{"id":55803,"date":"2026-08-20T09:42:26","date_gmt":"2026-08-20T01:42:26","guid":{"rendered":"https:\/\/www.1ai.net\/?p=55803"},"modified":"2026-08-11T11:28:45","modified_gmt":"2026-08-11T03:28:45","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%8cdepth-%e6%b7%b1%e6%8c%96-%e9%ab%98%e6%b8%85%e6%94%be%e5%a4%a7%ef%bc%88%e6%b8%85%e6%99%b0%e5%ba%a6%e5%86%8d","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/55803.html","title":{"rendered":"ComfyUI from installation entrance to mastery, Decth dig deep + high-resolution magnification (supplied)"},"content":{"rendered":"<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-55816\" title=\"6668087dj00tjl4i000p0d000p400dup\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2026\/08\/6668087dj00tjl4i000p0d000p400dup.jpg\" alt=\"6668087dj00tjl4i000p0d000p400dup\" width=\"904\" height=\"498\" \/><\/p>\n<p>Part 10 <a href=\"https:\/\/www.1ai.net\/en\/55801.html\/\">ComfyUI from installation to mastery, ControlNet, three scenes<\/a>You learned the OpenPose\/Dept\/Canny, and you know how screwth. But I left a pit unfilled:<strong>When will these three be used<\/strong><\/p>\n<p>I've stepped on my own real experience: I've thought about changing a real photo to Sabbunk style, using Canny. As a result, the wrinkles and edges of the new clothes were \u201cwelded to death\u201d and the style changes were so hard that they were like filtration filters. Then I found out that I had to do it\u00a0<strong>Decth (Depth)<\/strong>- It recognizes the back and forth spatial level, not the edge line. This is the one that tells Depth alone\u3002<\/p>\n<p><strong>I. Depth. What is it<\/strong><\/p>\n<p>One sentence:<strong>Depth controls the \"who's in the front, who's in the back, what's the size\" in the picture, not the edge line\u3002<\/strong>ControlNet's family is the easiest place for it to mix with Canny<\/p>\n<table>\n<tbody>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Comparison<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Canny<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Decth (Depth)<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">What<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">The contours, the wrinkles, all the hard edges<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Back and forth, body blocks in close proximity, shoulder neck stereos<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">When changing style<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">The edges are locked, and the new clothes won't move<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">It's only for space<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">suitability<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Colour on drafts, contours of protected products<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">People change their clothes, their style moves, they change their scenery<\/span><\/section>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><strong>Remember:<\/strong>Canny Pipeline, Decth Pipe Pipe Space. Want to change skin or bones, with Deepth\u3002<\/p>\n<p><strong>ii. Documentation \/ Environmental readiness (the misdirected model does not exist)<\/strong><\/p>\n<p>Decth is more dependent than OpenPose, with the highest probability of falling. Two things to prepare:<\/p>\n<p><strong>1 ControlNet model weight<\/strong>(Same as Canny\/OpenPose, here):<\/p>\n<p><a href=\"https:\/\/www.1ai.net\/en\/tag\/comfyui\" title=\"_Other Organiser\" target=\"_blank\" >ComfyUI<\/a>\/models\/controlnet\/<\/p>\n<p>SD1.5 Bottom model is used as control_v11f1p_sd15_depth_fp16; SDXL bottom model is used as diffuser_xl_depth_full\u3002<\/p>\n<p><strong>2 DevthAnythingV2 Pre-processing Weights<\/strong>(This one's the easiest to miss, node code installed, \u2260 weight added):<\/p>\n<p>ComfyUI\/kustom_nodes\/comfyui_controlnet_aux\/ckpts\/depth-anyth\/depth_anything_v2_vitl.pth<\/p>\n<p><strong>Trail experience:<\/strong>For the first time, I used Depth, only the ControlNet node, went down the ControlNet model, halfway to \"model not found.\" Half a day, it was found that the .pth pre-processing weight was missing -- node added to the canvas, and it ran short of material. The file name must be exactly the same\u3002<\/p>\n<p><strong>III. Hand-to-hand nodes (continuous)<\/strong><\/p>\n<p>And the basic chain of the 10th is a \"pre-treatment\" link: turn the reference into a Depth and feed it to ControlNet\u3002<\/p>\n<p>Load Image \u2192DepthAnythV2Processor Depth<br \/>\nZenium<br \/>\nControlNet Loader (select depth model) + Depth + Positive and Negative Plugin<br \/>\nZenium<br \/>\nKSampler<\/p>\n<p>1.