想在本地用中階顯卡完成「同一角色、不同姿勢與場景」的動漫插畫?這篇帶你用 Flux.1 Kontext Dev 的 GGUF 量化版,在 ComfyUI 透過「參考圖 → 以圖生圖 → 穩定角色特徵」的方式,快速生成 1024×768 成品。核心是 Kontext 的情境式生成能力:把角色立繪餵給模型,抽取髮色、臉型、配件等視覺語彙,再依提示詞換動作、表情與場景,同時維持角色一致性。
本文提供可直接匯入的 ComfyUI 工作流思路與參數建議;全流程只靠主模型 + 文字編碼 + VAE,不需 LoRA;在約 12GB VRAM 的環境即可順跑。提醒:Kontext-dev 為非商業授權,下載與使用前請先確認授權條款;內容安全過濾需自行把關。
環境配置與模型放置
請先完成 ComfyUI 的基本安裝並能順利開啟軟體介面 (教學點我)。接著需要額外安裝幾個自訂節點,分別是:
- ComfyUI-GGUF:支援
UnetLoaderGGUF、CLIPLoaderGGUF等節點,讓 ComfyUI 能夠直接讀取.gguf模型檔。 - ComfyUI-Flux:提供
FluxGuidance、FluxKontextImageScale、ModelSamplingFlux等專屬節點,是 Kontext 工作流的核心。 - ComfyUI-Detail-Daemon:附加
DetailDaemonSamplerNode,可強化影像細節、減少夜景或暗部場景常見的塗抹與雜點。
匯入工作流後會提示遺失的節點,建議可使用 ComfyUI-Manager 進行安裝,安裝完成後務必重啟 ComfyUI,節點才會出現在清單中。完成這些步驟後,即可使用 UnetLoaderGGUF 與 FluxGuidance 等專屬節點,順利載入 GGUF 權重並執行 Flux.1 Kontext Dev 的工作流。
模型與路徑
- Diffusion(GGUF):
flux1-kontext-dev-Q5_K_M.gguf[🔗 下載點]
📁 放置 →ComfyUI/models/diffusion_models/ - Text Encoders:
t5xxl_fp8_e4m3fn.safetensors、clip_l.safetensors[🔗 下載點]
📁 放置 →ComfyUI/models/clip/ - VAE:
ae.safetensors[🔗 下載點]
📁 放置 →ComfyUI/models/vae/
工作流重點與參數建議
角色一致性的核心在於讓模型同時接收「參考圖」與「文字提示」:① 透過 Load Image → FluxKontextImageScale → VAE Encode → ReferenceLatent,把原始角色立繪送入潛空間作為參考;② 使用 FluxGuidance 配合正向提示詞,描述角色的動作、姿勢、表情與場景背景;③ 在取樣過程中保持穩定性,由 Detail Daemon 幫助增添細節與控制雜訊,確保生成影像乾淨穩定。
- 解析度:1024×768(單張批量)
- 取樣器:
res_multistep(搭配 Detail Daemon 使用),建議步數 20 起 - 排程:
sgm_uniform - ModelSamplingFlux:
max_shift=1.15、base_shift=0.5 - 正向提示:主要以「角色要做的動作、姿勢、表情、場景描述」為主,例如坐姿、笑容、背景光影等。
這種設計省略了角色細節的文字綁定,讓參考圖成為維持一致性的主要依據,確保生成的角色在不同動作與場景下仍保持相同的外觀特徵。
📝 工作流(FluxKontextGGUF.json)
將下列內容複製到文字編輯器,儲存為 FluxKontextGGUF.json 後即可在 ComfyUI 匯入。
{"id":"ad18abd3-bdee-4f80-8fae-d15d4f845b9d","revision":0,"last_node_id":62,"last_link_id":166,"nodes":[{"id":42,"type":"ReferenceLatent","pos":[402.70611572265625,5.549673080444336],"size":[197.712890625,46],"flags":{},"order":16,"mode":0,"inputs":[{"name":"conditioning","type":"CONDITIONING","link":123},{"name":"latent","shape":7,"type":"LATENT","link":138}],"outputs":[{"name":"CONDITIONING","type":"CONDITIONING","links":[157]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.42","Node name for S&R":"ReferenceLatent","widget_ue_connectable":{}},"widgets_values":[]},{"id":41,"type":"LoadImage","pos":[-260.508544921875,43.22496795654297],"size":[274.080078125,314],"flags":{},"order":0,"mode":0,"inputs":[],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[158]},{"name":"MASK","type":"MASK","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"LoadImage","widget_ue_connectable":{}},"widgets_values":["萌芽娘(4比3).jpeg","image"],"color":"#332922","bgcolor":"#593930"},{"id":40,"type":"FluxKontextImageScale","pos":[40.24856948852539,5.769573211669922],"size":[187.75448608398438,26],"flags":{},"order":8,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":158}],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[166]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.42","Node name for S&R":"FluxKontextImageScale","widget_ue_connectable":{}},"widgets_values":[]},{"id":39,"type":"VAEEncode","pos":[248.9091339111328,1.1059212684631348],"size":[140,46],"flags":{},"order":13,"mode":0,"inputs":[{"name":"pixels","type":"IMAGE","link":166},{"name":"vae","type":"VAE","link":120}],"outputs":[{"name":"LATENT","type":"LATENT","links":[138]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"VAEEncode","widget_ue_connectable":{}},"widgets_values":[]},{"id":10,"type":"VAELoader","pos":[39.957279205322266,140.95301818847656],"size":[311.81634521484375,60.429901123046875],"flags":{},"order":1,"mode":0,"inputs":[],"outputs":[{"name":"VAE","type":"VAE","slot_index":0,"links":[12,120]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"VAELoader","widget_ue_connectable":{}},"widgets_values":["ae.safetensors"],"color":"#2a363b","bgcolor":"#3f5159"},{"id":11,"type":"DualCLIPLoader","pos":[43.029659271240234,243.72886657714844],"size":[308.3652648925781,130.82933044433594],"flags":{},"order":2,"mode":0,"inputs":[],"outputs":[{"name":"CLIP","type":"CLIP","slot_index":0,"links":[10]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"DualCLIPLoader","widget_ue_connectable":{}},"widgets_values":["t5xxl_fp8_e4m3fn.safetensors","clip_l.safetensors","flux","default"],"color":"#2a363b","bgcolor":"#3f5159"},{"id":59,"type":"UnetLoaderGGUF","pos":[44.689239501953125,421.2041320800781],"size":[305.9125061035156,72.30432891845703],"flags":{},"order":3,"mode":0,"inputs":[],"outputs":[{"name":"MODEL","type":"MODEL","links":[159]}],"properties":{"cnr_id":"comfyui-gguf","ver":"b3ec875a68d94b758914fd48d30571d953bb7a54","Node