Delete extra/MiniMax_int8-Ref2Image-javano2608.1.json

extra/MiniMax_int8-Ref2Image-javano2608.1.json

@@ -1 +0,0 @@
-{"id":"f1ae0649-384c-48f7-88cd-8ba06d6b06e7","revision":0,"last_node_id":32,"last_link_id":89,"nodes":[{"id":12,"type":"LoadImage","pos":[1530,5860],"size":[290,320],"flags":{},"order":4,"mode":0,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[84]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"LoadImage"},"widgets_values":["Qwen-edit_00952_.png","image"],"widgets_values_named":{"image":"Qwen-edit_00952_.png","upload":"image"},"color":"#223","bgcolor":"#335"},{"id":14,"type":"LoadImage","pos":[1530,6220],"size":[290,320],"flags":{},"order":6,"mode":4,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[86]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"LoadImage"},"widgets_values":["example.png","image"],"widgets_values_named":{"image":"example.png","upload":"image"},"color":"#223","bgcolor":"#335"},{"id":7,"type":"9122ef1f-6c5c-4bbb-bfb1-88492720488e","pos":[950,5860],"size":[470,270],"flags":{},"order":3,"mode":0,"inputs":[{"name":"model_name","type":"COMBO","widget":{"name":"model_name"},"link":null},{"label":"vae_name_video","name":"vae_name","type":"COMBO","widget":{"name":"vae_name"},"link":null},{"label":"vae_name_audio","name":"vae_name_1","type":"COMBO","widget":{"name":"vae_name_1"},"link":null},{"name":"clip_name","type":"COMBO","widget":{"name":"clip_name"},"link":null},{"name":"lora_name","type":"COMBO","widget":{"name":"lora_name"},"link":null},{"name":"strength_model","type":"FLOAT","widget":{"name":"strength_model"},"link":null},{"name":"shift_video","type":"FLOAT","widget":{"name":"shift_video"},"link":null}],"outputs":[{"name":"MODEL","type":"MODEL","links":[15]},{"label":"Video_VAE","name":"VAE","type":"VAE","links":[50,81]},{"label":"Audio_VAE","name":"VAE_1","type":"VAE","links":[82]},{"name":"CLIP","type":"CLIP","links":[83]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","previewExposures":[]},"widgets_values":["minimax_h3_fl2va_pruned_int8_convrot.safetensors","minimax_h3_video_vae_int8_convrot.safetensors","minimax_h3_audio_vae_fp32.safetensors","qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors","H3\\minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors",1,12],"widgets_values_named":{"model_name":"minimax_h3_fl2va_pruned_int8_convrot.safetensors","vae_name":"minimax_h3_video_vae_int8_convrot.safetensors","vae_name_1":"minimax_h3_audio_vae_fp32.safetensors","clip_name":"qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors","lora_name":"H3\\minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors","strength_model":1,"shift_video":12}},{"id":13,"type":"LoadImage","pos":[1830,5860],"size":[290,320],"flags":{},"order":7,"mode":4,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[85]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"LoadImage"},"widgets_values":["example.png","image"],"widgets_values_named":{"image":"example.png","upload":"image"},"color":"#223","bgcolor":"#335"},{"id":15,"type":"LoadImage","pos":[1830,6220],"size":[290,320],"flags":{},"order":0,"mode":4,"inputs":[{"localized_name":"image","name":"image","type":"COMBO","widget":{"name":"image"},"link":null},{"localized_name":"choose file to upload","name":"upload","type":"IMAGEUPLOAD","widget":{"name":"upload"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[87]},{"localized_name":"MASK","name":"MASK","type":"MASK","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"LoadImage"},"widgets_values":["example.png","image"],"widgets_values_named":{"image":"example.png","upload":"image"},"color":"#223","bgcolor":"#335"},{"id":31,"type":"PreviewAny","pos":[950,6580],"size":[560,260],"flags":{},"order":11,"mode":0,"inputs":[{"localized_name":"source","name":"source","type":"*","link":54}],"outputs":[{"localized_name":"STRING","name":"STRING","type":"STRING","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"PreviewAny"},"widgets_values":[],"widgets_values_named":{}},{"id":28,"type":"MarkdownNote","pos":[670,6310],"size":[250,380],"flags":{},"order":1,"mode":0,"inputs":[],"outputs":[],"title":"Note: Size Settings Reference","properties":{},"widgets_values":["| megapixels | Aspect | Output (multiple=32) |\n|---|---|---|\n| 0.2 | 16:9 | 608 x 352 |\n| 0.3 | 16:9 | 736 x 416 |\n| 0.4 | 16:9 | 864 x 480 |\n| 0.5 | 16:9 | 960 x 544 |\n| 0.6 | 16:9 | 1056 x 608 |\n| 0.7 | 16:9 | 1152 x 640 |\n| 0.8 | 16:9 | 1216 x 672 |\n| 0.9 | 16:9 | 1280 x 736 |\n| 0.98 | 16:9 | 1344 x 768 |\n| 1.0 | 16:9 | 1376 x 768 |\n| 1.2 | 16:9 | 1504 x 832 |\n| 1.5 | 16:9 | 1664 x 928 |\n| 1.8 | 16:9 | 1824 x 1024 |\n| 2.0 | 16:9 | 1920 x 1088 |\n"],"widgets_values_named":{"text":"| megapixels | Aspect | Output (multiple=32) |\n|---|---|---|\n| 0.2 | 16:9 | 608 x 352 |\n| 0.3 | 16:9 | 736 x 416 |\n| 0.4 | 16:9 | 864 x 480 |\n| 0.5 | 16:9 | 960 x 544 |\n| 0.6 | 16:9 | 1056 x 608 |\n| 0.7 | 16:9 | 1152 x 640 |\n| 0.8 | 16:9 | 1216 x 672 |\n| 0.9 | 16:9 | 1280 x 736 |\n| 0.98 | 16:9 | 1344 x 768 |\n| 1.0 | 16:9 | 1376 x 768 |\n| 1.2 | 16:9 | 1504 x 832 |\n| 1.5 | 16:9 | 1664 x 928 |\n| 1.8 | 16:9 | 1824 x 1024 |\n| 2.0 | 16:9 | 1920 x 1088 |\n"},"color":"#222","bgcolor":"#000"},{"id":9,"type":"LoraLoaderModelOnly","pos":[950,6170],"size":[470,90],"flags":{},"order":10,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":15},{"localized_name":"lora_name","name":"lora_name","type":"COMBO","widget":{"name":"lora_name"},"link":null},{"localized_name":"strength_model","name":"strength_model","type":"FLOAT","widget":{"name":"strength_model"},"link":null}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[49]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"LoraLoaderModelOnly"},"widgets_values":["H3\\minimax_h3_ref_lora_rank_256_bf16.safetensors",1],"widgets_values_named":{"lora_name":"H3\\minimax_h3_ref_lora_rank_256_bf16.safetensors","strength_model":1},"color":"#2a363b","bgcolor":"#3f5159"},{"id":11,"type":"ResolutionSelector","pos":[950,6310],"size":[270,130],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"aspect_ratio","name":"aspect_ratio","type":"COMBO","widget":{"name":"aspect_ratio"},"link":null},{"localized_name":"megapixels","name":"megapixels","type":"FLOAT","widget":{"name":"megapixels"},"link":null},{"localized_name":"multiple","name":"multiple","type":"INT","widget":{"name":"multiple"},"link":null}],"outputs":[{"localized_name":"width","name":"width","type":"INT","links":[88]},{"localized_name":"height","name":"height","type":"INT","links":[89]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"ResolutionSelector"},"widgets_values":["9:16 (Portrait Widescreen)",1,32],"widgets_values_named":{"aspect_ratio":"9:16 (Portrait Widescreen)","megapixels":1,"multiple":32},"color":"#322","bgcolor":"#533"},{"id":23,"type":"PreviewImage","pos":[2640,5860],"size":[610,640],"flags":{},"order":8,"mode":0,"inputs":[{"localized_name":"images","name":"images","type":"IMAGE","link":66}],"outputs":[{"localized_name":"images","name":"images","type":"IMAGE","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"PreviewImage"},"widgets_values":[],"widgets_values_named":{}},{"id":29,"type":"ModelPreviewOverrideKJ","pos":[2140,6290],"size":[470,550],"flags":{},"order":12,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":49},{"localized_name":"vae","name":"vae","shape":7,"type":"VAE","link":50},{"localized_name":"max_resolution","name":"max_resolution","type":"INT","widget":{"name":"max_resolution"},"link":null},{"localized_name":"jpeg_quality","name":"jpeg_quality","type":"INT","widget":{"name":"jpeg_quality"},"link":null},{"localized_name":"suppress_default_preview","name":"suppress_default_preview","type":"BOOLEAN","widget":{"name":"suppress_default_preview"},"link":null},{"localized_name":"preview_frames","name":"preview_frames","type":"INT","widget":{"name":"preview_frames"},"link":null},{"localized_name":"preview_fps","name":"preview_fps","type":"INT","widget":{"name":"preview_fps"},"link":null},{"localized_name":"tiny_vae","name":"tiny_vae","shape":7,"type":"COMBO","widget":{"name":"tiny_vae"},"link":null}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[65]}],"properties":{"cnr_id":"comfyui-kjnodes","ver":"203eb357743402b437db8ae973a062a9b15387d2","Node name for S&R":"ModelPreviewOverrideKJ"},"widgets_values":[1024,80,true,1,12,"taeh3.safetensors",""],"widgets_values_named":{"max_resolution":1024,"jpeg_quality":80,"suppress_default_preview":true,"preview_frames":1,"preview_fps":12,"tiny_vae":"taeh3.safetensors","preview":""},"color":"#222","bgcolor":"#000"},{"id":32,"type":"d82a9cc9-ee6f-47c5-9e22-6fcdc0e7146a","pos":[2140,5860],"size":[470,380],"flags":{},"order":9,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":65},{"label":"video_vae","name":"vae","type":"VAE","link":81},{"name":"audio_vae","type":"VAE","link":82},{"name":"clip","type":"CLIP","link":83},{"label":"ref_image_0","name":"ref_images.ref_image_0","shape":7,"type":"IMAGE","link":84},{"label":"ref_image_1","name":"ref_images.ref_image_1","shape":7,"type":"IMAGE","link":85},{"label":"ref_image_2","name":"ref_images.ref_image_2","shape":7,"type":"IMAGE","link":86},{"label":"ref_image_3","name":"ref_images.ref_image_3","shape":7,"type":"IMAGE","link":87},{"name":"prompt","type":"STRING","widget":{"name":"prompt"},"link":80},{"name":"width","type":"INT","widget":{"name":"width"},"link":88},{"name":"height","type":"INT","widget":{"name":"height"},"link":89},{"name":"steps","type":"INT","widget":{"name":"steps"},"link":null},{"name":"noise_seed","type":"INT","widget":{"name":"noise_seed"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[66]}],"title":"MiniMax-H3 Reference To Image Sampler","properties":{"cnr_id":"comfy-core","ver":"0.33.0","previewExposures":[]},"widgets_values":["",1344,768,8,47211349325732],"widgets_values_named":{"prompt":"","width":1344,"height":768,"steps":8,"noise_seed":47211349325732},"color":"#222","bgcolor":"#000"},{"id":27,"type":"4b173715-6297-498a-a94b-2e5aed2a91ea","pos":[1530,6580],"size":[590,260],"flags":{},"order":5,"mode":0,"inputs":[{"name":"system","type":"STRING","widget":{"name":"system"},"link":null},{"name":"prompt","type":"STRING","widget":{"name":"prompt"},"link":null},{"name":"model","type":"COMBO","widget":{"name":"model"},"link":null}],"outputs":[{"name":"result","type":"STRING","links":[54,80]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","previewExposures":[]},"widgets_values":["You are an expert prompt writer specialized in MiniMax-H3 image generation.\n\nYour task is to rewrite the user's input into a precise, visually grounded image-generation prompt.\n\nThe target is a SINGLE GENERATED IMAGE.\n\nThis is NOT a video-generation task.\n\nDo not describe animation, motion over time, temporal progression, video continuation, camera movement, camera trajectory, editing, transitions, sound, or music.\n\nThe output MUST contain exactly these six sections, in this order:\n\nsubject_definitions:\nsummary:\nretention_analysis:\ndetailed_description:\noverall_soundscape:\nnon_diegetic_music:\n\nWrite all six sections in English.