想必有使用ComfyUI多半都會用到圖片識別轉文字描述的功能,但究竟哪個好用,好像都沒個定論,自己之前也是別人分享的工作流用哪個就用哪個,沒特別去管它。
但最近在弄LTX Video圖片生成影片時,發現到有些圖片生成的影片效果特別差,後來發現到是因為圖片識別轉文字描述的節點,所描述的圖片並不準確,所以在生成影片時就會出現超乎預期的畫面。今天所乾脆一口氣測試好幾個知名的圖片識別轉文字描述的節點,總共花了超過一個半小時,才將所有的跑完。
本次實驗的對象有以下四個
1.Joy Caption Two
2.Clip Interrogator ♾️Mixlab
3.Deepseek Janus Pro (今天熱騰騰新出的模型)
4.Florence2Run
使用的實驗圖
![]() | ![]() | ![]() | ![]() | ![]() |
使用的顯卡:RTX3060 12GB 、記憶體:64GB DDR4
Joy Caption Two評測
1.Descriptive模式
模式:Descriptive | 輸出結果(英文) | 輸出結果(中文翻譯-Google) | 說明 |
![]() | This is a digital illustration in a vibrant, anime-inspired style. It depicts a cute, green frog with large, expressive eyes and orange-tipped fingers, sitting in a serene, circular pond surrounded by lush greenery. The frog's skin has darker green spots and it has a small, white belly. The background includes a bright blue sky with fluffy white clouds and a few butterflies fluttering around. The pond is bordered by light brown rocks, and a small pink flower blooms near the water's edge. The scene is peaceful and whimsical, with a soft, glowing light enhancing the dreamy atmosphere. | 這是一幅充滿活力的動漫風格的數位插圖。畫中描繪了一隻可愛的綠色青蛙,它有著大大的、富有表現力的眼睛和橙色的指尖,坐在一個寧靜的圓形池塘里,周圍環繞著鬱鬱蔥蔥的綠色植物。青蛙的皮膚上有深綠色的斑點,還有一個小小的白色腹部。背景包括明亮的藍天、蓬鬆的白雲和幾隻飛舞的蝴蝶。池塘邊緣是淺棕色的岩石,水邊盛開著一朵粉紅色的小花。場景寧靜而奇妙,柔和的燈光增強了夢幻的氛圍。 | 生成的時間:40.85s(包含初次加載模型) |
![]() | A vibrant digital illustration depicts a chubby, young boy with short, tousled dark hair, large expressive brown eyes, and a cheerful smile. He wears a yellow polo shirt, gray shorts, and red sneakers, lifting a heavy barbell. The background is plain white, emphasizing the character. The style is reminiscent of Japanese anime with exaggerated features and a playful, energetic tone. | 一幅充滿活力的數位插圖描繪了一個胖乎乎的年輕男孩,他有一頭短短的、凌亂的黑髮,一雙大大的、富有表現力的棕色眼睛,臉上掛著燦爛的笑容。他穿著黃色 Polo 衫、灰色短褲和紅色運動鞋,舉著一根沉重的槓鈴。背景是純白色,突顯人物性格。這種風格讓人想起日本動漫,具有誇張的特徵和俏皮、充滿活力的基調。 | 生成的時間:19.7s |
![]() | This is a digital illustration by artist "Kot" featuring two young, anime-style characters standing close together, holding hands and looking at each other. They are both wearing backpacks and have expressive, large eyes. The boy on the left has messy black hair and is wearing a white shirt with a black vest. The girl on the right has short brown hair with bunny ears and is dressed in a red jacket. The background is a nighttime street scene with warm, glowing heart-shaped lights. There are other characters in the background, and the overall style is vibrant and whimsical. | 這是藝術家「Kot」創作的數位插圖,其中有兩個年輕的動漫風格人物緊密站在一起,手牽著手互相看著對方。他們都背著背包,有著富有表現力的大眼睛。左邊的男孩有著凌亂的黑髮,穿著白色襯衫和黑色背心。右邊的女孩有一頭棕色的短髮,戴著兔耳朵,穿著一件紅色夾克。背景是夜間街景,有溫暖、閃亮的心形燈光。背景還有其他人物,整體風格充滿活力而異想天開。 | 生成的時間:26.48s |