name for S&R":"UnetLoaderGGUF","widget_ue_connectable":{}},"widgets_values":["flux1-kontext-dev-Q5_K_M.gguf"],"color":"#2a363b","bgcolor":"#3f5159"},{"id":34,"type":"PrimitiveNode","pos":[-200.1617889404297,401.81292724609375],"size":[210,82],"flags":{"collapsed":false},"order":4,"mode":0,"inputs":[],"outputs":[{"name":"INT","type":"INT","widget":{"name":"width"},"slot_index":0,"links":[112,115]}],"title":"width","properties":{"Run widget replace on values":false,"widget_ue_connectable":{}},"widgets_values":[1024,"fixed"],"color":"#323","bgcolor":"#535"},{"id":35,"type":"PrimitiveNode","pos":[-199.23150634765625,526.8526611328125],"size":[210,82],"flags":{"collapsed":false},"order":5,"mode":0,"inputs":[],"outputs":[{"name":"INT","type":"INT","widget":{"name":"height"},"slot_index":0,"links":[113,114]}],"title":"height","properties":{"Run widget replace on values":false,"widget_ue_connectable":{}},"widgets_values":[768,"fixed"],"color":"#323","bgcolor":"#535"},{"id":6,"type":"CLIPTextEncode","pos":[394.9665222167969,204.35877990722656],"size":[328.9437255859375,164.31304931640625],"flags":{},"order":9,"mode":0,"inputs":[{"name":"clip","type":"CLIP","link":10}],"outputs":[{"name":"CONDITIONING","type":"CONDITIONING","slot_index":0,"links":[41]}],"title":"CLIP Text Encode (Positive Prompt)","properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"CLIPTextEncode","widget_ue_connectable":{}},"widgets_values":["Wearing elegant lingerie, lying on a velvet sofa with one leg slightly raised. She gazes with a sultry expression under the glow of a dim chandelier, creating a seductive yet refined mood. Anime style, highly detailed, vibrant colors."],"color":"#232","bgcolor":"#353"},{"id":26,"type":"FluxGuidance","pos":[395.9127502441406,101.43500518798828],"size":[324.93951416015625,58],"flags":{},"order":14,"mode":0,"inputs":[{"name":"conditioning","type":"CONDITIONING","link":41}],"outputs":[{"name":"CONDITIONING","type":"CONDITIONING","slot_index":0,"links":[123]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"FluxGuidance","widget_ue_connectable":{}},"widgets_values":[2.5],"color":"#223","bgcolor":"#335"},{"id":16,"type":"KSamplerSelect","pos":[398.1559143066406,414.0418701171875],"size":[323.22491455078125,62.112449645996094],"flags":{},"order":6,"mode":0,"inputs":[],"outputs":[{"name":"SAMPLER","type":"SAMPLER","links":[160]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"KSamplerSelect","widget_ue_connectable":{}},"widgets_values":["res_multistep"]},{"id":17,"type":"BasicScheduler","pos":[401.2137451171875,518.5570068359375],"size":[318.42706298828125,106],"flags":{},"order":15,"mode":0,"inputs":[{"name":"model","type":"MODEL","link":55}],"outputs":[{"name":"SIGMAS","type":"SIGMAS","links":[20]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"BasicScheduler","widget_ue_connectable":{}},"widgets_values":["sgm_uniform",20,1]},{"id":30,"type":"ModelSamplingFlux","pos":[401.4834899902344,668.7677612304688],"size":[319.1124572753906,130],"flags":{},"order":11,"mode":0,"inputs":[{"name":"model","type":"MODEL","link":159},{"name":"width","type":"INT","widget":{"name":"width"},"link":115},{"name":"height","type":"INT","widget":{"name":"height"},"link":114}],"outputs":[{"name":"MODEL","type":"MODEL","slot_index":0,"links":[54,55]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"ModelSamplingFlux","widget_ue_connectable":{"width":true,"height":true}},"widgets_values":[1.15,0.5,1024,768]},{"id":22,"type":"BasicGuider","pos":[615.810302734375,5.704440593719482],"size":[222.3482666015625,46],"flags":{},"order":17,"mode":0,"inputs":[{"name":"model","type":"MODEL","link":54},{"name":"conditioning","type":"CONDITIONING","link":157