ComfyUI node for Qwen-Image-2.1: 17 photographic styles (camera language + mood) via a PE-T2I prompt rewriter. Not a model loader.
Python
17
9 commits
updated Sep 23, 2026
A single ComfyUI node — PE-T2I Photo Style Prompt — that turns a short prompt into a rich, styled one for Qwen-Image-2.1.
Pick a photographic style from a dropdown (17 styles), type a short prompt, and the node sends it to a PE-T2I rewriter (any OpenAI-compatible endpoint) together with that style's sheet. It returns a long English prompt, a matching negative prompt, and the width / height for the aspect ratio the rewriter chose.
Each style is described as camera language + mood — never a photographer's name. Your subject, and whatever text / counts / colours / positions you specify, stay yours; the style only decides what you left open. The shorter your prompt, the more the style shows through.
This is not a model loader. It only produces text (prompt / negative) and two integers (width / height). It has nothing to do with GGUF / NVFP4 / fp8 / bf16 — those only matter for the loader nodes downstream. Use it with any Qwen-Image-2.1 setup, in any format.
It stands at the very front and only shapes text: it takes your short line + the chosen style, has the PE-T2I rewriter expand it, and hands the finished prompt / negative / size to the ordinary Qwen-Image-2.1 nodes. It never loads a model or touches a pixel — think of it as the art director that writes the shot; the camera (the DiT) takes it.
your short line + style dropdown
|
v
+--------------------------------------+
| PE-T2I Photo Style Prompt (THIS) | <- text only, seconds, no VRAM
| short line + style sheet --> PE-T2I |
| out: long prompt . negative . size |
+--------------------------------------+
| prompt / negative | width / height
v v
TextEncodeQwenImage21 EmptyLatentImage
| (encode -> cond) | (make canvas)
+----------> KSampler (DiT denoise) <--------+
|
v
VAEDecode --> final image
Every image below is the whole prompt you write — pick a style, type one short line like the caption, and this is what comes out. Two photos and two "posters" per style; nothing was hand-tuned. (Generated with this node on a quantized Qwen-Image-2.1 setup.)
 photo 「午夜雨後漆黑的巷子，一隻野狗猛然回頭、雙眼直盯鏡頭」“a stray dog snaps its head around in a pitch-black alley after midnight rain” |  photo 「刺眼霓虹下擁擠的人潮，一張模糊的臉朝鏡頭踉蹌逼近」“a blurred face lurches toward the camera through a dense crowd under harsh neon” |  poster 「粗粒黑白的地下樂團演出海報，巨大斑駁的標題文字「NOISE」，下方一個模糊奔跑的身影」“a gritty B&W gig poster, huge distressed title “NOISE”, a blurred running figure” |  poster 「黑白街拍攝影展海報，頂端粗體標題「STREET」，中央一個逆光孤影被黑暗吞沒」“a B&W street-photography exhibition poster, bold title “STREET”, a lone backlit figure” |
 photo 「東京街頭，一道強烈夕陽斜光切過狹窄巷弄，一個行人拉出長長的影子，黑白高反差」“a Tokyo street, hard evening light slicing a narrow alley, a passer-by casting a long shadow (B&W)” |  photo 「長廊盡頭一個渺小的身影，一道光刃劈過整排廊柱」“a tiny figure at the end of a long arcade, a blade of light slicing the columns” |  poster 「極簡黑白攝影展海報，東京街頭一道光束與長影構成的幾何，大片留黑，底部小字標題「LIGHT AND SHADOW」」“a minimalist B&W exhibition poster, a Tokyo-street beam and long shadows, title “LIGHT AND SHADOW”” |  poster 「建築雙年展海報，一道光束與長影構成的幾何，頂端標題「幾何」」“an architecture-biennale poster, a beam and long shadows, title “Geometry”” |