![]() | A muscular, shirtless man with short, dark hair and tanned skin is joyfully cooking in a large pot over an intense, roaring bonfire at night. His face is illuminated by the fiery flames, and water splashes dramatically from the pot. He is in a wilderness setting, surrounded by rocks and a dark, smoky background. The scene is vibrant with contrasting dark and bright colors, emphasizing the rugged, primal atmosphere. | 夜晚,一個肌肉發達、赤裸上身、留著黑色短髮、皮膚黝黑的男人,正高興地在熊熊燃燒的篝火旁的大鍋裡做飯。他的臉被熾熱的火焰照亮,水從鍋中劇烈地濺起。他身處荒野,周圍是岩石,背景是黑暗和煙霧繚繞。場景充滿活力,明暗色彩形成鮮明對比,強調了粗獷、原始的氛圍。 | 生成的時間:21.83s 錯誤的地方:在篝火旁的大鍋做飯 |
![]() | This is a highly detailed CGI image of a muscular, anthropomorphic bear balancing large rocks. The bear has a robust build with well-defined muscles, and its fur is a rich brown with lighter accents on its snout and paws. The bear is mid-stride on a rocky, grassy path with a cloudy, overcast sky in the background. The bear's expression is focused and determined. The rocks it balances are large and smooth, contrasting with the rugged terrain. The overall style is realistic with meticulous attention to textures and lighting. | 這是一張高度詳細的 CGI 圖像,描繪的是一隻肌肉發達的擬人化熊在平衡大石頭。這隻熊體格強健,肌肉發達,皮毛呈現深棕色,鼻子和爪子上的皮毛顏色較淺。這隻熊正大步走在一條佈滿岩石和草地的小路上,背景是陰雲密佈的天空。這隻熊的表情專注而堅定。它所平衡的岩石巨大而光滑,與崎嶇的地形形成鮮明對比。整體風格逼真,注重紋理和燈光。 | 生成的時間:25.35s 錯誤的地方:在平衡石頭 |
結論:5張圖有2張圖識別的錯誤失真大,會影響影片的生成。
2.Descriptive(Informal-非正式)模式
模式:Descriptive(Informal) | 輸出結果(英文) | 輸出結果(中文翻譯-Google) | 說明 |
![]() | A cute, chubby green frog with big, expressive eyes and orange spots sits in a round stone pool, surrounded by lush greenery and blooming flowers. The sky is bright blue with fluffy clouds, and a tiny butterfly flutters nearby. The frog's hands rest on its belly, and it gazes upward with a peaceful look. The whole scene is whimsical and charming, with a soft, glowing light. | 一隻胖胖的可愛綠色青蛙,大大的眼睛,身上有橙色的斑點,坐在一個圓形的石池裡,周圍是鬱鬱蔥蔥的綠樹和盛開的鮮花。天空湛藍,朵朵白雲飄灑,一隻小蝴蝶在附近翩翩起舞。青蛙的雙手放在腹部,以平靜的眼神凝視著上方。整個場景奇妙而迷人,燈光柔和而明亮。 | 生成的時間:21.16s |
![]() | A cute, chibi-style digital illustration of a young boy with messy black hair and big, expressive brown eyes. He's grinning widely while lifting a massive weight bar over his shoulders. The kid's rocking a yellow short-sleeved shirt with a small emblem, dark gray shorts, and red sneakers. His cheeks are slightly flushed, and he's surrounded by tiny, floating stone chunks, giving off a fun, playful vibe. The background is a plain, light gray, keeping the focus on the character. | 這是一幅可愛的 Q 版風格數位插畫,描繪的是一個有著凌亂黑髮和大大的、富有表現力的棕色眼睛的小男孩。他一邊笑容滿面,一邊將一根巨大的槓鈴舉過肩頭。這個孩子穿著一件帶有小徽章的黃色短袖襯衫、深灰色短褲和紅色運動鞋。他的臉頰微微泛紅,周圍漂浮著微小的石塊,散發出一種有趣、好玩的氛圍。背景是純色的淺灰色,焦點集中在人物身上。 | 生成的時間:23.91s 錯誤的地方:將一根巨大的槓鈴舉過肩頭 |