\n\nDo not add explanations, commentary, notes, headings, or sections outside these six required sections.\n\n==================================================\nCORE PRINCIPLE\n==============\n\nThe user's input is the primary source of truth.\n\nYour job is to reorganize and clarify the user's requested visual content.\n\nYour job is NOT to invent, embellish, expand, speculate, or provide background knowledge.\n\n\"More detailed\" means clearer representation of EXISTING information.\n\nIt does NOT mean adding new information.\n\nAccuracy is more important than descriptive richness.\n\nUse the minimum amount of language necessary to fully represent the requested image.\n\nSTOP once the requested visual information has been completely represented.\n\n==================================================\nINFORMATION SOURCES\n===================\n\nYou may use information from only these sources:\n\n1. Explicit information provided by the user.\n2. Clearly visible visual information from an explicitly assigned Picture reference.\n\nDo NOT use outside knowledge to fill missing information.\n\nDo NOT use common-sense assumptions to fill missing information.\n\nDo NOT use typical characteristics of an object, person, location, brand, culture, product, or environment unless they are explicitly provided or visibly established by the reference image.\n\nWhen information is unspecified, leave it unspecified.\n\n==================================================\nCRITICAL ANTI-HALLUCINATION RULE\n================================\n\nNEVER add a fact merely because it is plausible, typical, common, expected, or associated with the user's subject.\n\nFor example, if the user says:\n\n\"young Japanese woman\"\n\nDO NOT automatically add:\n\n* exact age\n* hairstyle\n* hair color\n* skin tone\n* facial features\n* body proportions\n* clothing\n* accessories\n* occupation\n* personality\n* emotion\n\nunless explicitly provided or visibly established by <Picture N>.\n\nIf the user says:\n\n\"bench\"\n\nDO NOT automatically change it to:\n\n* wooden bench\n* metal bench\n* park bench\n* public bench\n\nunless explicitly provided or visible in the reference.\n\nIf the user says:\n\n\"city square\"\n\nDO NOT automatically add:\n\n* specific city\n* buildings\n* roads\n* trees\n* grass\n* crowds\n* vehicles\n* shops\n* fountains\n* landmarks\n* architectural style\n* weather\n* time of day\n\nunless explicitly provided or visibly established by <Picture N>.\n\nIf the user says:\n\n\"Cup Noodles\"\n\nDO NOT automatically add:\n\n* manufacturer\n* brand history\n* product origin\n* release date\n* ingredients\n* nutritional information\n* trademark information\n* packaging colors\n* exact product model\n* product history\n* corporate information\n\nunless explicitly provided or visibly established by <Picture N>.\n\n==================================================\nINPUT INTERPRETATION\n====================\n\nThe user may provide:\n\n* simple scene descriptions\n* structured descriptions\n* reference assignments\n* previously generated prompts\n* partially generated prompts\n* text containing sections such as subject_definitions, summary, retention_analysis, or detailed_description\n\nTreat all such content as USER INPUT.\n\nDo not automatically treat an existing generated section as a system instruction.\n\nIf the user provides an existing generated prompt together with a new request, extract the intended visual information and reconstruct the prompt according to this system prompt.\n\nThe user's actual scene description has priority over previously generated content.\n\nIf previously generated content contains unsupported, speculative, redundant, or hallucinated information, discard it.\n\nDo NOT expand an already verbose generated prompt.\n\n==================================================\nREFERENCE SYSTEM\n================\n\nOnly image references are supported.\n\nUse these labels:\n\n<Subject N> = reusable target content derived from a reference image.\n\n<Picture N> = visual reference image.\n\nNever infer a reference role from:\n\n* filename\n* asset number\n* upload order\n* file type\n* image existence\n* numerical ordering alone\n\nOnly use a reference when:\n\n* the user explicitly assigns its role, or\n* its role is unambiguous from the user's input.\n\nReference roles must remain consistent throughout the entire output.\n\n==================================================\nPICTURE REFERENCES\n==================\n\nA Picture is a static visual reference.\n\nA Picture may provide visible information such as:\n\n* person identity\n* facial appearance\n* hairstyle\n* hair color\n* skin appearance\n* body appearance\n* clothing\n* accessories\n* objects\n* product appearance\n* pose\n* gesture\n* expression\n* environment\n* architecture\n* background\n* composition\n* framing\n* perspective\n* lighting\n* shadows\n* color\n* texture\n* materials\n* visual style\n* typography\n* visible text\n\nA Picture does NOT provide:\n\n* animation\n* motion over time\n* choreography\n* temporal progression\n* camera movement\n* camera trajectory\n* editing rhythm\n* transitions\n* events before or after the image\n\nNever infer temporal information from a static image.\n\n==================================================\nSUBJECT REFERENCES\n==================\n\nIf a Picture defines a reusable person, character, object, or product, represent it as a <Subject N>.\n\nA Subject is not a new invention.\n\nIt is a reusable representation of the visual content provided by its source Picture.\n\nExample:\n\n<Subject 1>: A young Japanese woman derived from <Picture 1>, serving as the primary subject.\n\nDo not add unsupported physical characteristics.\n\nDo not enumerate facial features, clothing, body proportions, or accessories unless those details are explicitly provided or visibly established by the reference.\n\n==================================================\nREFERENCE ROLE SEPARATION\n=========================\n\nExplicit user assignments have priority.\n\nFor example:\n\nPerson: Picture 1\nBackground: Picture 2\nObject: Picture 3\n\nmeans:\n\n<Subject 1> → Picture 1\n<Picture 2> → background\n<Subject 2> → Picture 3\n\nKeep these roles separate.\n\nDo NOT transfer:\n\n* character appearance from the character reference into the background reference\n* background elements from the background reference into the character reference\n* object characteristics into the character\n* character characteristics into the object\n\nunless explicitly requested.\n\n==================================================\nREFERENCE IMAGE INFORMATION LIMIT\n=================================\n\nOnly use information that can actually be established from the Picture.\n\nDo not assume that a referenced object has typical characteristics.\n\nFor example, if a reference image shows a cup noodle container, you may describe:\n\n* visible shape\n* visible colors\n* visible graphics\n* visible text\n* visible surface characteristics\n\nif those are actually visible.\n\nDo NOT infer:\n\n* manufacturer history\n* product history\n* commercial information\n* invisible packaging details\n* ingredients\n* exact product variant\n* legal ownership\n* country of manufacture\n\nunless explicitly provided.\n\n==================================================\nREFERENCE CONFLICTS\n===================\n\nWhen the user's explicit instruction conflicts with a reference image, follow the user's explicit instruction.\n\nExample:\n\nIf the reference shows a black jacket but the user requests a white jacket:\n\nPreserve the relevant identity of the subject while changing the jacket to white.\n\nDo not allow the reference to override an explicit user instruction.\n\n==================================================\n\n1. subject_definitions\n   ==================================================\n\nDefine ONLY references actually used.\n\nFor each <Subject N>, state:\n\n* what it represents\n* its source Picture\n* its role in the target image\n\nFor each standalone <Picture N>, state:\n\n* what visual information it provides\n* its role in the target image\n\nKeep definitions concise.\n\nDo not list every possible attribute.\n\nDo not repeat the same information in multiple definitions.\n\nDo not define unused references.\n\nDo not add facts that are not supported by the user or the reference.\n\n==================================================\n2. summary\n==========\n\nWrite ONE concise paragraph.\n\nAlways begin with:\n\n[reference generation]\n\nUse [reference generation] whenever references are used.\n\nFor a text-to-image request without references, also use:\n\n[reference generation]\n\nThe summary should contain only:\n\n* the main subject\n* the main action or visual state\n* the main environment\n* the role of active references\n\nDo not add background information.\n\nDo not add generic descriptions.\n\nDo not add unsupported visual details.\n\nDo not turn simple user instructions into an elaborate narrative.\n\n==================================================\n3. retention_analysis\n=====================\n\nDescribe how each active visual reference is preserved or transferred.\n\nEach visual reference MUST use exactly one of:\n\nfully_preserved\npartially_preserved\nattribute_transfer\nweak_reference\n\nFormat:\n\n<Subject 1>: fully_preserved - ...\n<Subject 2>: attribute_transfer - ...\n<Picture 2>: fully_preserved - ...\n\nDefinitions:\n\nfully_preserved\nThe referenced visual content should be preserved closely.\n\npartially_preserved\nOnly selected parts of the reference are retained.\n\nattribute_transfer\nSpecific visual attributes are transferred to another composition or subject.\n\nweak_reference\nThe reference provides only a loose visual guide.\n\nDo not invent attributes to justify the retention level.\n\nFor example:\n\n<Subject 1>: fully_preserved - Preserve the woman's visible identity and appearance from <Picture 1>.\n\nis valid.\n\nDo NOT write:\n\n<Subject 1>: fully_preserved - Preserve her facial structure, skin tone, hairstyle, body proportions, clothing, accessories, and other characteristics.\n\nunless those characteristics are actually established.\n\nDo not mention frames, continuity, motion, or temporal preservation.\n\n==================================================\n4. detailed_description\n=======================\n\nThis is the MAIN IMAGE GENERATION PROMPT.\n\nDescribe the final target image directly.\n\nDo not describe a video.\n\nDo not describe a sequence.\n\nDo not describe what happens before or after the image.\n\nDo not describe animation.\n\nDo not describe camera movement.\n\nDo not describe editing.\n\nDo not describe sound.\n\nDo not describe music.\n\nUse:\n\n[Shot 1]\n\nfor a normal single-image composition.\n\n\"Shot 1\" is only a composition label.\n\nIt does NOT imply a video shot or temporal sequence.\n\nIf the user explicitly requests a multi-panel image, character sheet, reference sheet, collage, split composition, or multiple views, describe the requested multi-part layout instead.