 photo 「老農粗糙的雙手捧著秧苗，田邊一個孩子靜靜看著」“a farmer’s rough hands cupping seedlings, a child watching at the field’s edge” |  photo 「燈光下的傳統菜攤，一位老闆娘在垂掛的蔬菜之間笑著」“an old stall-keeper laughing among hanging vegetables under market lamplight” |  poster 「人文紀實攝影展海報，一張溫暖的黑白肖像，底部標題「土地與人」」“a humanist documentary exhibition poster, a warm B&W portrait, title “Land and People”” |  poster 「紀錄片影展海報，一雙勞動的手部特寫，上方標題「HANDS」」“a documentary film-festival poster, a close-up of working hands, title “HANDS”” |
 photo 「純白背景前，舞者在跳躍最高點被凝結的全身」“a dancer frozen at the peak of a leap, full body on pure white” |  photo 「佈滿深刻皺紋的臉部大特寫，每個毛孔清晰、直視鏡頭」“an extreme close-up of a deeply lined face, every pore sharp, eyes on the lens” |  poster 「純白極簡時尚海報，全身躍起的舞者，一側直排標題「LEAP」」“a pure-white minimalist fashion poster, a leaping dancer, vertical title “LEAP”” |  poster 「黑白肖像展海報，白底、銳利的臉孔特寫，頂端標題「FACE」」“a B&W portrait-show poster, white ground, a razor-sharp face, title “FACE”” |
 photo 「一位女子優雅地縮進空蕩的牆角，雙手安靜交疊」“a woman folded elegantly into a bare studio corner, hands quietly clasped” |  photo 「光滑的灰色背景前，一朵正在枯萎的鬱金香」“a single wilting tulip against a smooth grey seamless ground” |  poster 「極簡藝廊海報，大片留白，角落一個安靜的身影，小字標題「CORNER」」“a minimalist gallery poster, lots of white space, a quiet corner figure, title “CORNER”” |  poster 「靜物攝影集海報，灰底一朵枯萎鬱金香，底部細字標題「STILL」」“a still-life photobook poster, grey ground with a wilting tulip, title “STILL”” |
 photo 「深夜泛光燈的泳池邊，一位穿高跟鞋、氣場強大的女子全身站姿」“a commanding woman in heels, full figure, at the edge of a floodlit midnight pool” |  photo 「霓虹濕滑的街道旁，一位模特兒冷冷地倚著一輛跑車」“a model leaning coldly against a sports car on a neon-wet street” |  poster 「高冷時尚雜誌封面，全身女子強勢站姿，頂端刊名「NOIR」與幾行小字」“a high-fashion magazine cover, a full-figure power stance, masthead “NOIR”” |  poster 「黑白時尚海報，大理石長階上俯視鏡頭走下的高挑女子，直排標題「POWER」」“a B&W fashion poster, a woman descending marble stairs, vertical title “POWER”” |
 photo 「寫實時尚雜誌大片，一位全身模特兒穿著剪裁俐落的當代設計師禮服，站在乾淨棚拍佈景前，強勢時尚站姿」“a photorealistic editorial shot, a full-figure model in a sleek contemporary designer gown, power stance” |  photo 「寫實時尚攝影，全身模特兒穿著合身絲緞套裝，乾淨棚燈，濃艷珠寶色背景，醒目時尚站姿」“a photorealistic fashion shot, a model in a fitted satin suit, clean studio, jewel-tone backdrop” |  poster 「寫實時尚雜誌封面，全身模特兒穿當代高訂禮服，乾淨棚拍珠寶色背景，頂部放簡潔刊名「CHIC」」“a photoreal fashion magazine cover, a full-figure model in modern haute couture, masthead “CHIC”” |  poster 「寫實高彩時尚海報，全身模特兒穿極簡結構感當代禮服，珠寶色背景，一側放簡潔標題「CHIC EAST」」“a high-colour fashion poster, a model in a minimal structured gown, side title “CHIC EAST”” |
 photo 「雜亂的陽台上一小束紅花，襯著泛紫的黃昏天空」“a small bunch of red flowers on a cluttered balcony against a bruised evening sky” |  photo 「剛睡醒的人半裹在被單裡，被近距離的硬閃燈打亮」“someone just woken, half-wrapped in sheets, lit by a hard close flash” |  poster 「私寫真攝影集海報，閃燈直打的親密快照，底部手寫感標題「日記」」“a private-snapshot photobook poster, an intimate flash snapshot, title “Diary”” |  poster 「獨立寫真展海報，過曝的閃燈近景，頂端標題「DIARY」」“an indie photo-show poster, an overexposed flash close-up, title “DIARY”” |
 photo 「起霧又佈滿雨痕的咖啡館玻璃後，一把孤獨的紅傘」“a lone red umbrella seen through a fogged, rain-streaked café window” |  photo 「起霧的窗上倒映著模糊的街景，只有一盞暖燈亮著」“a blurred street reflected in a steamed-up window, one warm light glowing” |  poster 「柔調彩色攝影展海報，隔著雨窗的模糊街景，底部細字標題「GLASS」」“a soft-colour exhibition poster, a blurred street behind a rainy window, title “GLASS”” |  poster 「文青影展海報，霧面玻璃後一抹紅，直排標題「雨」」“an art-house film poster, a splash of red behind frosted glass, title “Rain”” |