\u00a0<strong>Add Node<\/strong>: Right-key search of DevthAnythingV2Preprocessor, ControlNet Loader, ControlNet Apply\u3002<br \/>\n2.\u00a0<strong>Connect<\/strong>: IMAGE of Load Image \u2192Preprocessor image; control_net of IMAGE \u2192Apply of Preprocessor; CONTROL_NET \u2192Apply of Loader _net; positive\/ negative alert of Apply\u3002<br \/>\n3.\u00a0<strong>Select Model<\/strong>: Loader selects control_v11f1p_sd15_depth_fp16 (SD1.5) or diffuser_xl_depth_full (SDXL)\u3002<br \/>\n4.\u00a0<strong>Scope<\/strong>: Apply has 0.65, Start_percent 0, End_percent 1\u3002<\/p>\n<p>One sentence:<strong>The reference diagram pre-processes the depth map through Decth, then goes to Apply Advanced with the ControlNet model and connects to the main chain\u3002<\/strong><\/p>\n<p><strong>Clean-up ring: 10 minutes, change the style of the photo with Deepth<\/strong><\/p>\n<p>Let's start with this one. After you've got an extra \"Spatial Structure Same But New Style\" map in your hand, you can confirm that Depth really works\u3002<\/p>\n<p><strong>Preconditions<\/strong>: You already have the 10th of the ControlNet Basic Chain. If you lose it, press this 5th node\u3002<br \/>\n<strong>Target<\/strong>: Take a photo of your real person, use the Decth control space, and change the style to \"water colour \/ Sabpunk \/ Whatever\" , the gestures, the necks, the contours of the original image are preserved, but the details of the clothes follow the style\u3002<\/p>\n<p>Step (sight, don't jump):<br \/>\n1.\u00a0<strong>Down Model<\/strong>: Depth ControlNet (SD1.5 with control_v11f1p_sd15_depth_fp16) dropped in Models\/controlnet\/\u3002<br \/>\n2.\u00a0<strong>Lower Preprocessing Weight<\/strong>: below depth_anyth_v2_vitl.pth, unnamed, comfyui_controlnet_aux\/ckpts\/depth-anything\/\u3002<br \/>\n3.\u00a0<strong>Add Node<\/strong>Photo by DepthAnythingV2Processor + ControlNet Loader + ControlNet Apply (Advanced)\u3002<br \/>\n4.\u00a0<strong>Hey, reference<\/strong>Load Image selects your photo; Loader chooses depth model; Preprocessor sets 768 (balance)\u3002<br \/>\n5.\u00a0<strong>Set parameters (most critical)<\/strong>Apply enter 0.65, start_percent0, end_percent1\u3002<br \/>\n6.\u00a0<strong>Write hints<\/strong>: positive-written \"for styles\" (e. g. watercolor painting) + original map master description; fixed seeds (e. g. 12345)\u3002<br \/>\n7.\u00a0<strong>Output<\/strong>: Point Queue Prompt\u3002<\/p>\n<p><strong>Checkpoint<\/strong>1 Generates a pattern\/ contour that is consistent with the original map, but the style changes = Depth; 2 returns \u201cmode not found\u201d = model or .pth path error (returns to section II); 3 produces nothing to do with the original = Apply is not connected to the title chain (returns to section III, step 2)\u3002<\/p>\n<p>job (selection): move stength from 0.65 to 0.4 and run one more -- you'll find the attitude has become loose, the original image has weakened; to 0.85, it's almost original. that's how the knob feels\u3002<\/p>\n<p><strong>core knob: stength and pre-treatment resolution<\/strong><\/p>\n<p><strong>strength (control intensity)<\/strong>\u2013 Depth's most understandable spin button, multi-source measurements:<\/p>\n<table>\n<tbody>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Area<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">effect<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">suitability<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">0.35 ~ 0.5<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Weak control. Just a general gesture<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">You want to be free, just refer to the drawings<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">0.55 ~ 0.7 (typical 0.65)<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Balance. Space's not rigid<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">People move like they change\/style<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">0.7 ~ 0.85<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Strong control, high-coherence redraw<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">To be precise about the original map space<\/span><\/section>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><strong>preprocessing resolution<\/strong>- Impact details and visibility:<br \/>\n512: Use when display is tight, speed is fast but depth is rough<br \/>\n768: Balancing, mostly adequate<br \/>\n1024: High detail, but visible, 4060<\/p>\n<p><strong>V. Common pits (all at once)<\/strong><\/p>\n<p>1.