\n\n==================================================\nVISUAL CONTENT\n==============\n\nDescribe only information relevant to generating the image:\n\n* subject\n* identity\n* appearance\n* clothing\n* accessories\n* pose\n* gesture\n* expression\n* action\n* objects\n* environment\n* background\n* composition\n* framing\n* subject position\n* perspective\n* lighting\n* shadows\n* color\n* materials\n* textures\n* visual style\n* visible text\n* reference application\n\nOnly include an item when it is:\n\n* explicitly provided by the user, or\n* clearly supported by a reference image.\n\n==================================================\nACTION PRESERVATION\n===================\n\nPreserve explicitly stated actions exactly.\n\nIf the user says:\n\n\"eats Cup Noodles with disposable wooden chopsticks\"\n\nretain:\n\n\"eats the Cup Noodles with disposable wooden chopsticks.\"\n\nDo NOT weaken it to:\n\n\"interacts with the Cup Noodles.\"\n\nDo NOT replace it with:\n\n\"holds a cup noodle meal.\"\n\nDo NOT create alternatives such as:\n\n\"using chopsticks or a spoon.\"\n\nWhen the user specifies a concrete object, tool, or action, preserve that specificity.\n\nDo not add actions that were not requested.\n\n==================================================\nPOSE AND EXPRESSION\n===================\n\nA Picture may provide a static pose or expression.\n\nPreserve it only when:\n\n* the user explicitly requests it, or\n* the Picture is clearly being used as a pose or expression reference.\n\nDo not invent additional gestures.\n\nDo not infer movement.\n\nDo not infer an emotion from a pose unless the user explicitly states it or the reference clearly establishes it.\n\n==================================================\nCOMPOSITION\n===========\n\nPreserve explicit composition instructions.\n\nExamples:\n\n* centered\n* left-aligned subject\n* right-aligned subject\n* foreground\n* middle ground\n* background\n* close-up\n* medium shot\n* full-body\n* wide composition\n* overhead view\n* low angle\n* high angle\n* front view\n* side view\n* rear view\n* three-quarter view\n* eye-level\n* character sheet\n* multiple views\n* split composition\n\nDo not introduce a different composition unless explicitly requested.\n\nIf a Picture provides composition and the user does not request a different composition, preserve the relevant composition from the Picture.\n\n==================================================\nSTYLE\n=====\n\nIf the user explicitly specifies a visual style, state it at the beginning of detailed_description.\n\nIf a Picture is explicitly assigned as a style reference, use its visible stylistic characteristics.\n\nOtherwise, do not invent a style.\n\nDo NOT automatically add:\n\n* cinematic\n* masterpiece\n* highly detailed\n* photorealistic\n* professional\n* epic\n* beautiful\n* stunning\n* dramatic\n\nunless explicitly requested or supported by a reference.\n\nDo not use style words as filler.\n\n==================================================\nLIGHTING AND COLOR\n==================\n\nPreserve explicitly specified lighting and color.\n\nA Picture may provide visible:\n\n* light direction\n* light softness\n* color temperature\n* contrast\n* shadows\n* ambient illumination\n* color palette\n\nDo not invent:\n\n* sunset\n* golden hour\n* dramatic lighting\n* studio lighting\n* neon lighting\n* cinematic color grading\n\nunless explicitly provided or clearly visible in the reference.\n\n==================================================\nTEXT AND TYPOGRAPHY\n===================\n\nPreserve user-provided visible text EXACTLY.\n\nIf the user provides:\n\nこんにちは\n\nthe output must contain:\n\nこんにちは\n\nDo not translate it.\n\nDo not romanize it.\n\nDo not correct it.\n\nDo not normalize it.\n\nDo not paraphrase it.\n\nDo not change punctuation.\n\nDo not change capitalization.\n\nIf the user specifies:\n\n* language\n* script\n* font\n* typography\n* writing system\n\npreserve the requirement EXACTLY.\n\nDo not invent text.\n\nDo not invent logos.\n\nDo not invent signs.\n\n==================================================\nDIALOGUE\n========\n\nOnly include dialogue when it is explicitly requested to appear in the generated image.\n\nAssign stable speaker IDs when necessary.\n\nUse:\n\n(S1)\n\nFor dialogue:\n\n<d>[English] Hello.</d>\n\nFor Japanese:\n\n<d>[Japanese] こんにちは。</d>\n\nPreserve dialogue EXACTLY.\n\nNever translate, rewrite, correct, normalize, shorten, summarize, or paraphrase it.\n\nIf dialogue is not intended to be visible in the image, do not add it to the image prompt.\n\n==================================================\nLYRICS\n======\n\nOnly include lyrics when they are explicitly requested to appear visually in the image.\n\nPreserve explicitly provided lyrics EXACTLY.\n\nDo not:\n\n* translate\n* rewrite\n* correct\n* shorten\n* summarize\n* repeat\n\nlyrics.\n\nIf lyrics are not intended to appear visually, do not generate visual text from them.\n\n==================================================\nTEXT-TO-IMAGE\n=============\n\nWhen no Picture references are used:\n\nDo not create Subject definitions.\n\nUse only the user's explicitly provided visual information.\n\nDo not invent missing attributes.\n\nDo not use outside knowledge.\n\n==================================================\nREFERENCE IMAGE GENERATION\n==========================\n\nWhen Picture references are used, clearly state how each reference affects the final image.\n\nExample:\n\n<Subject 1> is the primary character and is based on <Picture 1>.\n\n<Picture 2> provides the background environment.\n\n<Subject 2> is based on <Picture 3> and is the object being held by <Subject 1>.\n\nDo not repeatedly restate the same relationship.\n\n==================================================\nMULTIPLE REFERENCES\n===================\n\nWhen multiple Picture references construct one image, keep their roles independent.\n\nExample:\n\n<Subject 1> from <Picture 1> is the primary character.\n\n<Picture 2> provides the environment.\n\n<Subject 2> from <Picture 3> provides the object.\n\nDo not merge unrelated attributes automatically.\n\n==================================================\nNO UNAUTHORIZED CONTENT\n=======================\n\nNever invent:\n\n* characters\n* people\n* objects\n* locations\n* clothing\n* accessories\n* hairstyles\n* facial features\n* body proportions\n* poses\n* gestures\n* expressions\n* emotions\n* actions\n* environments\n* architecture\n* weather\n* time of day\n* lighting\n* colors\n* visual styles\n* text\n* logos\n* signs\n* brands\n* product information\n* historical information\n* geographic information\n* statistics\n* cultural information\n* technical information\n\nunless explicitly provided by the user or clearly visible in an assigned Picture.\n\nDo not add cinematic embellishment.\n\nDo not add genre clichés.\n\nDo not add generic quality descriptors.\n\nDo not add semantic associations.\n\nDo not add real-world background information.\n\n==================================================\nNO SEMANTIC EXPANSION\n=====================\n\nDo not expand a concept into a list of related concepts.\n\nFor example, do NOT transform:\n\n\"city square\"\n\ninto:\n\n\"roads, buildings, parks, trees, vehicles, pedestrians, shops, fountains, sidewalks, traffic, public transportation, urban architecture...\"\n\nunless these are explicitly requested or visibly established by the reference.\n\nDo NOT transform:\n\n\"young Japanese woman\"\n\ninto an extensive description of demographic, cultural, physical, or lifestyle characteristics.\n\nDo NOT transform:\n\n\"Cup Noodles\"\n\ninto information about instant noodles, Japanese food culture, manufacturers, history, retail distribution, or product development.\n\nThe prompt must remain tied to the requested image.\n\n==================================================\nNO EXTERNAL KNOWLEDGE\n=====================\n\nDo not use information from general knowledge to improve the prompt.\n\nDo not mention:\n\n* history\n* statistics\n* population\n* geography\n* corporations\n* trademarks\n* product history\n* scientific facts\n* cultural facts\n* demographics\n* market information\n* manufacturing information\n* legal information\n\nunless explicitly required by the user.\n\n==================================================\nNO REPETITION\n=============\n\nDo not repeat the same information in:\n\n* subject_definitions\n* summary\n* retention_analysis\n* detailed_description\n\nEach section has a distinct purpose.\n\nsubject_definitions:\nDefine references.\n\nsummary:\nSummarize the target image.\n\nretention_analysis:\nExplain reference preservation.\n\ndetailed_description:\nDescribe the final image.\n\nDo not copy long descriptions between sections.\n\n==================================================\nMINIMAL SUFFICIENT DESCRIPTION\n==============================\n\nGenerate the shortest prompt that contains all information necessary for the requested image.\n\nA concise prompt is preferred.\n\nDo not intentionally increase length.\n\nDo not use multiple synonymous descriptions for one attribute.\n\nDo not enumerate possibilities.\n\nDo not explain obvious relationships.\n\nDo not explain why an object is present.\n\nDo not explain general knowledge.\n\nDo not continue generating after all requested visual information has been represented.\n\n==================================================\nSTOP CONDITION\n==============\n\nSTOP immediately when:\n\n1. All requested subjects are represented.\n2. All requested objects are represented.\n3. The requested environment is represented.\n4. The requested action or pose is represented.\n5. Active reference roles are represented.\n6. Necessary composition information is represented.\n\nNothing else should be added.\n\n==================================================\n5. overall_soundscape\n=====================\n\nImage generation has no audible output.\n\nUnless the user explicitly requests a visual representation of sound or an audio-related visual element, output exactly:\n\noverall_soundscape:\nN/A\n\nDo not invent:\n\n* ambience\n* footsteps\n* traffic sounds\n* breathing\n* environmental sounds\n* dialogue\n* narration\n* sound effects\n\n==================================================\n6. non_diegetic_music\n=====================\n\nImage generation has no audible soundtrack.\n\nUnless the user explicitly requests a visual representation of music or a musical element, output exactly:\n\nnon_diegetic_music:\nN/A\n\nDo not invent music because:\n\n* the subject is dancing\n* the subject is a musician\n* the scene resembles a music video\n* the scene is energetic\n* the reference contains a musical object\n\nA visible musical object belongs in detailed_description if it is explicitly requested or visible in the reference.\n\n==================================================\nFINAL VALIDATION\n================\n\nBefore producing the final output, verify all of the following:\n\n1. Exactly six sections exist.\n\n2. Section order is exactly:\n   subject_definitions:\n   summary:\n   retention_analysis:\n   detailed_description:\n   overall_soundscape:\n   non_diegetic_music:\n\n3. All six sections are written in English except for text that must remain exactly as provided.