 photo 「藍色黃昏，空蕩郊區街道正中央一動不動站著的人」“a lone figure standing motionless in an empty suburban street at blue dusk” |  photo 「入夜無人的加油站，冷白燈光下獨自站著的女子」“a woman alone under the cold light of a deserted gas station at nightfall” |  poster 「電影感海報，藍調黃昏郊區、街心的孤立身影，底部片名「DUSK」與演職員小字」“a cinematic poster, a lone figure at blue dusk, film title “DUSK” with small credits” |  poster 「獨立電影海報，冷光加油站的孤獨女子，頂端片名「靜默」」“an indie film poster, a lone woman at a cold-lit gas station, title “Silence”” |
 photo 「一個渺小的人仰望比房子還高的巨大花朵，花瓣像陽傘」“a tiny figure gazing up at a flower taller than a house, petals like parasols” |  photo 「一個渺小的旅人走過一排比人還高的巨大蘑菇，夢幻粉彩佈景」“a tiny traveller walking past a row of giant mushrooms on a pastel set” |  poster 「夢幻時尚大片海報，巨大花朵旁的小小人物，花體標題「WONDER」」“a dreamy fashion-editorial poster, a tiny figure beside a giant flower, title “WONDER”” |  poster 「童話展覽海報，巨大茶杯與熟睡的女孩，頂端標題「巨物奇境」」“a fairytale-exhibition poster, a giant teacup and a sleeping girl, title “Land of Giants”” |
 photo 「午後柔光的房間裡逆光的孩子，塵埃在光中漂浮」“a backlit child in a soft afternoon room, dust drifting in the light” |  photo 「一個孩子小小的背影，朝著淡霧色的夏日海奔去」“the small back of a child running toward a pale, hazy summer sea” |  poster 「溫柔生活攝影集海報，逆光的孩子與漂浮的塵埃，底部細字「日常之光」」“a gentle lifestyle photobook poster, a backlit child and floating dust, title “Everyday Light”” |  poster 「文青攝影展海報，柔霧海邊奔跑的背影，頂端標題「夏日」」“a photo-exhibition poster, a small back running by a hazy sea, title “Summer”” |
 photo 「櫥窗疊映著三層街景，拍攝者的影子也落在其中」“a shop window stacking three layers of reflections, the photographer’s shadow among them” |  photo 「一根電線桿不偏不倚從路人頭頂長出來」“a lamp post growing precisely out of a passer-by’s head” |  poster 「街拍攝影展海報，層層反射疊映的街景，斜置標題「MIRROR」」“a street-photography exhibition poster, layered reflections, slanted title “MIRROR”” |  poster 「黑白攝影集海報，被窗框切開的臉與反射，頂端標題「REFLECT」」“a B&W photobook poster, a face sliced by a window frame, title “REFLECT”” |
 photo 「逆光明亮的葉尖，一顆將墜未墜、微微顫動的露珠」“a single trembling dew drop about to fall from a bright backlit leaf” |  photo 「一個孩子的小手，舉進蒼白到近乎過曝的晨光裡」“a child’s small hand lifted into pale, near-overexposed morning light” |  poster 「高調柔光攝影集海報，過曝晨光中的一顆露珠，底部淡字標題「微光」」“a high-key photobook poster, a dew drop in blown-out morning light, title “Faint Light”” |  poster 「極簡影展海報，蒼白光裡的一隻小手，頂端細字標題「LIGHT」」“a minimalist exhibition poster, a small hand in pale light, title “LIGHT”” |
 photo 「路人恰好被抓拍到一個湊巧荒謬的姿勢」“a passer-by caught in an accidentally absurd pose” |  photo 「一隻狗和主人並排，露出一模一樣的表情」“a dog and its owner side by side wearing the exact same expression” |  poster 「幽默攝影展海報，一個荒謬巧合的瞬間，大字標題「PUNCHLINE」」“a humour-photography exhibition poster, an absurd coincidence, title “PUNCHLINE”” |  poster 「生活抓拍集海報，狗與主人同款表情，頂端標題「日常笑點」」“a candid-life photobook poster, a dog and owner with matching faces, title “Everyday Punchline”” |
 photo 「濃到快要滿出來的緋紅與洋紅花海，金魚在其間穿游」“a drowning-dense field of crimson and magenta blooms with goldfish gliding through” |  photo 「一朵沾滿雨珠的紅色曼陀羅，飽和到近乎過度」“a single red datura flower beaded with rain, saturated to the edge of too-much” |  poster 「濃艷的當代藝術展海報，滿版飽和的花海與穿游的金魚，下方一條乾淨白色色帶放粗體標題「VIVID」」“a contemporary art poster, a full-bleed saturated flower field with goldfish, bold title “VIVID”” |  poster 「花卉攝影展海報，過飽和盛開的牡丹與發亮的金魚缸，頂部乾淨留白處放簡潔標題「BLOOM」」“a flower-photography poster, over-saturated peonies and a glowing goldfish bowl, title “BLOOM”” |