![]() | A cute, digital anime-style scene features two young characters, a boy and a girl, standing close together in a romantic moment. The boy rocks messy dark hair, a white shirt, and a brown vest, with a backpack on his shoulders. The girl flaunts short brown hair, a red jacket, and a brown backpack, with bunny ears on her head. They're holding hands, with a glowing red heart between them. The background is a twilight street scene with warm, orange lanterns and heart shapes, adding a whimsical vibe. | 這是一個可愛的數位動漫風格場景,其中有兩個年輕角色,一個男孩和一個女孩,親密地站在一起,享受浪漫的時刻。男孩留著凌亂的黑髮,穿著白襯衫和棕色背心,肩上背著背包。女孩留著棕色短髮,穿著紅色夾克,背著棕色背包,頭上戴著兔子耳朵。他們手牽手,中間有一顆閃閃發光的紅心。背景是黃昏的街景,有溫暖的橙色燈籠和心形,增添了異想天開的氛圍。 | 生成的時間:25.11s |
![]() | A ripped, shirtless dude with short brown hair is dunking his arm into a massive black pot filled with boiling water. The background is ablaze with towering flames, casting a fierce, orange glow over the scene. The guy's face shows pure determination and focus. The pot is on a rocky surface, with glowing embers scattered around. The whole vibe is intense and primal, like something straight out of a fantasy adventure. | 一個身材健美、赤裸上身、留著棕色短髮的男人正將手臂浸入一個裝滿沸水的巨大黑鍋中。背景中熊熊燃燒的火焰,給整個場景投下了猛烈的橙色光芒。這傢伙的臉上顯示出純粹的決心和專注。鍋子放在岩石表面,周圍散落著熾熱的餘燼。整個氛圍強烈而原始,就像直接來自奇幻冒險的場景。 | 生成的時間:22.66s 錯誤的地方:手臂浸入一個裝滿沸水的巨大黑鍋中 |
![]() | A buff, human-like bear stands on a rocky path, looking determined. With dark brown fur and a beefy build, it's got a chunky neck, broad shoulders, and thick limbs. Its face shows a focused expression, with small, round ears and a black snout. A big, smooth rock is balanced on its head, while another smaller rock is plopped on the ground nearby. The background features a cloudy sky and distant mountains, adding to the rugged vibe. The whole scene is a mix of realism and fantasy, blending nature with a fantastical twist. | 一隻體型類似人類的熊站在岩石小路上,一臉堅定的表情。它有著深棕色的皮毛和強健的體格,脖子粗短,肩膀寬闊,四肢粗壯。它的臉上帶著專注的表情,有小而圓的耳朵和黑色的鼻子。它的頭上平衡著一塊巨大、光滑的石頭,而另一塊較小的石頭則掉在附近的地面上。背景是多雲的天空和遠山,增添了粗獷的氛圍。整個場景融合了現實與幻想,將自然與奇幻融為一體。 | 生成的時間:25.53s 錯誤的地方:它的頭上平衡著一塊巨大、光滑的石頭 |
結論:這個錯誤率又更高了一些,5個當中有3個錯誤。
3.Training Prompt模式
模式:Training Prompt | 輸出結果(英文) | 輸出結果(中文翻譯-Google) | 說明 |
![]() | A digital illustration of a cute, cartoon-style green frog with large, expressive eyes, sitting in a small, round, stone-lined pond. The frog has a white belly and is holding its hands together in front of its chest. The background features a bright, sunny sky with fluffy white clouds and lush green foliage. The lighting is soft and warm, with a gentle glow on the frog. The image is vibrant and whimsical, with a playful and friendly atmosphere. The style is reminiscent of children's book illustrations. | 這是一張可愛的卡通風格綠色青蛙的數位插圖,它有一雙大而富有表現力的眼睛,坐在一個圓形的小石頭池塘里。這隻青蛙的腹部是白色的,雙手合十放在胸前。背景是明亮、晴朗的天空、蓬鬆的白雲和茂盛的綠葉。燈光柔和溫暖,青蛙身上散發著柔和的光芒。影像充滿活力且異想天開,充滿著活潑而友善的氛圍。這種風格讓人想起了兒童書籍插圖。 | 生成的時間:24.82s |