}],"outputs":[{"name":"GUIDER","type":"GUIDER","slot_index":0,"links":[30]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"BasicGuider","widget_ue_connectable":{}},"widgets_values":[]},{"id":9,"type":"SaveImage","pos":[1067.2725830078125,111.17443084716797],"size":[534.5103149414062,666.5654907226562],"flags":{},"order":20,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":165}],"outputs":[],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"SaveImage","widget_ue_connectable":{}},"widgets_values":["FluxKontext/ComfyUI"],"color":"#233","bgcolor":"#355"},{"id":27,"type":"EmptySD3LatentImage","pos":[45.176212310791016,563.6580200195312],"size":[315,106],"flags":{},"order":10,"mode":0,"inputs":[{"name":"width","type":"INT","widget":{"name":"width"},"link":112},{"name":"height","type":"INT","widget":{"name":"height"},"link":113}],"outputs":[{"name":"LATENT","type":"LATENT","slot_index":0,"links":[116]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"EmptySD3LatentImage","widget_ue_connectable":{"width":true,"height":true}},"widgets_values":[1024,768,1]},{"id":25,"type":"RandomNoise","pos":[744.817626953125,105.23249816894531],"size":[296.94219970703125,82],"flags":{},"order":7,"mode":0,"inputs":[],"outputs":[{"name":"NOISE","type":"NOISE","links":[37]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"RandomNoise","widget_ue_connectable":{}},"widgets_values":[1057506492236458,"randomize"],"color":"#432","bgcolor":"#653"},{"id":13,"type":"SamplerCustomAdvanced","pos":[744.6983032226562,232.98263549804688],"size":[294.8928527832031,106],"flags":{},"order":18,"mode":0,"inputs":[{"name":"noise","type":"NOISE","link":37},{"name":"guider","type":"GUIDER","link":30},{"name":"sampler","type":"SAMPLER","link":161},{"name":"sigmas","type":"SIGMAS","link":20},{"name":"latent_image","type":"LATENT","link":116}],"outputs":[{"name":"output","type":"LATENT","slot_index":0,"links":[24]},{"name":"denoised_output","type":"LATENT","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"SamplerCustomAdvanced","widget_ue_connectable":{}},"widgets_values":[]},{"id":8,"type":"VAEDecode","pos":[744.5648803710938,379.58099365234375],"size":[296.205810546875,47.1153564453125],"flags":{},"order":19,"mode":0,"inputs":[{"name":"samples","type":"LATENT","link":24},{"name":"vae","type":"VAE","link":12}],"outputs":[{"name":"IMAGE","type":"IMAGE","slot_index":0,"links":[165]}],"properties":{"cnr_id":"comfy-core","ver":"0.3.41","Node name for S&R":"VAEDecode","widget_ue_connectable":{}},"widgets_values":[]},{"id":60,"type":"DetailDaemonSamplerNode","pos":[744.7194213867188,471.7207946777344],"size":[294.1264343261719,274],"flags":{},"order":12,"mode":0,"inputs":[{"name":"sampler","type":"SAMPLER","link":160}],"outputs":[{"name":"SAMPLER","type":"SAMPLER","links":[161]}],"properties":{"cnr_id":"comfyui-detail-daemon","ver":"f391accbda2d309cdcbec65cb9fcc80a41197b20","Node name for S&R":"DetailDaemonSamplerNode","widget_ue_connectable":{}},"widgets_values":[0.06000000000000001,0.30000000000000004,0.7000000000000002,0.5,1,0,0,0,true,0]}],"links":[[10,11,0,6,0,"CLIP"],[12,10,0,8,1,"VAE"],[20,17,0,13,3,"SIGMAS"],[24,13,0,8,0,"LATENT"],[30,22,0,13,1,"GUIDER"],[37,25,0,13,0,"NOISE"],[41,6,0,26,0,"CONDITIONING"],[54,30,0,22,0,"MODEL"],[55,30,0,17,0,"MODEL"],[112,34,0,27,0,"INT"],[113,35,0,27,1,"INT"],[114,35,0,30,2,"INT"],[115,34,0,30,1,"INT"],[116,27,0,13,4,"LATENT"],[120,10,0,39,1,"VAE"],[123,26,0,42,0,"CONDITIONING"],[138,39,0,42,1,"LATENT"],[157,42,0,22,1,"CONDITIONING"],[158,41,0,40,0,"IMAGE"],[159,59,0,30,0,"MODEL"],[160,16,0,60,0,"SAMPLER"],[161,60,0,13,2,"SAMPLER"],[165,8,0,9,0,"IMAGE"],[166,40,0,39,0,"IMAGE"]],"groups":[{"id":1,"title":"Load model","bounding":[28.587430953979492,64.5455551147461,337.3845520019531,444.08306884765625],"color":"#3f789e","font_size":24,"flags":{}}],"config":{},"extra":{"ds":{"scale":0.724729500000007,"offset":[332.3715122539601,101.11918845824033]},"frontendVersion":"1.26.15","groupNodes":{},"VHS_latentpreview":false,"VHS_latentpreviewrate":0,"VHS_MetadataImage":true,"VHS_KeepIntermediate":true,"ue_links":[],"links_added_by_ue":[]},"version":0.4}