 photo 「陽光田野裡扮著鬼臉、頭髮飛揚跳起來的小孩」“a kid mid-goofy-face leaping through a sunlit field, hair flying” |  photo 「一個孩子咬著滴下汁的冰棒，笑得合不攏嘴」“a child laughing wide around a dripping popsicle” |  poster 「底片寫真集海報，跳躍扮鬼臉的孩子，手寫感標題「PLAY」」“a film-snapshot photobook poster, a kid leaping and pulling a face, title “PLAY”” |  poster 「夏日寫真展海報，咬著冰棒大笑的孩子，頂端標題「夏」」“a summer photo-show poster, a child laughing with a popsicle, title “Summer”” |
TextEncodeQwenImage21 and the
UNet / CLIP / VAE loaders you use)./v1/chat/completions endpoint, serving
Qwen/Qwen-Image-2.1-PE-T2I (any backend — vLLM, llama-server, LM Studio, ...; any weight
format). Point the node's endpoint at it (default http://127.0.0.1:8207).No Python dependencies to install.
cd ComfyUI/custom_nodes
git clone https://github.com/pottokao-dotcom/ComfyUI-QwenImage-PhotoStyles
Or install from the ComfyUI Registry via ComfyUI Manager (search "Qwen-Image Photo Styles").
Inputs — prompt (multiline), style (dropdown: None + 17 styles), endpoint, seed,
temperature; optional model, timeout_s, system_prompt_path.
Outputs — prompt (STRING), negative (STRING), width (INT), height (INT),
wh_ratio (STRING). seed also picks one of the style's mood sub-presets, so re-rolling it
varies the feeling.
Black Fury · Geometry of Light · Warm Documentary · White Leap · Corner Elegance · Cold Power · Hyper Chic East · Private Diary · Color Behind Glass · Frozen Film Still · Fairytale Giants · Soft Everyday Light · Street in the Mirror · Pale Light Moments · Everyday Punchline · Vivid Garden · Playful Film.
Each style sheet in styles/*.txt specifies, per style: Look, Lens language, Light, Emotion,
People, Scene & props, Colour, plus per-style negatives and — for film-look styles — a fixed
grain clause.
This node only needs the PE-T2I rewriter; everything else is your own Qwen-Image-2.1 graph. If you want a self-quantized set to run the whole thing, here are 4-bit builds — NVFP4 for NVIDIA (ComfyUI / vLLM) and GGUF Q4 for Mac / low-VRAM (llama.cpp / ComfyUI-GGUF). ("Heretic" = an abliterated / decensored build.)
| Component | NVFP4 · NVIDIA (ComfyUI / vLLM) | GGUF Q4 · Mac (llama.cpp / ComfyUI-GGUF) |
|---|---|---|
| PE-T2I rewriter (this node calls) | Qwen-Image-2.1-PE-T2I-Heretic-NVFP4 | Qwen-Image-2.1-PE-T2I-Heretic-GGUF |
| DiT (image model) | Qwen-Image-2.1-DiT-NVFP4-ComfyUI | Qwen-Image-2.1-DiT-GGUF |
| Text encoder (Qwen3-VL) | Qwen-Image-2.1-Text-Encoder-Heretic-NVFP4 | Qwen-Image-2.1-Text-Encoder-Heretic-GGUF |
The Qwen3-VL text-encoder GGUF needs this add-on in ComfyUI (fixes the 12288 shape error): https://github.com/pottokao-dotcom/ComfyUI-GGUF-Qwen3VL-TE
See example_workflow_api.json (API format — enable Dev mode in ComfyUI settings, then
Load (API Format)). The loader nodes and model file names there are just an example; swap in
whatever Qwen-Image-2.1 models and format you use.
presets.json, docs — MIT (see LICENSE).style_sp_plus.txt is adapted from Qwen's official Qwen-Image-2.1-PE-T2I system prompt and
remains under the Qwen Research License (see NOTICE). Review it before commercial use.ComfyUI node for Qwen-Image-2.1: 17 photographic styles (camera language + mood) via a PE-T2I prompt rewriter. Not a model loader.