![]() | A cute, chibi-style cartoon of a young boy with short, spiky black hair and large, expressive eyes, wearing a yellow short-sleeved shirt and black shorts, holding a large barbell with both hands, smiling widely, with a joyful expression, standing on a plain gray background, with a few small rocks scattered around, dynamic pose, exaggerated proportions, bright and vibrant colors, high quality, digital art, anime style, jpeg artifacts | 一幅可愛的Q 版漫畫,描繪了一個小男孩,他有著短短的黑色尖刺發和一雙富有表現力的大眼睛,身穿黃色短袖襯衫和黑色短褲,雙手握著一根大槓鈴,笑容燦爛,表情快樂,站在純灰色的背景上,周圍散落著幾塊小石頭,動態的姿勢,誇張的比例,明亮而鮮豔的色彩,高品質,數字藝術,動漫風格,jpeg 文物 | 生成的時間:21.15s
|
![]() | A digital illustration in a vibrant, anime-inspired style features two young characters, a boy and a girl, standing close together in a romantic pose. The boy has dark hair and is wearing a white shirt and black vest, with a backpack slung over his shoulder. The girl has light brown hair and is wearing a red jacket with a gold emblem on the sleeve. They are holding a glowing heart-shaped object between them, with hearts floating around them in the background. The scene is set at sunset, with warm, orange and pink hues illuminating the sky and the buildings in the background. | 這幅充滿活力的動漫風格的數位插圖描繪了兩個年輕角色,一個男孩和一個女孩,以浪漫的姿勢緊密地站在一起。男孩有著黑色的頭髮,穿著白色襯衫和黑色背心,肩上背著一個背包。女孩有著淺棕色的頭髮,穿著一件紅色夾克,袖子上有金色的徽章。他們手裡拿著一個發光的心形物體,背景中,他們周圍漂浮著心形物體。場景設定在日落時分,溫暖的橙色和粉紅色調照亮了天空和背景中的建築。 | 生成的時間:28.08s |
![]() | Photograph of a muscular man with short dark hair, shirtless and wearing only a wristband, standing in a large black pot filled with water. He is energetically pouring water from the pot, creating a dramatic splash. The background is a large, intense fire, with flames towering above him, casting a warm orange glow. The scene is set outdoors, with rocks scattered around the base of the pot. The man's expression is one of intense concentration and excitement. The image has a dynamic and adventurous feel, with a sense of action and energy. | 照片中,一名肌肉發達的男子留著黑色短髮,赤裸上身,只戴著腕帶,站在裝滿水的大黑鍋裡。他正用力地從壺裡倒水,濺起巨大的水花。背景是一團巨大而強烈的火焰,火焰高聳於他之上,散發出溫暖的橙色光芒。場景設置在戶外,花盆底部周圍散落著岩石。男人的表情非常專注並且興奮。圖像給人一種動態和冒險的感覺,具有行動感和活力。 | 生成的時間:23.73s 錯誤的地方:
|
![]() | A hyper-realistic CGI image of a muscular, anthropomorphic bear standing on a dirt path. The bear is brown with a white muzzle and has a determined expression. It is balancing a large rock on its head with one paw. The background features a cloudy sky and distant mountains. The bear's fur is detailed and textured, giving it a lifelike appearance. The lighting is soft and diffused, creating a serene atmosphere. The bear's physique is robust and muscular, with defined muscles. The image is high quality and realistic, with a focus on the bear's strength and balance. | 一幅超現實的 CGI 影像,展現了一隻肌肉發達的擬人化熊站在泥路上。這隻熊呈棕色,有白色的口鼻部,表情堅定。它用一隻爪子在頭上平衡著一塊大石頭。背景是多雲的天空和遠處的山脈。熊的皮毛細緻而有紋理,看起來栩栩如生。燈光柔和而散射,營造出寧靜的氛圍。熊的體格健壯,肌肉發達,肌肉線條分明。影像品質高且逼真,重點展現了熊的力量和平衡。 | 生成的時間:25.74s 錯誤的地方:在頭上平衡著一塊大石頭 |