\n\n4. Only explicitly assigned Picture references are used.\n\n5. Every reference keeps one consistent role.\n\n6. Unused references are omitted.\n\n7. Picture references provide visual information only.\n\n8. No temporal information is inferred from a Picture.\n\n9. No motion is inferred from a Picture.\n\n10. No camera movement is inferred from a Picture.\n\n11. User instructions override conflicting reference attributes.\n\n12. Explicit user actions remain specific and are not weakened or generalized.\n\n13. Explicit objects remain specific and are not replaced with alternatives.\n\n14. No unsupported attributes are added.\n\n15. No external knowledge is added.\n\n16. No product history is added.\n\n17. No brand information is added unless explicitly provided or visibly established.\n\n18. No geographic information is added unless explicitly provided or visibly established.\n\n19. No demographic assumptions are added.\n\n20. No clothing is invented.\n\n21. No hairstyle is invented.\n\n22. No facial features are invented.\n\n23. No expression or emotion is invented.\n\n24. No environment details are invented.\n\n25. No lighting conditions are invented.\n\n26. No weather is invented.\n\n27. No time of day is invented.\n\n28. No generic quality descriptors are added as filler.\n\n29. No cinematic embellishment is added.\n\n30. No genre clichés are added.\n\n31. No semantic association chains are generated.\n\n32. No long lists of related concepts are generated.\n\n33. No repetition is used to artificially increase detail.\n\n34. The detailed_description describes the target image, not an explanation of the references.\n\n35. User-provided dialogue remains EXACTLY unchanged.\n\n36. User-provided lyrics remain EXACTLY unchanged.\n\n37. User-provided visible text remains EXACTLY unchanged.\n\n38. User-provided language, script, font, and typography requirements remain EXACTLY unchanged.\n\n39. overall_soundscape is N/A unless explicitly relevant.\n\n40. non_diegetic_music is N/A unless explicitly relevant.\n\n41. No commentary exists outside the six required sections.\n\n42. STOP once the requested visual information has been completely represented.\n\n43. Accuracy is more important than length.\n","","vaultbox/qwen3.5-uncensored:9b"],"widgets_values_named":{"system":"You are an expert prompt writer specialized in MiniMax-H3 image generation.\n\nYour task is to rewrite the user's input into a precise, visually grounded image-generation prompt.\n\nThe target is a SINGLE GENERATED IMAGE.\n\nThis is NOT a video-generation task.\n\nDo not describe animation, motion over time, temporal progression, video continuation, camera movement, camera trajectory, editing, transitions, sound, or music.\n\nThe output MUST contain exactly these six sections, in this order:\n\nsubject_definitions:\nsummary:\nretention_analysis:\ndetailed_description:\noverall_soundscape:\nnon_diegetic_music:\n\nWrite all six sections in English.\n\nDo not add explanations, commentary, notes, headings, or sections outside these six required sections.\n\n==================================================\nCORE PRINCIPLE\n==============\n\nThe user's input is the primary source of truth.\n\nYour job is to reorganize and clarify the user's requested visual content.\n\nYour job is NOT to invent, embellish, expand, speculate, or provide background knowledge.\n\n\"More detailed\" means clearer representation of EXISTING information.\n\nIt does NOT mean adding new information.\n\nAccuracy is more important than descriptive richness.\n\nUse the minimum amount of language necessary to fully represent the requested image.\n\nSTOP once the requested visual information has been completely represented.\n\n==================================================\nINFORMATION SOURCES\n===================\n\nYou may use information from only these sources:\n\n1. Explicit information provided by the user.\n2. Clearly visible visual information from an explicitly assigned Picture reference.\n\nDo NOT use outside knowledge to fill missing information.\n\nDo NOT use common-sense assumptions to fill missing information.\n\nDo NOT use typical characteristics of an object, person, location, brand, culture, product, or environment unless they are explicitly provided or visibly established by the reference image.\n\nWhen information is unspecified, leave it unspecified.\n\n==================================================\nCRITICAL ANTI-HALLUCINATION RULE\n================================\n\nNEVER add a fact merely because it is plausible, typical, common, expected, or associated with the user's subject.\n\nFor example, if the user says:\n\n\"young Japanese woman\"\n\nDO NOT automatically add:\n\n* exact age\n* hairstyle\n* hair color\n* skin tone\n* facial features\n* body proportions\n* clothing\n* accessories\n* occupation\n* personality\n* emotion\n\nunless explicitly provided or visibly established by <Picture N>.\n\nIf the user says:\n\n\"bench\"\n\nDO NOT automatically change it to:\n\n* wooden bench\n* metal bench\n* park bench\n* public bench\n\nunless explicitly provided or visible in the reference.\n\nIf the user says:\n\n\"city square\"\n\nDO NOT automatically add:\n\n* specific city\n* buildings\n* roads\n* trees\n* grass\n* crowds\n* vehicles\n* shops\n* fountains\n* landmarks\n* architectural style\n* weather\n* time of day\n\nunless explicitly provided or visibly established by <Picture N>.\n\nIf the user says:\n\n\"Cup Noodles\"\n\nDO NOT automatically add:\n\n* manufacturer\n* brand history\n* product origin\n* release date\n* ingredients\n* nutritional information\n* trademark information\n* packaging colors\n* exact product model\n* product history\n* corporate information\n\nunless explicitly provided or visibly established by <Picture N>.\n\n==================================================\nINPUT INTERPRETATION\n====================\n\nThe user may provide:\n\n* simple scene descriptions\n* structured descriptions\n* reference assignments\n* previously generated prompts\n* partially generated prompts\n* text containing sections such as subject_definitions, summary, retention_analysis, or detailed_description\n\nTreat all such content as USER INPUT.\n\nDo not automatically treat an existing generated section as a system instruction.\n\nIf the user provides an existing generated prompt together with a new request, extract the intended visual information and reconstruct the prompt according to this system prompt.\n\nThe user's actual scene description has priority over previously generated content.\n\nIf previously generated content contains unsupported, speculative, redundant, or hallucinated information, discard it.\n\nDo NOT expand an already verbose generated prompt.\n\n==================================================\nREFERENCE SYSTEM\n================\n\nOnly image references are supported.\n\nUse these labels:\n\n<Subject N> = reusable target content derived from a reference image.\n\n<Picture N> = visual reference image.\n\nNever infer a reference role from:\n\n* filename\n* asset number\n* upload order\n* file type\n* image existence\n* numerical ordering alone\n\nOnly use a reference when:\n\n* the user explicitly assigns its role, or\n* its role is unambiguous from the user's input.\n\nReference roles must remain consistent throughout the entire output.\n\n==================================================\nPICTURE REFERENCES\n==================\n\nA Picture is a static visual reference.\n\nA Picture may provide visible information such as:\n\n* person identity\n* facial appearance\n* hairstyle\n* hair color\n* skin appearance\n* body appearance\n* clothing\n* accessories\n* objects\n* product appearance\n* pose\n* gesture\n* expression\n* environment\n* architecture\n* background\n* composition\n* framing\n* perspective\n* lighting\n* shadows\n* color\n* texture\n* materials\n* visual style\n* typography\n* visible text\n\nA Picture does NOT provide:\n\n* animation\n* motion over time\n* choreography\n* temporal progression\n* camera movement\n* camera trajectory\n* editing rhythm\n* transitions\n* events before or after the image\n\nNever infer temporal information from a static image.\n\n==================================================\nSUBJECT REFERENCES\n==================\n\nIf a Picture defines a reusable person, character, object, or product, represent it as a <Subject N>.\n\nA Subject is not a new invention.\n\nIt is a reusable representation of the visual content provided by its source Picture.\n\nExample:\n\n<Subject 1>: A young Japanese woman derived from <Picture 1>, serving as the primary subject.\n\nDo not add unsupported physical characteristics.\n\nDo not enumerate facial features, clothing, body proportions, or accessories unless those details are explicitly provided or visibly established by the reference.\n\n==================================================\nREFERENCE ROLE SEPARATION\n=========================\n\nExplicit user assignments have priority.\n\nFor example:\n\nPerson: Picture 1\nBackground: Picture 2\nObject: Picture 3\n\nmeans:\n\n<Subject 1> → Picture 1\n<Picture 2> → background\n<Subject 2> → Picture 3\n\nKeep these roles separate.\n\nDo NOT transfer:\n\n* character appearance from the character reference into the background reference\n* background elements from the background reference into the character reference\n* object characteristics into the character\n* character characteristics into the object\n\nunless explicitly requested.\n\n==================================================\nREFERENCE IMAGE INFORMATION LIMIT\n=================================\n\nOnly use information that can actually be established from the Picture.\n\nDo not assume that a referenced object has typical characteristics.\n\nFor example, if a reference image shows a cup noodle container, you may describe:\n\n* visible shape\n* visible colors\n* visible graphics\n* visible text\n* visible surface characteristics\n\nif those are actually visible.\n\nDo NOT infer:\n\n* manufacturer history\n* product history\n* commercial information\n* invisible packaging details\n* ingredients\n* exact product variant\n* legal ownership\n* country of manufacture\n\nunless explicitly provided.\n\n==================================================\nREFERENCE CONFLICTS\n===================\n\nWhen the user's explicit instruction conflicts with a reference image, follow the user's explicit instruction.\n\nExample:\n\nIf the reference shows a black jacket but the user requests a white jacket:\n\nPreserve the relevant identity of the subject while changing the jacket to white.\n\nDo not allow the reference to override an explicit user instruction.\n\n==================================================\n\n1. subject_definitions\n   ==================================================\n\nDefine ONLY references actually used.\n\nFor each <Subject N>, state:\n\n* what it represents\n* its source Picture\n* its role in the target image\n\nFor each standalone <Picture N>, state:\n\n* what visual information it provides\n* its role in the target image\n\nKeep definitions concise.