Python
17
9 commits
updated Sep 23, 2026
A single ComfyUI node — PE-T2I Photo Style Prompt — that turns a short prompt into a rich, styled one for Qwen-Image-2.1.
Pick a photographic style from a dropdown (17 styles), type a short prompt, and the node sends it to a PE-T2I rewriter (any OpenAI-compatible endpoint) together with that style's sheet. It returns a long English prompt, a matching negative prompt, and the width / height for the aspect ratio the rewriter chose.
Each style is described as camera language + mood — never a photographer's name. Your subject, and whatever text / counts / colours / positions you specify, stay yours; the style only decides what you left open. The shorter your prompt, the more the style shows through.
This is not a model loader. It only produces text (prompt / negative) and two integers (width / height). It has nothing to do with GGUF / NVFP4 / fp8 / bf16 — those only matter for the loader nodes downstream. Use it with any Qwen-Image-2.1 setup, in any format.
It stands at the very front and only shapes text: it takes your short line + the chosen style, has the PE-T2I rewriter expand it, and hands the finished prompt / negative / size to the ordinary Qwen-Image-2.1 nodes. It never loads a model or touches a pixel — think of it as the art director that writes the shot; the camera (the DiT) takes it.
your short line + style dropdown
|
v
+--------------------------------------+
| PE-T2I Photo Style Prompt (THIS) | <- text only, seconds, no VRAM
| short line + style sheet --> PE-T2I |
| out: long prompt . negative . size |
+--------------------------------------+
| prompt / negative | width / height
v v
TextEncodeQwenImage21 EmptyLatentImage
| (encode -> cond) | (make canvas)
+----------> KSampler (DiT denoise) <--------+
|
v
VAEDecode --> final image
Every image below is the whole prompt you write — pick a style, type one short line like the caption, and this is what comes out. Two photos and two "posters" per style; nothing was hand-tuned. (Generated with this node on a quantized Qwen-Image-2.1 setup.)
photo 「午夜雨後漆黑的巷子，一隻野狗猛然回頭、雙眼直盯鏡頭」“a stray dog snaps its head around in a pitch-black alley after midnight rain” | photo 「刺眼霓虹下擁擠的人潮，一張模糊的臉朝鏡頭踉蹌逼近」“a blurred face lurches toward the camera through a dense crowd under harsh neon” | poster 「粗粒黑白的地下樂團演出海報，巨大斑駁的標題文字「NOISE」，下方一個模糊奔跑的身影」“a gritty B&W gig poster, huge distressed title “NOISE”, a blurred running figure” | poster 「黑白街拍攝影展海報，頂端粗體標題「STREET」，中央一個逆光孤影被黑暗吞沒」“a B&W street-photography exhibition poster, bold title “STREET”, a lone backlit figure” |
photo 「東京街頭，一道強烈夕陽斜光切過狹窄巷弄，一個行人拉出長長的影子，黑白高反差」“a Tokyo street, hard evening light slicing a narrow alley, a passer-by casting a long shadow (B&W)” | photo 「長廊盡頭一個渺小的身影，一道光刃劈過整排廊柱」“a tiny figure at the end of a long arcade, a blade of light slicing the columns” | poster 「極簡黑白攝影展海報，東京街頭一道光束與長影構成的幾何，大片留黑，底部小字標題「LIGHT AND SHADOW」」“a minimalist B&W exhibition poster, a Tokyo-street beam and long shadows, title “LIGHT AND SHADOW”” | poster 「建築雙年展海報，一道光束與長影構成的幾何，頂端標題「幾何」」“an architecture-biennale poster, a beam and long shadows, title “Geometry”” |
