{"id":55805,"date":"2026-08-21T09:08:40","date_gmt":"2026-08-21T01:08:40","guid":{"rendered":"https:\/\/www.1ai.net\/?p=55805"},"modified":"2026-08-11T11:29:01","modified_gmt":"2026-08-11T03:29:01","slug":"comfyui-%e4%bb%8e%e5%ae%89%e8%a3%85%e5%85%a5%e9%97%a8%e5%88%b0%e7%b2%be%e9%80%9a%ef%bc%8cipadapter%ef%bc%8c%e4%b8%80%e5%bc%a0%e5%8f%82%e8%80%83%e5%9b%be%e9%94%81%e8%84%b8%e3%80%81%e6%90%ac%e9%a3%8e","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/55805.html","title":{"rendered":"ComfyUI from installation entrance to mastery, IPadapter, a reference map for face locks, moving styles, mattresses"},"content":{"rendered":"

\"d53bba6aj00tjl5la00phd000p300dup\"<\/p>\n

Title 11 ComfyUI from installation entrance to mastery, Decth dig deep + high-resolution magnification (supplementation)<\/a>You control space with Decth, and you know the ControlNet tube position. But there's one more troublesome thing: a set of different scenes of the same role, the first round face, the second square face, the third straight hair change -- I've been staring at three pictures for half a day, not three separated twins. Then I realized:ControlNet Tube and Posture, IPAdapter Stewardship and Style<\/strong>I don't know. If you want the same face, you have to rely on IPadapter, especially the FaceID series. That's what I'm talking about\u3002<\/p>\n

I. What the hell is IPAdapter<\/strong><\/p>\n

One sentence:Make a reference \"encoding\" a condition, place it in the bottom model, and leave it to the character of the reference map\u3002<\/strong><\/p>\n

\u00b7\u00a0Style Class<\/strong>(gypsy, poster)\u3002
\n\u00b7\u00a0Lockface<\/strong>(many faces each time): Yes\u00a0FaceID Series<\/strong>It's more specific to extracting human face features than the base version\u3002
\n\u00b7\u00a0Double<\/strong>One pipe style, one pipe face, two IPadapter strings, no fights\u3002<\/p>\n

IT'S JUST AS IT'S NOW THE BIGGEST PAIN OF THE AI COMIC -- \"SHARE A FACE\" ON A HOT SEARCH. LOCAL ComfyUI<\/a> + IPadapter, full, permanent free and unlimited, is the solution\u3002<\/p>\n

Remember:<\/strong>ControlNet position, IPadapter looks like you're the director\u3002<\/p>\n

ii. Documentation \/ Environmental readiness (dry in the wrong place)<\/strong><\/p>\n

Ipadapter doesn't run after the node, the model has to be put in the right place. The first thing I did was the model threw the wrong folder, the node was red, half an hour\u3002<\/p>\n\n\n\n\n\n\n\n
\n
Model<\/span><\/section>\n<\/td>\n
\n
Where<\/span><\/section>\n<\/td>\n
\n
clarification<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
IPAdapter Master Model<\/span><\/section>\n<\/td>\n
\n
i'm sorry<\/span><\/section>\n<\/td>\n
\n
It's all here<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
CLIP Vision Encoder<\/span><\/section>\n<\/td>\n
\n
models\/clip_vision\/<\/span><\/section>\n<\/td>\n
\n
Standard CLIP-ViT-H-14-laion2B; vit-G with bigG<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
inightface buffalo_l<\/span><\/section>\n<\/td>\n
\n
models\/insightface\/models\/buffalo_l\/<\/span><\/section>\n<\/td>\n
\n
The face must be locked<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
FaceID LoRA<\/span><\/section>\n<\/td>\n
\n
models\/loras\/<\/span><\/section>\n<\/td>\n
\n
Most FaceID to match, Unified Loader FACEID presets automatic calls<\/span><\/section>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Trail experience:<\/strong>I had only buffalo_l left Lora, and my face was still locked -- FaceID basically welded people's faces with Lora, and it wasn't enough with buffalo_l\u3002<\/p>\n

iii. Hand cuffing<\/strong><\/p>\n

Compared to the base chain of 05, the IPdapter is an additional \"encoded reference map\" link: the reference map goes past the IPdapter node and goes into KSampler with the bottom model\u3002<\/p>\n

Load Checkpoint
\n\u2022 CLIP \u2014 CLIP Text Encode (\u00b1) \u2014 KSampler.cond
\n\u2022 VAE \u2192 VAE Decode
\nRead Checkpoint
\nIpadapter Unified Loader
\nLoad Image (reference)
\nAdvanced. MODEL
\nKSampler<\/p>\n

1.\u00a0Add Node<\/strong>: Right-click search Load Image, IPadapter United Loader, IPadapter Advanced\u3002
\n2.\u00a0Connect<\/strong>:<\/p>\n

References<\/p>\n

Modular MODEL first to Ipadapter United Loader, Loader Output to Advanced.Model<\/p>\n

Advanced. MODEL\u00a0KSampler.model<\/strong>(mode not cond<\/p>\n

Unified Loader automatically pairs the Ipadapter model with the corresponding CLIP Vision\u3002
\n3.\u00a0Preset<\/strong>: Unified Loader chooses FACEID\u3002<\/p>\n

One sentence:Reference diagrams feed Advanced.image; Endimage Unified Loader to Advanced.Model; Endel out of Advanced collects Model from KSampler\u3002<\/strong>Small tip: IPadapter sets the reference figure into a square to extract the feature, using as much as possible a square or a clear face\u3002<\/p>\n