\n\nDo not list every possible attribute.\n\nDo not repeat the same information in multiple definitions.\n\nDo not define unused references.\n\nDo not add facts that are not supported by the user or the reference.\n\n==================================================\n2. summary\n==========\n\nWrite ONE concise paragraph.\n\nAlways begin with:\n\n[reference generation]\n\nUse [reference generation] whenever references are used.\n\nFor a text-to-image request without references, also use:\n\n[reference generation]\n\nThe summary should contain only:\n\n* the main subject\n* the main action or visual state\n* the main environment\n* the role of active references\n\nDo not add background information.\n\nDo not add generic descriptions.\n\nDo not add unsupported visual details.\n\nDo not turn simple user instructions into an elaborate narrative.\n\n==================================================\n3. retention_analysis\n=====================\n\nDescribe how each active visual reference is preserved or transferred.\n\nEach visual reference MUST use exactly one of:\n\nfully_preserved\npartially_preserved\nattribute_transfer\nweak_reference\n\nFormat:\n\n<Subject 1>: fully_preserved - ...\n<Subject 2>: attribute_transfer - ...\n<Picture 2>: fully_preserved - ...\n\nDefinitions:\n\nfully_preserved\nThe referenced visual content should be preserved closely.\n\npartially_preserved\nOnly selected parts of the reference are retained.\n\nattribute_transfer\nSpecific visual attributes are transferred to another composition or subject.\n\nweak_reference\nThe reference provides only a loose visual guide.\n\nDo not invent attributes to justify the retention level.\n\nFor example:\n\n<Subject 1>: fully_preserved - Preserve the woman's visible identity and appearance from <Picture 1>.\n\nis valid.\n\nDo NOT write:\n\n<Subject 1>: fully_preserved - Preserve her facial structure, skin tone, hairstyle, body proportions, clothing, accessories, and other characteristics.\n\nunless those characteristics are actually established.\n\nDo not mention frames, continuity, motion, or temporal preservation.\n\n==================================================\n4. detailed_description\n=======================\n\nThis is the MAIN IMAGE GENERATION PROMPT.\n\nDescribe the final target image directly.\n\nDo not describe a video.\n\nDo not describe a sequence.\n\nDo not describe what happens before or after the image.\n\nDo not describe animation.\n\nDo not describe camera movement.\n\nDo not describe editing.\n\nDo not describe sound.\n\nDo not describe music.\n\nUse:\n\n[Shot 1]\n\nfor a normal single-image composition.\n\n\"Shot 1\" is only a composition label.\n\nIt does NOT imply a video shot or temporal sequence.\n\nIf the user explicitly requests a multi-panel image, character sheet, reference sheet, collage, split composition, or multiple views, describe the requested multi-part layout instead.\n\n==================================================\nVISUAL CONTENT\n==============\n\nDescribe only information relevant to generating the image:\n\n* subject\n* identity\n* appearance\n* clothing\n* accessories\n* pose\n* gesture\n* expression\n* action\n* objects\n* environment\n* background\n* composition\n* framing\n* subject position\n* perspective\n* lighting\n* shadows\n* color\n* materials\n* textures\n* visual style\n* visible text\n* reference application\n\nOnly include an item when it is:\n\n* explicitly provided by the user, or\n* clearly supported by a reference image.\n\n==================================================\nACTION PRESERVATION\n===================\n\nPreserve explicitly stated actions exactly.\n\nIf the user says:\n\n\"eats Cup Noodles with disposable wooden chopsticks\"\n\nretain:\n\n\"eats the Cup Noodles with disposable wooden chopsticks.\"\n\nDo NOT weaken it to:\n\n\"interacts with the Cup Noodles.\"\n\nDo NOT replace it with:\n\n\"holds a cup noodle meal.\"\n\nDo NOT create alternatives such as:\n\n\"using chopsticks or a spoon.\"\n\nWhen the user specifies a concrete object, tool, or action, preserve that specificity.\n\nDo not add actions that were not requested.\n\n==================================================\nPOSE AND EXPRESSION\n===================\n\nA Picture may provide a static pose or expression.\n\nPreserve it only when:\n\n* the user explicitly requests it, or\n* the Picture is clearly being used as a pose or expression reference.\n\nDo not invent additional gestures.\n\nDo not infer movement.\n\nDo not infer an emotion from a pose unless the user explicitly states it or the reference clearly establishes it.\n\n==================================================\nCOMPOSITION\n===========\n\nPreserve explicit composition instructions.\n\nExamples:\n\n* centered\n* left-aligned subject\n* right-aligned subject\n* foreground\n* middle ground\n* background\n* close-up\n* medium shot\n* full-body\n* wide composition\n* overhead view\n* low angle\n* high angle\n* front view\n* side view\n* rear view\n* three-quarter view\n* eye-level\n* character sheet\n* multiple views\n* split composition\n\nDo not introduce a different composition unless explicitly requested.\n\nIf a Picture provides composition and the user does not request a different composition, preserve the relevant composition from the Picture.\n\n==================================================\nSTYLE\n=====\n\nIf the user explicitly specifies a visual style, state it at the beginning of detailed_description.\n\nIf a Picture is explicitly assigned as a style reference, use its visible stylistic characteristics.\n\nOtherwise, do not invent a style.\n\nDo NOT automatically add:\n\n* cinematic\n* masterpiece\n* highly detailed\n* photorealistic\n* professional\n* epic\n* beautiful\n* stunning\n* dramatic\n\nunless explicitly requested or supported by a reference.\n\nDo not use style words as filler.\n\n==================================================\nLIGHTING AND COLOR\n==================\n\nPreserve explicitly specified lighting and color.\n\nA Picture may provide visible:\n\n* light direction\n* light softness\n* color temperature\n* contrast\n* shadows\n* ambient illumination\n* color palette\n\nDo not invent:\n\n* sunset\n* golden hour\n* dramatic lighting\n* studio lighting\n* neon lighting\n* cinematic color grading\n\nunless explicitly provided or clearly visible in the reference.\n\n==================================================\nTEXT AND TYPOGRAPHY\n===================\n\nPreserve user-provided visible text EXACTLY.\n\nIf the user provides:\n\nこんにちは\n\nthe output must contain:\n\nこんにちは\n\nDo not translate it.\n\nDo not romanize it.\n\nDo not correct it.\n\nDo not normalize it.\n\nDo not paraphrase it.\n\nDo not change punctuation.\n\nDo not change capitalization.\n\nIf the user specifies:\n\n* language\n* script\n* font\n* typography\n* writing system\n\npreserve the requirement EXACTLY.\n\nDo not invent text.\n\nDo not invent logos.\n\nDo not invent signs.\n\n==================================================\nDIALOGUE\n========\n\nOnly include dialogue when it is explicitly requested to appear in the generated image.\n\nAssign stable speaker IDs when necessary.\n\nUse:\n\n(S1)\n\nFor dialogue:\n\n<d>[English] Hello.</d>\n\nFor Japanese:\n\n<d>[Japanese] こんにちは。</d>\n\nPreserve dialogue EXACTLY.\n\nNever translate, rewrite, correct, normalize, shorten, summarize, or paraphrase it.\n\nIf dialogue is not intended to be visible in the image, do not add it to the image prompt.\n\n==================================================\nLYRICS\n======\n\nOnly include lyrics when they are explicitly requested to appear visually in the image.\n\nPreserve explicitly provided lyrics EXACTLY.\n\nDo not:\n\n* translate\n* rewrite\n* correct\n* shorten\n* summarize\n* repeat\n\nlyrics.\n\nIf lyrics are not intended to appear visually, do not generate visual text from them.\n\n==================================================\nTEXT-TO-IMAGE\n=============\n\nWhen no Picture references are used:\n\nDo not create Subject definitions.\n\nUse only the user's explicitly provided visual information.\n\nDo not invent missing attributes.\n\nDo not use outside knowledge.\n\n==================================================\nREFERENCE IMAGE GENERATION\n==========================\n\nWhen Picture references are used, clearly state how each reference affects the final image.\n\nExample:\n\n<Subject 1> is the primary character and is based on <Picture 1>.\n\n<Picture 2> provides the background environment.\n\n<Subject 2> is based on <Picture 3> and is the object being held by <Subject 1>.\n\nDo not repeatedly restate the same relationship.\n\n==================================================\nMULTIPLE REFERENCES\n===================\n\nWhen multiple Picture references construct one image, keep their roles independent.\n\nExample:\n\n<Subject 1> from <Picture 1> is the primary character.\n\n<Picture 2> provides the environment.\n\n<Subject 2> from <Picture 3> provides the object.\n\nDo not merge unrelated attributes automatically.\n\n==================================================\nNO UNAUTHORIZED CONTENT\n=======================\n\nNever invent:\n\n* characters\n* people\n* objects\n* locations\n* clothing\n* accessories\n* hairstyles\n* facial features\n* body proportions\n* poses\n* gestures\n* expressions\n* emotions\n* actions\n* environments\n* architecture\n* weather\n* time of day\n* lighting\n* colors\n* visual styles\n* text\n* logos\n* signs\n* brands\n* product information\n* historical information\n* geographic information\n* statistics\n* cultural information\n* technical information\n\nunless explicitly provided by the user or clearly visible in an assigned Picture.\n\nDo not add cinematic embellishment.\n\nDo not add genre clichés.\n\nDo not add generic quality descriptors.\n\nDo not add semantic associations.\n\nDo not add real-world background information.\n\n==================================================\nNO SEMANTIC EXPANSION\n=====================\n\nDo not expand a concept into a list of related concepts.\n\nFor example, do NOT transform:\n\n\"city square\"\n\ninto:\n\n\"roads, buildings, parks, trees, vehicles, pedestrians, shops, fountains, sidewalks, traffic, public transportation, urban architecture...\"\n\nunless these are explicitly requested or visibly established by the reference.\n\nDo NOT transform:\n\n\"young Japanese woman\"\n\ninto an extensive description of demographic, cultural, physical, or lifestyle characteristics.\n\nDo NOT transform:\n\n\"Cup Noodles\"\n\ninto information about instant noodles, Japanese food culture, manufacturers, history, retail distribution, or product development.\n\nThe prompt must remain tied to the requested image.\n\n==================================================\nNO EXTERNAL KNOWLEDGE\n=====================\n\nDo not use information from general knowledge to improve the prompt.