photo 「老農粗糙的雙手捧著秧苗，田邊一個孩子靜靜看著」“a farmer’s rough hands cupping seedlings, a child watching at the field’s edge” | photo 「燈光下的傳統菜攤，一位老闆娘在垂掛的蔬菜之間笑著」“an old stall-keeper laughing among hanging vegetables under market lamplight” | poster 「人文紀實攝影展海報，一張溫暖的黑白肖像，底部標題「土地與人」」“a humanist documentary exhibition poster, a warm B&W portrait, title “Land and People”” | poster 「紀錄片影展海報，一雙勞動的手部特寫，上方標題「HANDS」」“a documentary film-festival poster, a close-up of working hands, title “HANDS”” |
photo 「純白背景前，舞者在跳躍最高點被凝結的全身」“a dancer frozen at the peak of a leap, full body on pure white” | photo 「佈滿深刻皺紋的臉部大特寫，每個毛孔清晰、直視鏡頭」“an extreme close-up of a deeply lined face, every pore sharp, eyes on the lens” | poster 「純白極簡時尚海報，全身躍起的舞者，一側直排標題「LEAP」」“a pure-white minimalist fashion poster, a leaping dancer, vertical title “LEAP”” | poster 「黑白肖像展海報，白底、銳利的臉孔特寫，頂端標題「FACE」」“a B&W portrait-show poster, white ground, a razor-sharp face, title “FACE”” |
photo 「一位女子優雅地縮進空蕩的牆角，雙手安靜交疊」“a woman folded elegantly into a bare studio corner, hands quietly clasped” | photo 「光滑的灰色背景前，一朵正在枯萎的鬱金香」“a single wilting tulip against a smooth grey seamless ground” | poster 「極簡藝廊海報，大片留白，角落一個安靜的身影，小字標題「CORNER」」“a minimalist gallery poster, lots of white space, a quiet corner figure, title “CORNER”” | poster 「靜物攝影集海報，灰底一朵枯萎鬱金香，底部細字標題「STILL」」“a still-life photobook poster, grey ground with a wilting tulip, title “STILL”” |
photo 「深夜泛光燈的泳池邊，一位穿高跟鞋、氣場強大的女子全身站姿」“a commanding woman in heels, full figure, at the edge of a floodlit midnight pool” | photo 「霓虹濕滑的街道旁，一位模特兒冷冷地倚著一輛跑車」“a model leaning coldly against a sports car on a neon-wet street” | poster 「高冷時尚雜誌封面，全身女子強勢站姿，頂端刊名「NOIR」與幾行小字」“a high-fashion magazine cover, a full-figure power stance, masthead “NOIR”” | poster 「黑白時尚海報，大理石長階上俯視鏡頭走下的高挑女子，直排標題「POWER」」“a B&W fashion poster, a woman descending marble stairs, vertical title “POWER”” |
photo 「寫實時尚雜誌大片，一位全身模特兒穿著剪裁俐落的當代設計師禮服，站在乾淨棚拍佈景前，強勢時尚站姿」“a photorealistic editorial shot, a full-figure model in a sleek contemporary designer gown, power stance” | photo 「寫實時尚攝影，全身模特兒穿著合身絲緞套裝，乾淨棚燈，濃艷珠寶色背景，醒目時尚站姿」“a photorealistic fashion shot, a model in a fitted satin suit, clean studio, jewel-tone backdrop” | poster 「寫實時尚雜誌封面，全身模特兒穿當代高訂禮服，乾淨棚拍珠寶色背景，頂部放簡潔刊名「CHIC」」“a photoreal fashion magazine cover, a full-figure model in modern haute couture, masthead “CHIC”” | poster 「寫實高彩時尚海報，全身模特兒穿極簡結構感當代禮服，珠寶色背景，一側放簡潔標題「CHIC EAST」」“a high-colour fashion poster, a model in a minimal structured gown, side title “CHIC EAST”” |
photo 「雜亂的陽台上一小束紅花，襯著泛紫的黃昏天空」“a small bunch of red flowers on a cluttered balcony against a bruised evening sky” | photo 「剛睡醒的人半裹在被單裡，被近距離的硬閃燈打亮」“someone just woken, half-wrapped in sheets, lit by a hard close flash” | poster 「私寫真攝影集海報，閃燈直打的親密快照，底部手寫感標題「日記」」“a private-snapshot photobook poster, an intimate flash snapshot, title “Diary”” | poster 「獨立寫真展海報，過曝的閃燈近景，頂端標題「DIARY」」“an indie photo-show poster, an overexposed flash close-up, title “DIARY”” |
photo 「起霧又佈滿雨痕的咖啡館玻璃後，一把孤獨的紅傘」“a lone red umbrella seen through a fogged, rain-streaked café window” | photo 「起霧的窗上倒映著模糊的街景，只有一盞暖燈亮著」“a blurred street reflected in a steamed-up window, one warm light glowing” | poster 「柔調彩色攝影展海報，隔著雨窗的模糊街景，底部細字標題「GLASS」」“a soft-colour exhibition poster, a blurred street behind a rainy window, title “GLASS”” | poster 「文青影展海報，霧面玻璃後一抹紅，直排標題「雨」」“an art-house film poster, a splash of red behind frosted glass, title “Rain”” |