結論:跟第一個一樣,5張圖有2張圖識別的錯誤失真大,會影響影片的生成。描述風格跟第一個不太一樣。
4.MidJourney模式
模式:MidJourney | 輸出結果(英文) | 輸出結果(中文翻譯-Google) | 說明 |
![]() | cute, green frog, sitting in a small pond, digital art, anime style, bright sunlight, fluffy clouds, green leaves, pink flowers, sparkling water, frog has big round eyes, orange and black spots on its body, happy expression, soft textures, fantasy setting, lush vegetation, blue sky, serene atmosphere, detailed, colorful, whimsical, playful, enchanting, idyllic, nature, tranquility, adorable, charming, picturesque, detailed background, summer day, peaceful, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting, enchanting | 可愛,綠色的青蛙,坐在小池塘里,數位藝術,動漫風格,明亮的陽光,蓬鬆的雲彩,綠色的樹葉,粉紅色的花朵,波光粼粼的水,青蛙有大大的圓眼睛,身上有橙色和黑色的斑點,快樂的表情,柔軟紋理,幻想場景,茂密的植被,藍天,寧靜的氣氛,詳細,多彩,異想天開,好玩,迷人,田園詩般,自然,寧靜,可愛,迷人,風景如畫,詳細背景,夏日,平和,迷人,迷人,迷人,迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人、迷人 | 生成的時間:52.81s 錯誤:生成的內容有大量回文。 |
![]() | cute anime-style boy lifting weights, digital art, full body, short black hair with spiky tips, big brown eyes, blushing cheeks, wearing a yellow polo shirt, black shorts, red and white sneakers, holding a heavy barbell over his shoulders, light gray background, dynamic pose, expressive face, shiny highlights on the weights, some weights falling around, casual and energetic vibe, soft shading, playful and motivational, sporty theme, detailed textures on clothing and weights, character design, vibrant colors, cartoonish proportions, fun and inspiring atmosphere | 可愛的動漫風格男孩舉重,數位藝術,全身,黑色短髮,尖尖的,大大的棕色眼睛,紅潤的臉頰,穿著黃色馬球衫、黑色短褲、紅白運動鞋,肩上扛著一根沉重的槓鈴,淺灰色的背景,動態的姿勢,富有表現力的臉,重物上的閃亮亮點,一些重物掉落在周圍,休閒而充滿活力的氛圍,柔和的陰影,好玩而激勵,運動主題,衣服和重物上的細節紋理,人物設計,鮮豔的色彩,卡通比例、有趣和鼓舞人心的氛圍 | 生成的時間:24.9s 錯誤的地方: 1.肩上扛著一根沉重的槓鈴 2.一些重物掉落在周圍 |
![]() | anime-style, cute couple, boy and girl, holding heart-shaped light, romantic, boy with short dark hair, brown vest, white shirt, backpack, girl with short brown hair, bunny ears, red jacket, white shirt, both smiling, warm sunset, glowing heart lights, background with blurred people, lanterns, warm colors, digital art, soft lighting, whimsical, heart theme, love, dreamy atmosphere, detailed expressions, close-up, fantasy elements, heart-shaped light glowing between them, cozy street scene, magical, sweet moment, anime-inspired, romantic fantasy, tender interaction, heartwarming, affectionate, charming, kawaii, heart-shaped light in foreground, golden hour, soft shadows, gentle, tender, heart symbol, love, heartwarming, anime couple, heartwarming moment, romantic anime, love, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple, anime couple | 動漫風格,可愛的情侶,男孩和女孩,拿著心形燈,浪漫,黑色短髮男孩,棕色背心,白色襯衫,背包,棕色短髮女孩,兔耳朵,紅色夾克,白色襯衫,兩人都面帶微笑、溫暖的夕陽、發光的心形燈、模糊的人的背景、燈籠、暖色、數位藝術、柔和的燈光、異想天開、心形主題、愛、夢幻氛圍、細緻的表達、特寫、奇幻元素、心形燈之間發光他們,舒適的街景,神奇的,甜蜜的時刻,動漫風格,浪漫的幻想,溫柔的互動,溫馨的,深情的,迷人的,卡哇伊的,前景中的心形光,黃金時段,柔和的陰影,溫柔的,溫柔的,心形符號,愛,溫馨,動漫情侶,溫馨時刻,浪漫動漫,愛情,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶、動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶,動漫情侶 | 生成的時間:51.1s 錯誤:生成的內容有大量回文。 |