▲ ComfyUI 的工作流截圖:以 ReferenceLatent 導入角色潛空間,中段透過 FluxGuidance 與 Detail Daemon 提升穩定度,最後再解碼輸出。整體流程乾淨直覺,方便替換提示詞並維持角色一致性。
提示詞示範與生成成果(以圖生圖)
以下展示的是角色「萌芽娘」在不同服裝與場景下的生成成果,全部皆以同一張立繪作為參考圖,並透過文字提示控制角色的動作、表情與場景氛圍。這九張作品均為一次直出、未經後製修圖,能清楚看出 Flux.1 Kontext Dev 在維持角色一致性的同時,仍能靈活切換不同情境與風格。每張圖約生成 3 分鐘,還可以接受。本次使用的提示詞如下:
1. Wearing a frilly maid outfit, kneeling gracefully on the floor while holding a silver tray with tea. She has a shy yet charming smile, surrounded by soft candlelight in a cozy room. Anime style, highly detailed, vibrant colors.
2. Wearing a soft oversized sweater, lying on a bed with her chin resting on her hands. Her playful eyes and gentle smile create a warm, intimate mood in a softly lit bedroom. Anime style, highly detailed, vibrant colors.
3. In a magical girl costume, standing confidently with one hand on her hip and the other pointing a glowing wand toward the sky. The night is filled with sparkling stars and glowing magic effects. Anime style, highly detailed, vibrant colors.
4. Wearing a bikini and sheer sarong, lounging on a beach chair with sunglasses in hand. She stretches lazily, with waves and palm trees in the background under golden sunlight. Anime style, highly detailed, vibrant colors.
5. Dressed in a school uniform with a short pleated skirt, sitting on a swing with legs slightly crossed. She smiles softly as cherry blossom petals fall around her in the spring breeze. Anime style, highly detailed, vibrant colors.
6. In a stylish yukata, sitting on wooden steps by a festival stall, holding cotton candy with both hands. Fireworks glow in the night sky, reflecting in her joyful eyes. Anime style, highly detailed, vibrant colors.
7. Wearing a sexy bunny suit with fishnet stockings, sitting cross-legged on a barstool. She leans forward with a playful wink, neon lights glowing behind her. Anime style, highly detailed, vibrant colors.
8. In a flowing white sundress, running barefoot across a meadow. She holds her hat with one hand while laughing freely, surrounded by butterflies in warm sunlight. Anime style, highly detailed, vibrant colors.
9. Wearing elegant lingerie, lying on a velvet sofa with one leg slightly raised. She gazes with a sultry expression under the glow of a dim chandelier, creating a seductive yet refined mood. Anime style, highly detailed, vibrant colors.