photo 「藍色黃昏，空蕩郊區街道正中央一動不動站著的人」“a lone figure standing motionless in an empty suburban street at blue dusk” | photo 「入夜無人的加油站，冷白燈光下獨自站著的女子」“a woman alone under the cold light of a deserted gas station at nightfall” | poster 「電影感海報，藍調黃昏郊區、街心的孤立身影，底部片名「DUSK」與演職員小字」“a cinematic poster, a lone figure at blue dusk, film title “DUSK” with small credits” | poster 「獨立電影海報，冷光加油站的孤獨女子，頂端片名「靜默」」“an indie film poster, a lone woman at a cold-lit gas station, title “Silence”” |
photo 「一個渺小的人仰望比房子還高的巨大花朵，花瓣像陽傘」“a tiny figure gazing up at a flower taller than a house, petals like parasols” | photo 「一個渺小的旅人走過一排比人還高的巨大蘑菇，夢幻粉彩佈景」“a tiny traveller walking past a row of giant mushrooms on a pastel set” | poster 「夢幻時尚大片海報，巨大花朵旁的小小人物，花體標題「WONDER」」“a dreamy fashion-editorial poster, a tiny figure beside a giant flower, title “WONDER”” | poster 「童話展覽海報，巨大茶杯與熟睡的女孩，頂端標題「巨物奇境」」“a fairytale-exhibition poster, a giant teacup and a sleeping girl, title “Land of Giants”” |
photo 「午後柔光的房間裡逆光的孩子，塵埃在光中漂浮」“a backlit child in a soft afternoon room, dust drifting in the light” | photo 「一個孩子小小的背影，朝著淡霧色的夏日海奔去」“the small back of a child running toward a pale, hazy summer sea” | poster 「溫柔生活攝影集海報，逆光的孩子與漂浮的塵埃，底部細字「日常之光」」“a gentle lifestyle photobook poster, a backlit child and floating dust, title “Everyday Light”” | poster 「文青攝影展海報，柔霧海邊奔跑的背影，頂端標題「夏日」」“a photo-exhibition poster, a small back running by a hazy sea, title “Summer”” |
photo 「櫥窗疊映著三層街景，拍攝者的影子也落在其中」“a shop window stacking three layers of reflections, the photographer’s shadow among them” | photo 「一根電線桿不偏不倚從路人頭頂長出來」“a lamp post growing precisely out of a passer-by’s head” | poster 「街拍攝影展海報，層層反射疊映的街景，斜置標題「MIRROR」」“a street-photography exhibition poster, layered reflections, slanted title “MIRROR”” | poster 「黑白攝影集海報，被窗框切開的臉與反射，頂端標題「REFLECT」」“a B&W photobook poster, a face sliced by a window frame, title “REFLECT”” |
photo 「逆光明亮的葉尖，一顆將墜未墜、微微顫動的露珠」“a single trembling dew drop about to fall from a bright backlit leaf” | photo 「一個孩子的小手，舉進蒼白到近乎過曝的晨光裡」“a child’s small hand lifted into pale, near-overexposed morning light” | poster 「高調柔光攝影集海報，過曝晨光中的一顆露珠，底部淡字標題「微光」」“a high-key photobook poster, a dew drop in blown-out morning light, title “Faint Light”” | poster 「極簡影展海報，蒼白光裡的一隻小手，頂端細字標題「LIGHT」」“a minimalist exhibition poster, a small hand in pale light, title “LIGHT”” |
photo 「路人恰好被抓拍到一個湊巧荒謬的姿勢」“a passer-by caught in an accidentally absurd pose” | photo 「一隻狗和主人並排，露出一模一樣的表情」“a dog and its owner side by side wearing the exact same expression” | poster 「幽默攝影展海報，一個荒謬巧合的瞬間，大字標題「PUNCHLINE」」“a humour-photography exhibition poster, an absurd coincidence, title “PUNCHLINE”” | poster 「生活抓拍集海報，狗與主人同款表情，頂端標題「日常笑點」」“a candid-life photobook poster, a dog and owner with matching faces, title “Everyday Punchline”” |
photo 「濃到快要滿出來的緋紅與洋紅花海，金魚在其間穿游」“a drowning-dense field of crimson and magenta blooms with goldfish gliding through” | photo 「一朵沾滿雨珠的紅色曼陀羅，飽和到近乎過度」“a single red datura flower beaded with rain, saturated to the edge of too-much” | poster 「濃艷的當代藝術展海報，滿版飽和的花海與穿游的金魚，下方一條乾淨白色色帶放粗體標題「VIVID」」“a contemporary art poster, a full-bleed saturated flower field with goldfish, bold title “VIVID”” | poster 「花卉攝影展海報，過飽和盛開的牡丹與發亮的金魚缸，頂部乾淨留白處放簡潔標題「BLOOM」」“a flower-photography poster, over-saturated peonies and a glowing goldfish bowl, title “BLOOM”” |