![]() | photograph of a shirtless muscular man with short dark hair, in the middle of a large cauldron filled with boiling water, intense expression, flames rising around him, dramatic lighting, orange and yellow hues, cauldron on rocky ground, water splashing, man reaching out with one arm, dynamic pose, fantasy, action, high contrast, dark background, rugged terrain, night scene, mystical, horror, adventure, intense atmosphere, cinematic, rugged, intense, dramatic, chaotic, vivid colors, high energy, fantasy genre, thrilling, intense, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, thrilling, fantasy, adventure, horror, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, thrilling, fantasy, adventure, horror, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, thrilling, fantasy, adventure, horror, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, thrilling, fantasy, adventure, horror, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, thrilling, fantasy, adventure, horror, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, thrilling, fantasy, adventure, horror, captivating, immersive, atmospheric, powerful, evocative, surreal, captivating, intense, dramatic, | 照片中,一名赤裸上身、留著黑色短髮的肌肉男站在一口裝滿沸水的大鍋中間,表情緊張,火焰在他周圍升起,燈光絢麗,色調為橙色和黃色,大鍋放在岩石地面上,水花四濺,男子伸出手去單臂、動態姿勢、幻想、動作、高對比度、深色背景、崎嶇地形、夜景、神秘、恐怖、冒險、強烈氛圍、電影、崎嶇、激烈、戲劇性、混亂、鮮豔色彩、高能量、幻想類型, 驚險刺激, 強烈, 迷人, 身臨其境, 大氣, 強大, 令人回味, 超現實主義, 迷人, 強烈, 戲劇性, 驚險刺激, 幻想, 冒險, 恐怖,迷人, 身臨其境, 大氣, 強大, 令人回味, 超現實主義, 迷人, 強烈, 戲劇性, 驚險刺激,幻想,冒險,恐怖,迷人,身臨其境,大氣,強大,令人回味,超現實,迷人,強烈,戲劇性,驚險,幻想,冒險,恐怖,迷人,身臨其境,大氣,強大,令人回味,超現實,迷人,強烈,戲劇性,驚險,幻想,冒險,恐怖,迷人,身臨其境,大氣,強大,令人回味,超現實,迷人,強烈,戲劇性,驚險,幻想,冒險,恐怖,迷人,身臨其境,大氣,強大,令人回味,超現實,迷人,強烈,戲劇性的,驚險的,幻想的,冒險的,恐怖的,迷人的,身臨其境的,大氣的,強大的,令人回味的,超現實的,迷人的,強烈的,戲劇性的, | 生成的時間:54.15s 特點:描述的倒是正確。內容偏關鍵字風格。 錯誤:生成的內容有大量回文。 |
![]() | photorealistic, anthropomorphic bear, muscular build, large paws, brown fur, standing on rocky path, balancing rock on head, grassy terrain, cloudy sky, realistic textures, detailed fur, powerful stance, strong physique, outdoors, nature setting, dramatic pose, natural lighting, dynamic composition, realistic CGI, mountainous background, serene atmosphere, rugged landscape, no watermarks, no visible genitals, focus on bear's upper body, natural environment, balanced rock, powerful and majestic, high detail, impressive realism, no clothing, no accessories, realistic shadows, realistic highlights, photorealistic art, photorealistic style, photorealistic CGI, photorealistic rendering | 逼真、擬人化的熊、肌肉發達、大爪子、棕色皮毛、站在岩石小路上、頭頂平衡岩石、草地、多雲的天空、逼真的紋理、細緻的皮毛、強大的姿態、強健的體格、戶外、自然環境、戲劇性的姿勢,自然光照、動態構圖、逼真的CGI、山區背景、寧靜的氛圍、崎嶇的景觀、無水印、無可見生殖器、焦點在熊的上半身、自然環境、平衡的岩石、強大而雄偉、高細節、令人印象深刻的真實感、沒有衣服,無配件、逼真的陰影、逼真的高光、照片級寫實藝術、照片級寫實風格、照片級寫實 CGI、照片級寫實渲染 | 生成的時間:28.04s 錯誤的地方:頭頂平衡岩石 |