▲ 女僕服端盤跪坐(燭光室內)。角色臉型與髮飾特徵清晰一致,氛圍溫暖柔和,細節表現穩定。

▲ 寬鬆毛衣趴床。角色五官比例與表情保持連貫,場景氛圍呈現輕鬆舒適的居家感,可惜服裝沒有替換。

▲ 魔法少女夜空指星。服裝與特效元素完整呈現,角色神情自信,與前面作品仍保持同一角色的特徵。服裝沒有更換到,也許要再調整提示詞。

▲ 沙灘日落與躺椅。服裝有更換,畫面展現悠閒氛圍,角色外觀穩定,依舊能清楚辨識為同一角色。

▲ 校服坐鞦韆。原本的服裝就類似校服,因此沒更換。櫻花與春風吹拂的場景氛圍自然,角色表情柔和,臉型與特徵維持一致。

▲ 浴衣祭典與煙火。這張服裝也沒更換到。夜晚場景光影鮮明,角色神情充滿喜悅,與其他畫面中的角色保持同樣的面貌。

▲ 兔女郎吧檯。目前看來用「Wearing」換裝成功率較高。角色在霓虹燈背景下展現俏皮表情,特徵與細節穩定,角色辨識度高。

▲ 白色洋裝奔跑草地。這次全身都有替換到服裝啦!畫面充滿動態與自由感,角色臉部與髮飾保持穩定。手指有些細節上的不完美。

▲ 絲緞沙發與吊燈。場景氛圍高雅,角色表情帶有成熟氣質,與其他生成圖保持一致的角色形象。太性感的服裝就出不來了。
結語
Flux.1 Kontext Dev 讓「以圖生圖+角色一致」的流程變得前所未有的簡單且可控。只要先透過參考圖抽取角色特徵,再搭配提示詞去描述表情、姿勢與場景,就能在不同的情境下維持同一角色的外觀與辨識度。加上取樣器與 Detail Daemon 的輔助,畫面細節能夠保持乾淨穩定,不容易出現常見的塗抹或崩壞問題,對於想要建立完整角色視覺資料庫的創作者而言,這是一條非常高效的工作流。
更重要的是,這樣的設計大幅降低了學習門檻。你不需要額外訓練 LoRA 或管理複雜的多模型切換,只要在 ComfyUI 安裝必要的節點,按照本文的模型放置路徑與工作流配置,就能直接開始生成。即使是家用等級的顯卡,在約 12GB VRAM 的條件下,也能順暢地輸出 1024×768 的插畫成果,兼顧速度與品質。這讓 Flux.1 Kontext Dev 成為許多插畫師、同人創作者、以及想要快速做專案視覺草圖的最佳工具之一。
《上一篇》Qwen Chat:線上免費體驗 Qwen-Image-Edit 為任一角色換裝|AI 換裝工具推薦
《下一篇》【Gigastone 立達】512GB microSDXC UHS-I U3 A2 V30 高速記憶卡 









留言區 / Comments
萌芽論壇