photo 「陽光田野裡扮著鬼臉、頭髮飛揚跳起來的小孩」“a kid mid-goofy-face leaping through a sunlit field, hair flying” | photo 「一個孩子咬著滴下汁的冰棒，笑得合不攏嘴」“a child laughing wide around a dripping popsicle” | poster 「底片寫真集海報，跳躍扮鬼臉的孩子，手寫感標題「PLAY」」“a film-snapshot photobook poster, a kid leaping and pulling a face, title “PLAY”” | poster 「夏日寫真展海報，咬著冰棒大笑的孩子，頂端標題「夏」」“a summer photo-show poster, a child laughing with a popsicle, title “Summer”” |
TextEncodeQwenImage21 and the
UNet / CLIP / VAE loaders you use)./v1/chat/completions endpoint, serving
Qwen/Qwen-Image-2.1-PE-T2I (any backend — vLLM, llama-server, LM Studio, ...; any weight
format). Point the node's endpoint at it (default http://127.0.0.1:8207).No Python dependencies to install.
cd ComfyUI/custom_nodes
git clone https://github.com/pottokao-dotcom/ComfyUI-QwenImage-PhotoStyles
Or install from the ComfyUI Registry via ComfyUI Manager (search "Qwen-Image Photo Styles").
Inputs — prompt (multiline), style (dropdown: None + 17 styles), endpoint, seed,
temperature; optional model, timeout_s, system_prompt_path.
Outputs — prompt (STRING), negative (STRING), width (INT), height (INT),
wh_ratio (STRING). seed also picks one of the style's mood sub-presets, so re-rolling it
varies the feeling.
Black Fury · Geometry of Light · Warm Documentary · White Leap · Corner Elegance · Cold Power · Hyper Chic East · Private Diary · Color Behind Glass · Frozen Film Still · Fairytale Giants · Soft Everyday Light · Street in the Mirror · Pale Light Moments · Everyday Punchline · Vivid Garden · Playful Film.
Each style sheet in styles/*.txt specifies, per style: Look, Lens language, Light, Emotion,
People, Scene & props, Colour, plus per-style negatives and — for film-look styles — a fixed
grain clause.
This node only needs the PE-T2I rewriter; everything else is your own Qwen-Image-2.1 graph. If you want a self-quantized set to run the whole thing, here are 4-bit builds — NVFP4 for NVIDIA (ComfyUI / vLLM) and GGUF Q4 for Mac / low-VRAM (llama.cpp / ComfyUI-GGUF). ("Heretic" = an abliterated / decensored build.)
| Component | NVFP4 · NVIDIA (ComfyUI / vLLM) | GGUF Q4 · Mac (llama.cpp / ComfyUI-GGUF) |
|---|---|---|
| PE-T2I rewriter (this node calls) | Qwen-Image-2.1-PE-T2I-Heretic-NVFP4 | Qwen-Image-2.1-PE-T2I-Heretic-GGUF |
| DiT (image model) | Qwen-Image-2.1-DiT-NVFP4-ComfyUI | Qwen-Image-2.1-DiT-GGUF |
| Text encoder (Qwen3-VL) | Qwen-Image-2.1-Text-Encoder-Heretic-NVFP4 | Qwen-Image-2.1-Text-Encoder-Heretic-GGUF |
The Qwen3-VL text-encoder GGUF needs this add-on in ComfyUI (fixes the 12288 shape error): https://github.com/pottokao-dotcom/ComfyUI-GGUF-Qwen3VL-TE
See example_workflow_api.json (API format — enable Dev mode in ComfyUI settings, then
Load (API Format)). The loader nodes and model file names there are just an example; swap in
whatever Qwen-Image-2.1 models and format you use.
presets.json, docs — MIT (see LICENSE).style_sp_plus.txt is adapted from Qwen's official Qwen-Image-2.1-PE-T2I system prompt and
remains under the Qwen Research License (see NOTICE). Review it before commercial use.