Weld the star's face in 10 minutes<\/strong><\/p>\n

Let's start with this one. After you've got a bunch of \"same face, but scenes\/styles change.\"\u3002<\/p>\n

Preconditions<\/strong>: You've run Load Checkpoint \u2192 KSampler \u2192 VAE Decode \u2192 Save. This article adds nodes to it\u3002
\nTarget<\/strong>: Use a leading character reference to share the same face, the same style with all the following characters\u3002<\/p>\n

Step (sight, don't jump):
\n1.\u00a0Down model right in position<\/strong>:<\/p>\n

ipadapter models\/ipadapter\/<\/p>\n

clip_vision \u2192 models\/clip_vision\/<\/p>\n

buffalo_l models\/insightface\/models\/buffalo_l\/<\/p>\n

FaceID LoRA \u2192 models\/loras\/\u3002
\n2.\u00a0Plugin<\/strong>:Manager search for \"IPAdapter Plus\" one key, restart\u3002
\n3.\u00a0Add Node<\/strong>: Load Image (reference diagram) + IPadapter United Loader + IPadapter Advanced\u3002
\n4.\u00a0Connect<\/strong>: MODEL for Load Checkpoint first to IPadapter Unified Loader, Loader output to Advanced.Model; Load Image to Advanced.image; Advanced.MODEL \u2192 KSampler.Model\u3002
\n5.\u00a0Participation<\/strong>: Unified Loader preset FACEID (lockface); Weight style 0.6~0.8, FaceID 0.9~1.0; start 0\/ end 1. Queue Prompt\u3002<\/p>\n

Checkpoint<\/strong><\/p>\n

1 Face\/style with reference diagrams, with no change in hint face = success<\/p>\n

2 Pit 1 \"face unlocked\":light is too low, or Advanced is wrong (must pick up KSampler's model), or FaceID LoRA does not play models\/loras\/<\/p>\n

3 Pit 2 \"Facedown\/Backback\": FaceID does not have a buffalo_l, go back and replace section 2<\/p>\n

4 pit 3 \"significant bomb\": 4060 folds \u2013medvram with resolution down to 512\u3002<\/p>\n

job (selection): same reference, view 0.6, one each, and 0.9, see how much different face welding is\u3002<\/p>\n

IV. QUIREMENTS OF PARLIAMENTS<\/strong><\/p>\n\n\n\n\n\n\n\n
\n
The effect you want<\/span><\/section>\n<\/td>\n
\n
Which model<\/span><\/section>\n<\/td>\n
\n
right<\/span><\/section>\n<\/td>\n
\n
Remark<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
style migration<\/span><\/section>\n<\/td>\n
\n
this post is part of our special coverage syria protests 2011<\/span><\/section>\n<\/td>\n
\n
0.6~0.8<\/span><\/section>\n<\/td>\n
\n
Follow style reference<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
Lockface<\/span><\/section>\n<\/td>\n
\n
faced-plus + Insightface + FaceID LoRA<\/span><\/section>\n<\/td>\n
\n
0.9~1.0<\/span><\/section>\n<\/td>\n
\n
Foreface reference<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
Blanket Separation<\/span><\/section>\n<\/td>\n
\n
Two Advanced<\/span><\/section>\n<\/td>\n
\n
Style 0.7 \/ Face 0.95<\/span><\/section>\n<\/td>\n
\n
I don't care<\/span><\/section>\n<\/td>\n<\/tr>\n
\n
4060 Conservative<\/span><\/section>\n<\/td>\n
\n
fp16 above<\/span><\/section>\n<\/td>\n
\n
0.7 cases<\/span><\/section>\n<\/td>\n
\n
staple -medvram<\/span><\/section>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Note: Model paths, node names are cross-checked by cubiq\/CommyUI_IPAdapter_plus Official Document + Community Curriculum 2 Sources (validated 2026-07-21); between weights are community general sweet areas, finely adjusted upwards and downs\u3002<\/p>\n

V, 4060 user-specific<\/strong><\/p>\n

:: Use all models\u00a0fp16<\/strong>don't touch fp32, double the balance\u3002
\n\u2022 CLIP Vision also eats visible (about 1-2G) not to have too many reference nodes at the same time\u3002
\n\u2022 FaceID's insightface reasoning is light, with a head in the bottom + crip_vision\u3002
\nResolution\u00a0512 x 768 Start<\/strong>;to magnify under 13\u3002
\n\u2022 Multi-IPadapter serial display, run single and double; start plus\u00a0\u2013medvram<\/strong>.<\/p>","protected":false},"excerpt":{"rendered":"

Title 11 ComfyUI, from installation to mastery, Depth dig deep + high-resolution magnification (supplied) you control space with Decth and know the ControlNet tube position. But there's one more troublesome thing: a set of different scenes of the same role, the first round face, the second square face, the third straight hair change -- I've been staring at three pictures for half a day, not three separated twins. Then it became clear that the ControlNet tube chart and pose, the IPadapter tube looks and styles. If you want the same face, you have to rely on IPadapter, especially the FaceID series. This one<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[149,144],"tags":[1989,6276,4749],"collection":[],"class_list":["post-55805","post","type-post","status-publish","format-standard","hentry","category-jiaocheng","category-baike","tag-comfyui"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/55805","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=55805"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/55805\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=55805"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=55805"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=55805"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=55805"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}