passersby with umbrellas blur past. Mixed tungsten and neon light, slight motion blur on the noodles, high ISO grain, documentary street photography.
Z Image TurboZ Image Turbo：招牌兩行文字完全正確，英文「LUCKY NOODLE」與中文「好运面馆」都精準呈現。
Candid street photograph at night, 35mm lens at chest height. A noodle cook in a white cap and a splattered canvas apron lifts a tangle of noodles out of a steaming wok at a tiny street stall, above him a glowing red and green neon sign reads "LUCKY NOODLE" over the Chinese characters "好运面馆". Rain slick pavement in the foreground mirrors the neon, clouds of steam rise into the light, two passersby with umbrellas blur past. Mixed tungsten and neon light, slight motion blur on the noodles, high ISO grain, documentary street photography.
Flux 2 Klein 4BFlux 2 Klein 4B：英文「LUCKY NOODLE」在熱鬧夜攤上方發光，四周充滿蒸氣與雨幕。
Candid street photograph at night, 35mm lens at chest height. A noodle cook in a white cap and a splattered canvas apron lifts a tangle of noodles out of a steaming wok at a tiny street stall, above him a glowing red and green neon sign reads "LUCKY NOODLE" over the Chinese characters "好运面馆". Rain slick pavement in the foreground mirrors the neon, clouds of steam rise into the light, two passersby with umbrellas blur past. Mixed tungsten and neon light, slight motion blur on the noodles, high ISO grain, documentary street photography.
相同提示詞： Medium close shot at eye level, 50mm lens. A florist with rolled denim sleeves and a green canvas apron ties brown twine around a bouquet of orange ranunculus on a zinc counter, both hands visible, while a hand painted chalkboard behind her reads "FRESH CUT TODAY" in white letters. Soft window light from the right, buckets of tulips and eucalyptus in the background, shallow depth of field, natural color, 35mm film grain.
Z Image TurboZ Image Turbo：黑板上的「FRESH CUT TODAY」拼字完全正確，橘色陸蓮花也依提示詞要求用麻繩綁好。
Medium close shot at eye level, 50mm lens. A florist with rolled denim sleeves and a green canvas apron ties brown twine around a bouquet of orange ranunculus on a zinc counter, both hands visible, while a hand painted chalkboard behind her reads "FRESH CUT TODAY" in white letters. Soft window light from the right, buckets of tulips and eucalyptus in the background, shallow depth of field, natural color, 35mm film grain.
Flux 2 Klein 4BFlux 2 Klein 4B：花藝師站在工作檯前，呈現溫暖、構圖完整的人像，每張 0.1 點數。
Medium close shot at eye level, 50mm lens. A florist with rolled denim sleeves and a green canvas apron ties brown twine around a bouquet of orange ranunculus on a zinc counter, both hands visible, while a hand painted chalkboard behind her reads "FRESH CUT TODAY" in white letters. Soft window light from the right, buckets of tulips and eucalyptus in the background, shallow depth of field, natural color, 35mm film grain.
| Z Image Turbo | Flux 2 Klein 4B | |
|---|---|---|
| 每張圖片點數 | 0.25 | 0.1 |
| 雙語霓虹招牌 | LUCKY NOODLE 與 好运面馆 皆正確 | LUCKY NOODLE 正確 |
| 花店場景 | 黑板上的 FRESH CUT TODAY 拼字完全正確 | 花藝師人像更溫暖、構圖更完整 |
| 9:16 像素尺寸 | 704 x 1280 | 768 x 1360 |
| 參考圖片 | 文字生成圖片 | 最多 5 張 |
| 最適合 | 招牌、標籤、海報與中文文字 | 低成本生圖與參考圖片創作 |
Z Image Turbo 出圖技巧
用同一組皮革筆記本提示詞，固定種子值 1729，分別以 2、4、8 推論步數生成，再用相同種子值以 8 步生成一次。無論步數多少，金色「ATLAS OF TIDES」字樣都精準呈現；步數越高，細節越豐富，而相同種子值能重現同一張圖片。
2 步，種子值 1729：「ATLAS OF TIDES」字樣已完全正確，皮革與地圖呈現平滑柔和的質感。
Studio photograph from above at a slight angle, 70mm lens. A hand bound leather journal with a brass corner clasp lies on a dark walnut desk, its cover embossed in gold with the words "ATLAS OF TIDES", a fountain pen, a folded paper nautical chart and a small brass compass arranged beside it. Warm lamp light from the upper left, soft shadows, rich leather grain and wood texture, crisp focus on the lettering.
