Nodes Browser

ComfyDeploy: How comfyui-selector works in ComfyUI?

What is comfyui-selector?

Nodes:Selector. Quick and dirty parameter generator node for ComfyUI.

How to install it in ComfyDeploy?

Head over to the machine page

  1. Click on the "Create a new machine" button
  2. Select the Edit build steps
  3. Add a new step -> Custom Node
  4. Search for comfyui-selector and select it
  5. Close the build step dialig and then click on the "Save" button to rebuild the machine

comfyui-selector

簡單的離線節點,用於ComfyUI。[選擇器]預設值[資源]開關 Quick and dirty nodes for ComfyUI 這些節點將會在images/utilities下列出,或者你可以輸入“.”來在搜索/雙擊菜單中的網格上找到它。 The nodes will be listed under images/utils, or you can type . to find it in search/double-click menu on the grid.

[!Note] 注釋 git clone https://github.com/exdysa/comfyui-selector/ 或將.py檔案放在...內 or place the .py file inside custom_nodes.

"資源"節點 - 從python-goss/ltdrdata建立。如果輸入被静音/跳過/刪除,它使用'fallback'中的替代輸入。這裡還有一個額外的布林出口,你可以在邏輯混亂中使用它。請確保輸入、失敗和輸出類型相符... "Recourse" node - building from pythongoss/ltdrdata. If the input has been muted/bypassed/deleted, it uses the alternative input in 'fallback'. There is an additional boolean out that you can use for logic mayhem as well. You probably want to make sure input, fallback and output types match...

Screenshot 2024-06-04 020106

"資源"現在有6個fallback朋友,大部分都由一個模式識別節點稱為Check控制,並根據檢查點類型重新導向輸出。輸入1是sd,2是xl,3是sd3,而輸入4是refiner。這大大簡化了自動化任務。 "Recourse" now has 6 fallback friends, most of which are controlled by a model-identifying node called Check that redirects output based on checkpoint type. input 1 is sd, 2 is xl, 3 is sd3 and input 4 is refiner. This greatly simplifies automation tasks. recoursecheck forkmodel forkclip unite recoursepolar recourseimg

"選擇器"節點 - 生成用於其他節點的簡單參數 "Selector" node - Generates simple parameters to use in other nodes Screenshot_2024-05-26_16-12-36 Screenshot_2024-05-26_17-44-08 Screenshot_2024-05-26_16-13-26

Data output todo :

  • ~SD 3.0 resolutions~
  • ~sampler options~
  • ~support for Pixart alpha/sigma, playground, kandinsky, etc...~
  • ~upscale dimensions and factors~
  • ~multiple steps~
  • ~multiple cfgs~
  • ~refiner steps~
  • ~multiple denoises~
  • ~...toast??~

Initial plans I will likely not attempt :

  • custom values
  • multiple seeds
  • nested menus for model types