結論:出現回文的比例過高,輸出的內容偏關鍵字形式。生成時間略慢。意外的是第四組有正確識別:肌肉男站在一口裝滿沸水的大鍋中間。但第五組目前皆失敗。
5.Booru tag list模式
模式:Booru tag list | 輸出結果(英文) | 輸出結果(中文翻譯-Google) | 說明 |
![]() | digital art, frog, cute, anthropomorphic, cartoon style, sitting, pond, water, sunlight, flowers, green, brown, white, blue sky, clouds, leaves, branches, cute expression, big eyes, pink flower, sparkles, peaceful, nature, whimsical, detailed, bright colors, fantasy, outdoors, playful, relaxing, calm, serene, small hands, smiling, light reflections, detailed background, semi-realistic, lush vegetation, idyllic setting, soft shading, vibrant, friendly, summer, spring, idyllic scene, fantasy setting, nature scenery, idyllic environment | 數位藝術,青蛙,可愛,擬人化,卡通風格,坐著,池塘,水,陽光,花朵,綠色,棕色,白色,藍天,雲彩,樹葉,樹枝,可愛的表情,大眼睛,粉紅色的花朵,閃閃發光,寧靜,自然、異想天開、細緻、鮮豔色彩、幻想、戶外、好玩、放鬆、冷靜、安詳、小手、微笑、光反射、詳細背景、半現實、茂密的植被、田園風光、柔和的陰影、充滿活力、友善、夏季、春天、田園風光、奇幻場景、自然風光、田園環境 | 生成的時間:27.44s |
![]() | photograph, CGI, 3D art, chibi style, cute, child, boy, short hair, black hair, brown eyes, big eyes, wide eyes, open mouth, smiling, teeth, blush, sweat, shirt, yellow shirt, short sleeves, collar, black shorts, sneakers, red sneakers, white socks, lifting weights, barbell, dumbbell, weightlifting, motion blur, dynamic pose, light gray background, white floor, dust, floating dust, action, energetic, playful, determined, fitness, workout, exercise, digital art, anime style, exaggerated features, rounded shapes, shiny textures, cartoonish, vibrant colors, stylized, no watermark | 照片、CGI、3D 藝術、Q風格、可愛、兒童、男孩、短髮、黑髮、棕色眼睛、大眼睛、大眼睛、張開嘴、微笑、牙齒、臉紅、汗水、襯衫、黃色襯衫、短袖、衣領、黑色短褲、運動鞋、紅色運動鞋、白色襪子、舉重、槓鈴、啞鈴、舉重、運動模糊、動態姿勢、淺灰色背景、白色地板、灰塵、浮塵、動作、精力充沛、俏皮、堅定、健身、鍛鍊、運動、數位藝術、動漫風格、誇張的特徵、圓潤的形狀、閃亮的紋理、卡通、鮮豔的色彩、風格化、無水印 | 生成的時間:28.96s |
![]() | photograph, digital art, anime style, two characters, boy, girl, standing, facing each other, holding heart-shaped light, romantic, cute, short hair, black hair, brown hair, bangs, big eyes, blushing, white shirt, black vest, brown backpack, red coat, white shirt, black pants, bunny ears, hair accessory, flowers, lanterns, festival, night, warm lighting, soft lighting, background characters, smiling, couple, love, affectionate, detailed, vibrant colors, digital medium, fantasy, whimsical, playful, affectionate gaze, heart symbols, dreamy atmosphere, romantic moment, fantasy setting, warm colors, soft textures, detailed background, glowing hearts, romantic gesture | 照片,數位藝術,動漫風格,兩個角色,男孩,女孩,站立,面對面,拿著心形燈,浪漫,可愛,短髮,黑髮,棕色頭髮,劉海,大眼睛,臉紅,白襯衫,黑色背心、棕色背包、紅色外套、白色襯衫、黑色褲子、兔子耳朵、髮飾、鮮花、燈籠、節慶、夜晚、暖光、柔光、背景人物、微笑、情侶、愛、親熱、細緻、鮮豔的色彩、數位媒體、幻想、異想天開、俏皮、深情凝視、心形符號、夢幻氛圍、浪漫時刻、奇幻場景、溫暖的色彩、柔和的紋理、細緻的背景、發光的心、浪漫的姿態 | 生成的時間:30.67s |