4 步，相同種子值：皮革紋理與地圖線條更加豐富。
Studio photograph from above at a slight angle, 70mm lens. A hand bound leather journal with a brass corner clasp lies on a dark walnut desk, its cover embossed in gold with the words "ATLAS OF TIDES", a fountain pen, a folded paper nautical chart and a small brass compass arranged beside it. Warm lamp light from the upper left, soft shadows, rich leather grain and wood texture, crisp focus on the lettering.
預設 8 步：皮革摺痕最豐富，地圖線條最銳利，同樣只要 0.25 點數。
Studio photograph from above at a slight angle, 70mm lens. A hand bound leather journal with a brass corner clasp lies on a dark walnut desk, its cover embossed in gold with the words "ATLAS OF TIDES", a fountain pen, a folded paper nautical chart and a small brass compass arranged beside it. Warm lamp light from the upper left, soft shadows, rich leather grain and wood texture, crisp focus on the lettering.
再次以種子值 1729 跑 8 步：生成同一張圖，讓你隨時重現喜歡的畫面。
Studio photograph from above at a slight angle, 70mm lens. A hand bound leather journal with a brass corner clasp lies on a dark walnut desk, its cover embossed in gold with the words "ATLAS OF TIDES", a fountain pen, a folded paper nautical chart and a small brass compass arranged beside it. Warm lamp light from the upper left, soft shadows, rich leather grain and wood texture, crisp focus on the lettering.
- 所有必須出現在圖中的文字，不論英文或中文，都用雙引號框起來。我們的霓虹招牌印出了「LUCKY NOODLE」和「好运面馆」，茶罐則印出了「OOLONG」和「乌龙茶」。
- 指定文字的用途與樣式。我們的海報提示詞要求標題使用超大黑色窄體字，場地與日期則用較小字體，最後加上「每個字都必須與原文完全一致」。
- 步數維持 8 就好。無論設定幾步，價格都是 0.25 點數；我們的日誌在 8 步時呈現最豐富的皮革質感，地圖也最銳利。
- 找到喜歡的圖，就設定種子值。我們以種子值 1729 跑了兩次 8 步，得到的是同一張圖。
- 像寫分鏡清單一樣寫提示詞：先寫鏡頭與角度，再寫主體、一個動作、光線與質感。本頁每個提示詞都依照這個順序。
- 用自訂 LoRA 加入你的風格或角色，每次生成最多可用 3 個，價格不變。
Z Image Turbo 費用：所有方案皆包含
Pixel Dojo 所有方案都包含 Z Image Turbo，每月 $10 起，每張圖片使用方案內的 0.25 點數。訂閱期間，剩餘點數可累積。無論長寬比或步數，價格都相同，1 點數就能生成 4 張圖。
自訂 LoRA 與固定種子值都不加價，讓你精準調出想要的風格，之後也能以相同的 0.25 點數重現。
Z Image Turbo常見問題
什麼是 Z Image Turbo？
Z Image Turbo 是 Alibaba 通義團隊推出的 6B 參數文字生成圖片模型，專為快速生成寫實圖片與清晰文字打造。在 Pixel Dojo 輸入提示詞，從 5 種長寬比中選一種，就能生成圖片，也可透過 Pixel Dojo API 使用。
Z Image Turbo 擅長生成圖片中的文字，包括中文嗎？
是，這正是它最出色的強項。我們的霓虹招牌精準印出了「LUCKY NOODLE」和「好运面馆」，茶罐標籤印出了「OOLONG」和「乌龙茶」，演唱會海報的 4 行文字也全部與原文一致。將每段文字用雙引號框起來，並指定放置位置。
Z Image Turbo 生成的圖片可以商用嗎？
可以。無論使用哪個方案，你都擁有自己創作的內容。
Z Image Turbo 免費嗎？
免費帳號每天可免費使用 3 次效果，體驗 Pixel Dojo。每月 $10 起的所有方案都包含 Z Image Turbo，訂閱期間剩餘點數可累積。每張圖片使用 0.25 點數，不論長寬比、步數，或是否使用自訂 LoRA，價格都一樣。
Z Image Turbo 支援哪些圖片尺寸？
共 5 種長寬比。我們取得的檔案尺寸為：1:1 是 1024 x 1024、16:9 是 1344 x 768、9:16 是 704 x 1280、4:3 是 1376 x 1024、3:4 是 1024 x 1392。
步數、種子值與 LoRA 設定有什麼作用？
步數可設為 1 到 8，預設為 8；在相同價格下，8 步讓我們的圖片呈現最豐富的質感。種子值能讓你再次執行提示詞時，重現同一張圖。自訂 LoRA 可加入你的風格或角色，每次生成最多可用 3 個。
Z Image Turbo 是開源模型嗎？
是。Alibaba 通義團隊已在 Hugging Face 上，以 Apache 2.0 授權釋出 Z Image Turbo 模型權重。