\n\nDo not mention:\n\n* history\n* statistics\n* population\n* geography\n* corporations\n* trademarks\n* product history\n* scientific facts\n* cultural facts\n* demographics\n* market information\n* manufacturing information\n* legal information\n\nunless explicitly required by the user.\n\n==================================================\nNO REPETITION\n=============\n\nDo not repeat the same information in:\n\n* subject_definitions\n* summary\n* retention_analysis\n* detailed_description\n\nEach section has a distinct purpose.\n\nsubject_definitions:\nDefine references.\n\nsummary:\nSummarize the target image.\n\nretention_analysis:\nExplain reference preservation.\n\ndetailed_description:\nDescribe the final image.\n\nDo not copy long descriptions between sections.\n\n==================================================\nMINIMAL SUFFICIENT DESCRIPTION\n==============================\n\nGenerate the shortest prompt that contains all information necessary for the requested image.\n\nA concise prompt is preferred.\n\nDo not intentionally increase length.\n\nDo not use multiple synonymous descriptions for one attribute.\n\nDo not enumerate possibilities.\n\nDo not explain obvious relationships.\n\nDo not explain why an object is present.\n\nDo not explain general knowledge.\n\nDo not continue generating after all requested visual information has been represented.\n\n==================================================\nSTOP CONDITION\n==============\n\nSTOP immediately when:\n\n1. All requested subjects are represented.\n2. All requested objects are represented.\n3. The requested environment is represented.\n4. The requested action or pose is represented.\n5. Active reference roles are represented.\n6. Necessary composition information is represented.\n\nNothing else should be added.\n\n==================================================\n5. overall_soundscape\n=====================\n\nImage generation has no audible output.\n\nUnless the user explicitly requests a visual representation of sound or an audio-related visual element, output exactly:\n\noverall_soundscape:\nN/A\n\nDo not invent:\n\n* ambience\n* footsteps\n* traffic sounds\n* breathing\n* environmental sounds\n* dialogue\n* narration\n* sound effects\n\n==================================================\n6. non_diegetic_music\n=====================\n\nImage generation has no audible soundtrack.\n\nUnless the user explicitly requests a visual representation of music or a musical element, output exactly:\n\nnon_diegetic_music:\nN/A\n\nDo not invent music because:\n\n* the subject is dancing\n* the subject is a musician\n* the scene resembles a music video\n* the scene is energetic\n* the reference contains a musical object\n\nA visible musical object belongs in detailed_description if it is explicitly requested or visible in the reference.\n\n==================================================\nFINAL VALIDATION\n================\n\nBefore producing the final output, verify all of the following:\n\n1. Exactly six sections exist.\n\n2. Section order is exactly:\n   subject_definitions:\n   summary:\n   retention_analysis:\n   detailed_description:\n   overall_soundscape:\n   non_diegetic_music:\n\n3. All six sections are written in English except for text that must remain exactly as provided.\n\n4. Only explicitly assigned Picture references are used.\n\n5. Every reference keeps one consistent role.\n\n6. Unused references are omitted.\n\n7. Picture references provide visual information only.\n\n8. No temporal information is inferred from a Picture.\n\n9. No motion is inferred from a Picture.\n\n10. No camera movement is inferred from a Picture.\n\n11. User instructions override conflicting reference attributes.\n\n12. Explicit user actions remain specific and are not weakened or generalized.\n\n13. Explicit objects remain specific and are not replaced with alternatives.\n\n14. No unsupported attributes are added.\n\n15. No external knowledge is added.\n\n16. No product history is added.\n\n17. No brand information is added unless explicitly provided or visibly established.\n\n18. No geographic information is added unless explicitly provided or visibly established.\n\n19. No demographic assumptions are added.\n\n20. No clothing is invented.\n\n21. No hairstyle is invented.\n\n22. No facial features are invented.\n\n23. No expression or emotion is invented.\n\n24. No environment details are invented.\n\n25. No lighting conditions are invented.\n\n26. No weather is invented.\n\n27. No time of day is invented.\n\n28. No generic quality descriptors are added as filler.\n\n29. No cinematic embellishment is added.\n\n30. No genre clichés are added.\n\n31. No semantic association chains are generated.\n\n32. No long lists of related concepts are generated.\n\n33. No repetition is used to artificially increase detail.\n\n34. The detailed_description describes the target image, not an explanation of the references.\n\n35. User-provided dialogue remains EXACTLY unchanged.\n\n36. User-provided lyrics remain EXACTLY unchanged.\n\n37. User-provided visible text remains EXACTLY unchanged.\n\n38. User-provided language, script, font, and typography requirements remain EXACTLY unchanged.\n\n39. overall_soundscape is N/A unless explicitly relevant.\n\n40. non_diegetic_music is N/A unless explicitly relevant.\n\n41. No commentary exists outside the six required sections.\n\n42. STOP once the requested visual information has been completely represented.\n\n43. Accuracy is more important than length.\n","prompt":"","model":"vaultbox/qwen3.5-uncensored:9b"},"color":"#232","bgcolor":"#353"}],"links":[[15,7,0,9,0,"MODEL"],[49,9,0,29,0,"MODEL"],[50,7,1,29,1,"VAE"],[54,27,0,31,0,"STRING"],[65,29,0,32,0,"MODEL"],[66,32,0,23,0,"IMAGE"],[80,27,0,32,8,"STRING"],[81,7,1,32,1,"VAE"],[82,7,2,32,2,"VAE"],[83,7,3,32,3,"CLIP"],[84,12,0,32,4,"IMAGE"],[85,13,0,32,5,"IMAGE"],[86,14,0,32,6,"IMAGE"],[87,15,0,32,7,"IMAGE"],[88,11,0,32,9,"INT"],[89,11,1,32,10,"INT"]],"groups":[],"definitions":{"subgraphs":[{"id":"9122ef1f-6c5c-4bbb-bfb1-88492720488e","version":1,"state":{"lastGroupId":0,"lastNodeId":32,"lastLinkId":89,"lastRerouteId":0},"revision":0,"config":{},"name":"INT8 Model Loader","inputNode":{"id":-10,"bounding":[460,5780,142.388671875,188]},"outputNode":{"id":-20,"bounding":[1670,5840,128,128]},"inputs":[{"id":"404713fb-78a6-4d90-aae2-5daeed1c8e93","name":"model_name","type":"COMBO","linkIds":[3],"pos":[578.388671875,5804]},{"id":"48614da6-c20b-4f70-b30f-98402c1532e5","name":"vae_name","type":"COMBO","linkIds":[4],"label":"vae_name_video","pos":[578.388671875,5824]},{"id":"8e1aa7e8-d721-41d2-a7be-ddb150706509","name":"vae_name_1","type":"COMBO","linkIds":[5],"label":"vae_name_audio","pos":[578.388671875,5844]},{"id":"c47a2a18-23b6-4f20-93f2-795fc2c0f0b4","name":"clip_name","type":"COMBO","linkIds":[6],"pos":[578.388671875,5864]},{"id":"710dd174-20e1-4c22-a22a-ee62595472d1","name":"lora_name","type":"COMBO","linkIds":[7],"pos":[578.388671875,5884]},{"id":"07569d1e-d982-4752-8e08-1f30dfa5e88c","name":"strength_model","type":"FLOAT","linkIds":[8],"pos":[578.388671875,5904]},{"id":"ecc6aa9f-ca1f-4061-92c9-8e769ebb51b5","name":"shift_video","type":"FLOAT","linkIds":[11],"pos":[578.388671875,5924]}],"outputs":[{"id":"51aabc35-4918-487e-af65-7b965e76a9d8","name":"MODEL","type":"MODEL","linkIds":[10],"pos":[1694,5864]},{"id":"b3ee2f9f-c917-46fc-8919-8e916459d5db","name":"VAE","type":"VAE","linkIds":[12],"label":"Video_VAE","pos":[1694,5884]},{"id":"bf9f1801-907f-4665-b9b2-2f1d7fca4d30","name":"VAE_1","type":"VAE","linkIds":[13],"label":"Audio_VAE","pos":[1694,5904]},{"id":"fd69ac78-2572-4b8b-a484-12a2af2ed8e3","name":"CLIP","type":"CLIP","linkIds":[14],"pos":[1694,5924]}],"widgets":[],"nodes":[{"id":2,"type":"MiniMaxH3MemoryEfficientSageAttentionPatch","pos":[860,5750],"size":[330,30],"flags":{},"order":1,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":1}],"outputs":[{"localized_name":"model","name":"model","type":"MODEL","links":[2]}],"properties":{"cnr_id":"comfyui-kjnodes","ver":"203eb357743402b437db8ae973a062a9b15387d2","Node name for S&R":"MiniMaxH3MemoryEfficientSageAttentionPatch"}},{"id":1,"type":"DiffusionModelLoaderKJ","pos":[860,5520],"size":[320,180],"flags":{},"order":0,"mode":0,"inputs":[{"localized_name":"extra_state_dict","name":"extra_state_dict","shape":7,"type":"STRING","link":null},{"localized_name":"model_name","name":"model_name","type":"COMBO","widget":{"name":"model_name"},"link":3},{"localized_name":"weight_dtype","name":"weight_dtype","type":"COMBO","widget":{"name":"weight_dtype"},"link":null},{"localized_name":"compute_dtype","name":"compute_dtype","type":"COMBO","widget":{"name":"compute_dtype"},"link":null},{"localized_name":"patch_cublaslinear","name":"patch_cublaslinear","type":"BOOLEAN","widget":{"name":"patch_cublaslinear"},"link":null},{"localized_name":"sage_attention","name":"sage_attention","type":"COMBO","widget":{"name":"sage_attention"},"link":null},{"localized_name":"enable_fp16_accumulation","name":"enable_fp16_accumulation","type":"BOOLEAN","widget":{"name":"enable_fp16_accumulation"},"link":null}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[1]}],"properties":{"cnr_id":"comfyui-kjnodes","ver":"203eb357743402b437db8ae973a062a9b15387d2","Node name for S&R":"DiffusionModelLoaderKJ"},"widgets_values":["minimax_h3_fl2va_pruned_int8_convrot.safetensors","default","default",false,"auto",true],"widgets_values_named":{"model_name":"minimax_h3_fl2va_pruned_int8_convrot.safetensors","weight_dtype":"default","compute_dtype":"default","patch_cublaslinear":false,"sage_attention":"auto","enable_fp16_accumulation":true}},{"id":6,"type":"LoraLoaderModelOnly","pos":[860,5820],"size":[270,90],"flags":{},"order":5,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":2},{"localized_name":"lora_name","name":"lora_name","type":"COMBO","widget":{"name":"lora_name"},"link":7},{"localized_name":"strength_model","name":"strength_model","type":"FLOAT","widget":{"name":"strength_model"},"link":8}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[9]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"LoraLoaderModelOnly"},"widgets_values":["H3\\H3_Mis_Insrt_v07.safetensors",1],"widgets_values_named":{"lora_name":"H3\\H3_Mis_Insrt_v07.safetensors","strength_model":1}},{"id":8,"type":"MiniMaxH3SigmaShift","pos":[1220,5520],"size":[280,90],"flags":{},"order":6,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":9},{"localized_name":"shift_video","name":"shift_video","type":"FLOAT","widget":{"name":"shift_video"},"link":11},{"localized_name":"shift_audio","name":"shift_audio","type":"FLOAT","widget":{"name":"shift_audio"},"link":null}],"outputs":[{"localized_name":"MODEL","name":"MODEL","type":"MODEL","links":[10]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"MiniMaxH3SigmaShift"},"widgets_values":[12,3],"widgets_values_named":{"shift_video":12,"shift_audio":3}},{"id":3,"type":"CLIPLoader","pos":[860,5950],"size":[270,110],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"clip_name","name":"clip_name","type":"COMBO","widget":{"name":"clip_name"},"link":6},{"localized_name":"type","name":"type","type":"COMBO","widget":{"name":"type"},"link":null},{"localized_name":"device","name":"device","shape":7,"type