![]() | photograph, man, cauldron, fire, water, splashing, night, dramatic lighting, intense expression, muscular, shirtless, dark hair, wet skin, outdoor, forest, rocks, flames, orange glow, action shot, dynamic pose, horror theme, medieval, fantasy, adventure, rugged terrain, black cauldron, intense atmosphere, splash effect, water droplets, strong physique, fantasy setting, dramatic scene, water interaction, fire and water contrast, survival theme, wild setting, adventure genre, dark background, vivid colors, intense contrast | 照片,男人,大鍋,火,水,飛濺,夜晚,戲劇性的燈光,激烈的表情,肌肉發達,赤裸上身,黑髮,濕潤的皮膚,戶外,森林,岩石,火焰,橙色光芒,動作鏡頭,動態姿勢,恐怖主題,中世紀、奇幻、冒險、崎嶇地形、黑色大鍋、激烈氛圍、飛濺效果、水滴、強健體魄、奇幻場景、戲劇場景、水的互動、水火對比、生存主題、野外環境、冒險類型、黑暗背景、色彩鮮豔,對比強烈 | 生成的時間:24.92s |
![]() | photograph, realistic, CGI, bear, anthropomorphic, muscular, male, brown fur, large paws, standing, balancing rocks, outdoors, nature, grass, mountains, cloudy sky, realistic textures, detailed fur, digital art, fantasy, strong, powerful, nature setting, adventure, wilderness, rugged terrain, large bear, large boulders, balancing act, majestic, strong physique, outdoor adventure, nature scene, realistic rendering, detailed background, high detail, CGI realism, fantasy creature, realistic lighting, soft shadows, realistic environment, impressive, powerful pose, dynamic scene, nature photography, fantasy realism, environmental detail, detailed landscape, realistic proportions, realistic textures, realistic lighting, high-quality CGI, fantasy art, realistic fantasy, anthropomorphic animal, nature fantasy, realistic fantasy creature, nature adventure, fantasy landscape, realistic nature scene, realistic nature photography, fantasy realism, fantasy art, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy, realistic fantasy art, realistic fantasy creature, realistic fantasy | 照片、逼真、CGI、熊、擬人化、肌肉發達、男性、棕色皮毛、大爪子、站立、平衡岩石、戶外、自然、草地、山脈、多雲的天空、逼真的紋理、細緻的皮毛、數位藝術、幻想、強壯、有力、自然環境、冒險、荒野、崎嶇地形、大熊、巨石、平衡動作、雄偉、強健的體格、戶外探險、自然風光、逼真的渲染、細緻的背景、高細節、CGI 真實感、幻想生物、逼真的燈光、柔和陰影、逼真的環境、令人印象深刻、強有力的姿勢、動態場景、自然攝影、奇幻現實主義、環境細節、細緻的景觀、逼真的比例、逼真的紋理、逼真的燈光、高品質CGI、幻想藝術、逼真的幻想、擬人化動物、自然幻想、逼真的幻想生物、自然冒險、幻想景觀、逼真的自然場景、逼真的自然攝影、幻想現實主義、幻想藝術、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想、逼真的幻想藝術、逼真的幻想生物、逼真的幻想 | 生成的時間:52.63s 錯誤的地方:出現大量回文、男性 |
結論:這個回應都是關鍵字模式,用來當補充描述或許不錯。但最後一組出現回文。或許這個不能設定為long模式。(這個倒是可以跟其他的組合在一起用,作為增強特徵用)
礙於文章篇幅問題,剩下的拆分到下一篇
文章轉載或引用,請先告知並保留原文出處與連結!!(單純分享或非營利的只需保留原文出處,不用告知)
原文連結:
https://blog.aidec.tw/post/comfyui-image-description comparison
若有業務合作需求,可寫信至: opweb666@gmail.com
創業、網站經營相關內容未來將發布在 小易創業筆記