\u00a0<strong>...pth pre-processing weights are omitted<\/strong>\u2014 Nodal energy plus, a run-off model does not exist (first pit, back to section II)\u3002<br \/>\n2.\u00a0<strong>Synchronising folder<\/strong>_depth_anything_v2_vitl.pth cannot be mislettered and the wrong folder can not be found\u3002<br \/>\n3.\u00a0<strong>settings 0<\/strong>- It's like being uncontrollable, creating a picture and a reference\u3002<br \/>\n4.\u00a0<strong>Use that, Canny. Change<\/strong>- Welded on the sideline and the style had to be hard; change the skin, please\u3002<br \/>\n5.\u00a0<strong>More than one ControlNet without visible<\/strong>\u2014Depth + OpenPose at the same time, 4060 recommends a resolution below 768\u3002<\/p>\n<p><strong>VI. QUIREMENTS OF PARLIAMENTS<\/strong><\/p>\n<table>\n<tbody>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">take<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Bottom Model<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">strength<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Preprocessing Resolution<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">People changing clothes\/style migration<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">SDXL<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">0.65<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">768<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">High-consistent redraw<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">SD1.5<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">0.8<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">1024<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Weak control through attitude<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Any<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">0.45<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">512<\/span><\/section>\n<\/td>\n<\/tr>\n<tr>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Deep + OpenPose Double Control<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">SDXL<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">Each 0.6<\/span><\/section>\n<\/td>\n<td>\n<section data-nest-level=\"6\"><span data-nest-level=\"7\">512<\/span><\/section>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><strong>VII. 4060 \/ 8G SPECIALIZED &amp; I STEPPED ON PITS<\/strong><\/p>\n<p>Skills: Depth pre-treatment itself takes a bit of visibility, up 4060\u00a0<strong>resolute with 768<\/strong>; if the OpenPose is folded at the same time, the resolution of both ControlNet is down to 512, and the visible occupancy can be pressed. Model Selection\u00a0<strong>fp16 version<\/strong>could not close temporary folder: %s starts with \u2013medvram more steady\u3002<\/p>\n<p>Five pits say all at once: 1.pth (No. 1); 2 file name\/path error; 3 stength set 0 unresponsive; 4 take Canny to replace it and cause welding; 5 stacks of ControlNet without direct resolution OOM\u3002<\/p>","protected":false},"excerpt":{"rendered":"<p>Book 10 ComfyUI, from installation to mastery, ControlNet, three scenes (in your direction) you learned the ControlNet Three Brothers, and you know howstrength is screwed. But then I left a pit unfilled: which are the three? I've stepped on my own real experience: I've thought about changing a real photo to Sabbunk style, using Canny. As a result, the wrinkles and edges of the new clothes were \u201cwelded to death\u201d and the style changes were so hard that they were like filtration filters. And then I found out that I was going to change the skin<\/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-55803","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\/55803","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=55803"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/55803\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=55803"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=55803"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=55803"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=55803"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}