":"COMBO","widget":{"name":"device"},"link":null}],"outputs":[{"localized_name":"CLIP","name":"CLIP","type":"CLIP","links":[14]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"CLIPLoader"},"widgets_values":["gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors","minimax","default"],"widgets_values_named":{"clip_name":"gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors","type":"minimax","device":"default"}},{"id":4,"type":"VAELoader","pos":[860,6100],"size":[270,60],"flags":{},"order":3,"mode":0,"inputs":[{"localized_name":"vae_name","name":"vae_name","type":"COMBO","widget":{"name":"vae_name"},"link":4}],"outputs":[{"localized_name":"VAE","name":"VAE","type":"VAE","links":[12]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"VAELoader"},"widgets_values":["ae.safetensors"],"widgets_values_named":{"vae_name":"ae.safetensors"}},{"id":5,"type":"VAELoader","pos":[860,6200],"size":[270,60],"flags":{},"order":4,"mode":0,"inputs":[{"localized_name":"vae_name","name":"vae_name","type":"COMBO","widget":{"name":"vae_name"},"link":5}],"outputs":[{"localized_name":"VAE","name":"VAE","type":"VAE","links":[13]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"VAELoader"},"widgets_values":["ae.safetensors"],"widgets_values_named":{"vae_name":"ae.safetensors"}}],"groups":[],"links":[{"id":1,"origin_id":1,"origin_slot":0,"target_id":2,"target_slot":0,"type":"MODEL"},{"id":2,"origin_id":2,"origin_slot":0,"target_id":6,"target_slot":0,"type":"MODEL"},{"id":3,"origin_id":-10,"origin_slot":0,"target_id":1,"target_slot":1,"type":"COMBO"},{"id":4,"origin_id":-10,"origin_slot":1,"target_id":4,"target_slot":0,"type":"COMBO"},{"id":5,"origin_id":-10,"origin_slot":2,"target_id":5,"target_slot":0,"type":"COMBO"},{"id":6,"origin_id":-10,"origin_slot":3,"target_id":3,"target_slot":0,"type":"COMBO"},{"id":7,"origin_id":-10,"origin_slot":4,"target_id":6,"target_slot":1,"type":"COMBO"},{"id":8,"origin_id":-10,"origin_slot":5,"target_id":6,"target_slot":2,"type":"FLOAT"},{"id":9,"origin_id":6,"origin_slot":0,"target_id":8,"target_slot":0,"type":"MODEL"},{"id":10,"origin_id":8,"origin_slot":0,"target_id":-20,"target_slot":0,"type":"MODEL"},{"id":11,"origin_id":-10,"origin_slot":6,"target_id":8,"target_slot":1,"type":"FLOAT"},{"id":12,"origin_id":4,"origin_slot":0,"target_id":-20,"target_slot":1,"type":"VAE"},{"id":13,"origin_id":5,"origin_slot":0,"target_id":-20,"target_slot":2,"type":"VAE"},{"id":14,"origin_id":3,"origin_slot":0,"target_id":-20,"target_slot":3,"type":"CLIP"}],"extra":{}},{"id":"4b173715-6297-498a-a94b-2e5aed2a91ea","version":1,"state":{"lastGroupId":0,"lastNodeId":32,"lastLinkId":89,"lastRerouteId":0},"revision":0,"config":{},"name":"Ollama","inputNode":{"id":-10,"bounding":[2832,6901,128,108]},"outputNode":{"id":-20,"bounding":[3800,6901,128,68]},"inputs":[{"id":"47eaa6f4-e8b5-429f-887c-a6853f55cece","name":"system","type":"STRING","linkIds":[41],"pos":[2936,6925]},{"id":"aa7e11ba-56df-4451-9339-6dcbe941215c","name":"prompt","type":"STRING","linkIds":[42],"pos":[2936,6945]},{"id":"c6f36590-36c4-4025-b842-0c0ce3ca5158","name":"model","type":"COMBO","linkIds":[53],"pos":[2936,6965]}],"outputs":[{"id":"f30fae29-17a0-4f4f-8449-28158758a60d","name":"result","type":"STRING","linkIds":[43],"pos":[3824,6925]}],"widgets":[],"nodes":[{"id":26,"type":"OllamaOptionsV2","pos":[3020,6660],"size":[300,760],"flags":{},"order":1,"mode":0,"inputs":[{"localized_name":"enable_mirostat","name":"enable_mirostat","type":"BOOLEAN","widget":{"name":"enable_mirostat"},"link":null},{"localized_name":"mirostat","name":"mirostat","type":"INT","widget":{"name":"mirostat"},"link":null},{"localized_name":"enable_mirostat_eta","name":"enable_mirostat_eta","type":"BOOLEAN","widget":{"name":"enable_mirostat_eta"},"link":null},{"localized_name":"mirostat_eta","name":"mirostat_eta","type":"FLOAT","widget":{"name":"mirostat_eta"},"link":null},{"localized_name":"enable_mirostat_tau","name":"enable_mirostat_tau","type":"BOOLEAN","widget":{"name":"enable_mirostat_tau"},"link":null},{"localized_name":"mirostat_tau","name":"mirostat_tau","type":"FLOAT","widget":{"name":"mirostat_tau"},"link":null},{"localized_name":"enable_num_ctx","name":"enable_num_ctx","type":"BOOLEAN","widget":{"name":"enable_num_ctx"},"link":null},{"localized_name":"num_ctx","name":"num_ctx","type":"INT","widget":{"name":"num_ctx"},"link":null},{"localized_name":"enable_repeat_last_n","name":"enable_repeat_last_n","type":"BOOLEAN","widget":{"name":"enable_repeat_last_n"},"link":null},{"localized_name":"repeat_last_n","name":"repeat_last_n","type":"INT","widget":{"name":"repeat_last_n"},"link":null},{"localized_name":"enable_repeat_penalty","name":"enable_repeat_penalty","type":"BOOLEAN","widget":{"name":"enable_repeat_penalty"},"link":null},{"localized_name":"repeat_penalty","name":"repeat_penalty","type":"FLOAT","widget":{"name":"repeat_penalty"},"link":null},{"localized_name":"enable_temperature","name":"enable_temperature","type":"BOOLEAN","widget":{"name":"enable_temperature"},"link":null},{"localized_name":"temperature","name":"temperature","type":"FLOAT","widget":{"name":"temperature"},"link":null},{"localized_name":"enable_seed","name":"enable_seed","type":"BOOLEAN","widget":{"name":"enable_seed"},"link":null},{"localized_name":"seed","name":"seed","type":"INT","widget":{"name":"seed"},"link":null},{"localized_name":"enable_stop","name":"enable_stop","type":"BOOLEAN","widget":{"name":"enable_stop"},"link":null},{"localized_name":"stop","name":"stop","type":"STRING","widget":{"name":"stop"},"link":null},{"localized_name":"enable_tfs_z","name":"enable_tfs_z","type":"BOOLEAN","widget":{"name":"enable_tfs_z"},"link":null},{"localized_name":"tfs_z","name":"tfs_z","type":"FLOAT","widget":{"name":"tfs_z"},"link":null},{"localized_name":"enable_num_predict","name":"enable_num_predict","type":"BOOLEAN","widget":{"name":"enable_num_predict"},"link":null},{"localized_name":"num_predict","name":"num_predict","type":"INT","widget":{"name":"num_predict"},"link":null},{"localized_name":"enable_top_k","name":"enable_top_k","type":"BOOLEAN","widget":{"name":"enable_top_k"},"link":null},{"localized_name":"top_k","name":"top_k","type":"INT","widget":{"name":"top_k"},"link":null},{"localized_name":"enable_top_p","name":"enable_top_p","type":"BOOLEAN","widget":{"name":"enable_top_p"},"link":null},{"localized_name":"top_p","name":"top_p","type":"FLOAT","widget":{"name":"top_p"},"link":null},{"localized_name":"enable_min_p","name":"enable_min_p","type":"BOOLEAN","widget":{"name":"enable_min_p"},"link":null},{"localized_name":"min_p","name":"min_p","type":"FLOAT","widget":{"name":"min_p"},"link":null},{"localized_name":"debug","name":"debug","type":"BOOLEAN","widget":{"name":"debug"},"link":null}],"outputs":[{"localized_name":"options","name":"options","type":"OLLAMA_OPTIONS","links":[39]}],"properties":{"cnr_id":"comfyui-ollama","ver":"6db7560576e5a59488708e6be13e07b5aba2432a","Node name for S&R":"OllamaOptionsV2"},"widgets_values":[false,0,false,0.1,false,5,false,2048,false,64,true,1.1,false,0.8,true,1405107337,"randomize",false,"",false,1,false,-1,false,40,false,0.9,false,0,false],"widgets_values_named":{"enable_mirostat":false,"mirostat":0,"enable_mirostat_eta":false,"mirostat_eta":0.1,"enable_mirostat_tau":false,"mirostat_tau":5,"enable_num_ctx":false,"num_ctx":2048,"enable_repeat_last_n":false,"repeat_last_n":64,"enable_repeat_penalty":true,"repeat_penalty":1.1,"enable_temperature":false,"temperature":0.8,"enable_seed":true,"seed":1405107337,"control_after_generate":"randomize","enable_stop":false,"stop":"","enable_tfs_z":false,"tfs_z":1,"enable_num_predict":false,"num_predict":-1,"enable_top_k":false,"top_k":40,"enable_top_p":false,"top_p":0.9,"enable_min_p":false,"min_p":0,"debug":false}},{"id":24,"type":"OllamaGenerateV2","pos":[3340,6460],"size":[400,300],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"connectivity","name":"connectivity","shape":7,"type":"OLLAMA_CONNECTIVITY","link":38},{"localized_name":"options","name":"options","shape":7,"type":"OLLAMA_OPTIONS","link":39},{"localized_name":"images","name":"images","shape":7,"type":"IMAGE","link":null},{"localized_name":"context","name":"context","shape":7,"type":"OLLAMA_CONTEXT","link":null},{"localized_name":"meta","name":"meta","shape":7,"type":"OLLAMA_META","link":null},{"localized_name":"system","name":"system","type":"STRING","widget":{"name":"system"},"link":41},{"localized_name":"prompt","name":"prompt","type":"STRING","widget":{"name":"prompt"},"link":42},{"localized_name":"think","name":"think","type":"BOOLEAN","widget":{"name":"think"},"link":null},{"localized_name":"keep_context","name":"keep_context","type":"BOOLEAN","widget":{"name":"keep_context"},"link":null},{"localized_name":"format","name":"format","type":"COMBO","widget":{"name":"format"},"link":null}],"outputs":[{"localized_name":"result","name":"result","type":"STRING","links":[43]},{"localized_name":"thinking","name":"thinking","type":"STRING","links":null},{"localized_name":"context","name":"context","type":"OLLAMA_CONTEXT","links":null},{"localized_name":"meta","name":"meta","type":"OLLAMA_META","links":null}],"properties":{"cnr_id":"comfyui-ollama","ver":"6db7560576e5a59488708e6be13e07b5aba2432a","Node name for S&R":"OllamaGenerateV2"},"widgets_values":["","",false,false,"text"],"widgets_values_named":{"system":"","prompt":"","think":false,"keep_context":false,"format":"text"}},{"id":25,"type":"OllamaConnectivityV2","pos":[3020,6460],"size":[270,160],"flags":{},"order":0,"mode":0,"inputs":[{"localized_name":"url","name":"url","type":"STRING","widget":{"name":"url"},"link":null},{"localized_name":"model","name":"model","type":"COMBO","widget":{"name":"model"},"link":53},{"localized_name":"keep_alive","name":"keep_alive","type":"INT","widget":{"name":"keep_alive"},"link":null},{"localized_name":"keep_alive_unit","name":"keep_alive_unit","type":"COMBO","widget":{"name":"keep_alive_unit"},"link":null}],"outputs":[{"localized_name":"connection","name":"connection","type":"OLLAMA_CONNECTIVITY","links":[38]}],"properties":{"cnr_id":"comfyui-ollama","ver":"6db7560576e5a59488708e6be13e07b5aba2432a","Node name for S&R":"OllamaConnectivityV2"},"widgets_values":["http://127.0.0.1:11434","vaultbox/qwen3.5-uncensored:9b",5,"minutes",null],"widgets_values_named":{"url":"http://127.0.0.1:11434","model":"vaultbox/qwen3.5-uncensored:9b","keep_alive":5,"keep_alive_unit":"minutes","🔄 Reconnect":null}}],"groups":[],"links":[{"id":38,"origin_id":25,"origin_slot":0,"target_id":24,"target_slot":0,"type":"OLLAMA_CONNECTIVITY"},{"id":39,"origin_id":26,"origin_slot":0,"target_id":24,"target_slot":1,"type":"OLLAMA_OPTIONS"},{"id":41,"origin_id":-10,"origin_slot":0,"target_id":24,"target_slot":5,"type":"STRING"},{"id":42,"origin_id":-10,"origin_slot":1,"target_id":24,"target_slot":6,"type":"STRING"},{"id":43,"origin_id":24,"origin_slot":0,"target_id":-20,"target_slot":0,"type":"STRING"},{"id":53,"origin_id":-10,"origin_slot":2,"target_id":25,"target_slot":1,"type":"COMBO"}],"extra":{}},{"id":"d82a9cc9-ee6f-47c5-9e22-6fcdc0e7146a","version":1,"state":{"lastGroupId":0,"lastNodeId":32,"lastLinkId":89,"lastRerouteId":0},"revision":0,"config":{},"name":"MiniMax-H3 Reference To Image Sampler","inputNode":{"id":-10,"bounding":[1830,5920,128,308]},"outputNode":{"id":-20,"bounding":[3570,6051,128,68]},"inputs":[{"id":"4b491ca5-62e6-4cad-ba67-390f5a069e90","name":"model","type":"MODEL","linkIds":[47,48],"localized_name":"model","pos":[1934,5944]},{"id":"ff411466-ad31-4edb-a078-461bd72b8b97","name":"vae","type":"VAE","linkIds":[67],"label":"video_vae","pos":[1934,5964]},{"id":"475b7dae-729f-4aaf-a7cf-3e9bdfbd5e9d","name":"audio_vae","type":"VAE","linkIds":[68],"pos":[1934,5984]},{"id":"02e734cc-77ba-4c0c-aa8f-337e96831146","name":"clip","type":"CLIP","linkIds":[69],"pos":[1934,6004]},{"id":"8f556e30-3520-4677-85d6-5f062e1334c5","name":"ref_images.ref_image_0","type":"IMAGE","linkIds":[70],"label":"ref_image_0","pos":[1934,6024]},{"id":"1e10315e-1e4b-4780-a0d8-f13a5f10c5e1","name":"ref_images.ref_image_1","type":"IMAGE","linkIds":[71],"label":"ref_image_1","pos":[1934,6044]},{"id":"639bae1a-8cd2-41c9-bbdb-c3bc14edc873","name":"ref_images.ref_image_2","type":"IMAGE","linkIds":[72],"label":"ref_image_2","pos":[1934,6064]},{"id":"dac63311-7bee-4983-9100-ec03488db0b8","name":"ref_images.ref_image_3","type":"IMAGE","linkIds":[73],"label":"ref_image_3","pos":[1934,6084]},{"id":"1372cfe1-d71e-446a-9e23-949ce69b173f","name":"prompt","type":"STRING","linkIds":[74],"pos":[1934,6104]},{"id":"f3004773-aec8-4443-928e-31f21c31c60d","name":"width","type":"INT","linkIds":[79],"pos":[1934,6124]},{"id":"abfd492d-f5e7-41fa-9a77-160e3faaa794","name":"height","type":"INT","linkIds":[76],"pos":[1934,6144]},{"id":"28f690cb-1937-4b31-8890-6800715670de","name":"steps","type":"INT","linkIds":[77],"pos":[1934,6164]},{"id":"7732e1b4-ba82-4648-a403-8a6636640d15","name":"noise_seed","type":"INT","linkIds":[78],"pos":[1934,6184]}],"outputs":[{"id":"684346cc-747b-445e-8f89-3344bbcc5425","name":"IMAGE","type":"IMAGE","linkIds":[52],"localized_name":"IMAGE","pos":[3594,6075]}],"widgets":[],"nodes":[{"id":18,"type":"CFGGuider","pos":[2640,5990],"size":[270,100],"flags":{},"order":3,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":47},{"localized_name":"positive","name":"positive","type":"CONDITIONING","link":22},{"localized_name":"negative","name":"negative","type":"CONDITIONING","link":24},{"localized_name":"cfg","name":"cfg","type":"FLOAT","widget":{"name":"cfg"},"link":null}],"outputs":[{"localized_name":"GUIDER","name":"GUIDER","type":"GUIDER","links":[26]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"CFGGuider"},"widgets_values":[1],"widgets_values_named":{"cfg":1}},{"id":17,"type":"KSamplerSelect","pos":[2640,6130],"size":[270,60],"flags":{},"order":1,"mode":0,"inputs":[{"localized_name":"sampler_name","name":"sampler_name","type":"COMBO","widget":{"name":"sampler_name"},"link":null}],"outputs":[{"localized_name":"SAMPLER","name":"SAMPLER","type":"SAMPLER","links":[27]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"KSamplerSelect"},"widgets_values":["er_sde"],"widgets_values_named":{"sampler_name":"er_sde"}},{"id":21,"type":"SamplerCustomAdvanced","pos":[2990,5850],"size":[220,330],"flags":{},"order":6,"mode":0,"inputs":[{"localized_name":"noise","name":"noise","type":"NOISE","link":25},{"localized_name":"guider","name":"guider","type":"GUIDER","link":26},{"localized_name":"sampler","name":"sampler","type":"SAMPLER","link":27},{"localized_name":"sigmas","name":"sigmas","type":"SIGMAS","link":28},{"localized_name":"latent_image","name":"latent_image","type":"LATENT","link":29}],"outputs":[{"localized_name":"output","name":"output","type":"LATENT","links":[30]},{"localized_name":"denoised_output","name":"denoised_output","type":"LATENT","links":null}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"SamplerCustomAdvanced"}},{"id":19,"type":"ConditioningZeroOut","pos":[2170,6320],"size":[210,30],"flags":{},"order":4,"mode":0,"inputs":[{"localized_name":"conditioning","name":"conditioning","type":"CONDITIONING","link":23}],"outputs":[{"localized_name":"CONDITIONING","name":"CONDITIONING","type":"CONDITIONING","links":[24]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"ConditioningZeroOut"}},{"id":22,"type":"VAEDecode","pos":[3240,5860],"size":[140,50],"flags":{},"order":7,"mode":0,"inputs":[{"localized_name":"samples","name":"samples","type":"LATENT","link":30},{"localized_name":"vae","name":"vae","type":"VAE","link":31}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[51]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"VAEDecode"}},{"id":30,"type":"ImageFromBatch","pos":[3240,5960],"size":[270,90],"flags":{},"order":8,"mode":0,"inputs":[{"localized_name":"image","name":"image","type":"IMAGE","link":51},{"localized_name":"batch_index","name":"batch_index","type":"INT","widget":{"name":"batch_index"},"link":null},{"localized_name":"length","name":"length","type":"INT","widget":{"name":"length"},"link":null}],"outputs":[{"localized_name":"IMAGE","name":"IMAGE","type":"IMAGE","links":[52]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"ImageFromBatch"},"widgets_values":[0,1],"widgets_values_named":{"batch_index":0,"length":1}},{"id":20,"type":"BasicScheduler","pos":[2640,6230],"size":[270,110],"flags":{},"order":5,"mode":0,"inputs":[{"localized_name":"model","name":"model","type":"MODEL","link":48},{"localized_name":"scheduler","name":"scheduler","type":"COMBO","widget":{"name":"scheduler"},"link":null},{"localized_name":"steps","name":"steps","type":"INT","widget":{"name":"steps"},"link":77},{"localized_name":"denoise","name":"denoise","type":"FLOAT","widget":{"name":"denoise"},"link":null}],"outputs":[{"localized_name":"SIGMAS","name":"SIGMAS","type":"SIGMAS","links":[28]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"BasicScheduler"},"widgets_values":["simple",8,1],"widgets_values_named":{"scheduler":"simple","steps":8,"denoise":1}},{"id":16,"type":"RandomNoise","pos":[2640,5860],"size":[270,90],"flags":{},"order":0,"mode":0,"inputs":[{"localized_name":"noise_seed","name":"noise_seed","type":"INT","widget":{"name":"noise_seed"},"link":78}],"outputs":[{"localized_name":"NOISE","name":"NOISE","type":"NOISE","links":[25]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"RandomNoise"},"widgets_values":[47211349325732,"randomize"],"widgets_values_named":{"noise_seed":47211349325732,"control_after_generate":"randomize"}},{"id":10,"type":"MiniMaxH3ReferenceToVideo","pos":[2170,5860],"size":[400,384],"flags":{},"order":2,"mode":0,"inputs":[{"localized_name":"clip","name":"clip","type":"CLIP","link":69},{"localized_name":"vae","name":"vae","type":"VAE","link":67},{"localized_name":"audio_vae","name":"audio_vae","type":"VAE","link":68},{"label":"ref_image_0","localized_name":"ref_images.ref_image_0","name":"ref_images.ref_image_0","shape":7,"type":"IMAGE","link":70},{"label":"ref_image_1","localized_name":"ref_images.ref_image_1","name":"ref_images.ref_image_1","shape":7,"type":"IMAGE","link":71},{"label":"ref_image_2","localized_name":"ref_images.ref_image_2","name":"ref_images.ref_image_2","shape":7,"type":"IMAGE","link":72},{"label":"ref_image_3","localized_name":"ref_images.ref_image_3","name":"ref_images.ref_image_3","shape":7,"type":"IMAGE","link":73},{"label":"ref_image_4","localized_name":"ref_images.ref_image_4","name":"ref_images.ref_image_4","shape":7,"type":"IMAGE","link":null},{"label":"ref_video_0","localized_name":"ref_videos.ref_video_0","name":"ref_videos.ref_video_0","shape":7,"type":"IMAGE","link":null},{"label":"ref_video_audio_0","localized_name":"ref_video_audios.ref_video_audio_0","name":"ref_video_audios.ref_video_audio_0","shape":7,"type":"AUDIO","link":null},{"label":"ref_audio_0","localized_name":"ref_audios.ref_audio_0","name":"ref_audios.ref_audio_0","shape":7,"type":"AUDIO","link":null},{"localized_name":"prompt","name":"prompt","type":"STRING","widget":{"name":"prompt"},"link":74},{"localized_name":"width","name":"width","type":"INT","widget":{"name":"width"},"link":79},{"localized_name":"height","name":"height","type":"INT","widget":{"name":"height"},"link":76},{"localized_name":"length","name":"length","type":"INT","widget":{"name":"length"},"link":null},{"localized_name":"ref_image_size","name":"ref_image_size","type":"COMBO","widget":{"name":"ref_image_size"},"link":null}],"outputs":[{"localized_name":"positive","name":"positive","type":"CONDITIONING","links":[22,23]},{"localized_name":"LATENT","name":"LATENT","type":"LATENT","links":[29]}],"properties":{"cnr_id":"comfy-core","ver":"0.33.0","Node name for S&R":"MiniMaxH3ReferenceToVideo"},"widgets_values":["",1344,768,5,"match"],"widgets_values_named":{"prompt":"","width":1344,"height":768,"length":5,"ref_image_size":"match"}}],"groups":[],"links":[{"id":22,"origin_id":10,"origin_slot":0,"target_id":18,"target_slot":1,"type":"CONDITIONING"},{"id":24,"origin_id":19,"origin_slot":0,"target_id":18,"target_slot":2,"type":"CONDITIONING"},{"id":25,"origin_id":16,"origin_slot":0,"target_id":21,"target_slot":0,"type":"NOISE"},{"id":26,"origin_id":18,"origin_slot":0,"target_id":21,"target_slot":1,"type":"GUIDER"},{"id":27,"origin_id":17,"origin_slot":0,"target_id":21,"target_slot":2,"type":"SAMPLER"},{"id":28,"origin_id":20,"origin_slot":0,"target_id":21,"target_slot":3,"type":"SIGMAS"},{"id":29,"origin_id":10,"origin_slot":1,"target_id":21,"target_slot":4,"type":"LATENT"},{"id":23,"origin_id":10,"origin_slot":0,"target_id":19,"target_slot":0,"type":"CONDITIONING"},{"id":30,"origin_id":21,"origin_slot":0,"target_id":22,"target_slot":0,"type":"LATENT"},{"id":51,"origin_id":22,"origin_slot":0,"target_id":30,"target_slot":0,"type":"IMAGE"},{"id":31,"origin_id":-10,"origin_slot":0,"target_id":22,"target_slot":1,"type":"VAE"},{"id":47,"origin_id":-10,"origin_slot":0,"target_id":18,"target_slot":0,"type":"MODEL"},{"id":48,"origin_id":-10,"origin_slot":0,"target_id":20,"target_slot":0,"type":"MODEL"},{"id":52,"origin_id":30,"origin_slot":0,"target_id":-20,"target_slot":0,"type":"IMAGE"},{"id":67,"origin_id":-10,"origin_slot":1,"target_id":10,"target_slot":1,"type":"VAE"},{"id":68,"origin_id":-10,"origin_slot":2,"target_id":10,"target_slot":2,"type":"VAE"},{"id":69,"origin_id":-10,"origin_slot":3,"target_id":10,"target_slot":0,"type":"CLIP"},{"id":70,"origin_id":-10,"origin_slot":4,"target_id":10,"target_slot":3,"type":"IMAGE"},{"id":71,"origin_id":-10,"origin_slot":5,"target_id":10,"target_slot":4,"type":"IMAGE"},{"id":72,"origin_id":-10,"origin_slot":6,"target_id":10,"target_slot":5,"type":"IMAGE"},{"id":73,"origin_id":-10,"origin_slot":7,"target_id":10,"target_slot":6,"type":"IMAGE"},{"id":74,"origin_id":-10,"origin_slot":8,"target_id":10,"target_slot":11,"type":"STRING"},{"id":76,"origin_id":-10,"origin_slot":10,"target_id":10,"target_slot":13,"type":"INT"},{"id":77,"origin_id":-10,"origin_slot":11,"target_id":20,"target_slot":2,"type":"INT"},{"id":78,"origin_id":-10,"origin_slot":12,"target_id":16,"target_slot":0,"type":"INT"},{"id":79,"origin_id":-10,"origin_slot":9,"target_id":10,"target_slot":12,"type":"INT"}],"extra":{}}]},"config":{},"extra":{"ds":{"scale":0.6431033558016328,"offset":[-580.0411377391031,-5634.075563083232]}},"version":0.4}