ComfyUI: Intro to Control Net- Sketch to Render (Part III)

Поделиться
HTML-код
  • Опубликовано: 10 июл 2024
  • In this video I will go over how you can develop more advance workflows with comfy UI with control net to allow more structural control based on input images and they are use in popular workflows such as sketch to image AI generation.
    This follows on from a previous video on the basics of comfy so please take a look at that for more background. Control net is a core extension that I often use as part of my architectural workflows so is definitely worth getting acquainted with.
    00:00 Intro
    00:35 Control net Set-up
    01:34 Control net Model install
    03:30 Sketch to Image Workflow
    08:18 Using multiple control nets
    12:02 Comparisons in ComfyUI
    Install: • One-click installs for...
    Part I: • ComfyUI: Getting start...
    Part II: • ComfyUI: Intro to Mask...
    Control net models: huggingface.co/lllyasviel/Con...
    Instagram account: / urbandecoders
    #ai #architecture #generativedesign #parametric #stablediffusion #animation #comfui #image #design #generativedesign #painting
  • НаукаНаука

Комментарии • 15

  • @kaiforging
    @kaiforging Месяц назад +1

    very good ,thanks

  • @davidbell6101
    @davidbell6101 4 месяца назад +2

    nicely done.

  • @Arquiteto3d
    @Arquiteto3d 3 месяца назад +1

    Congratulations! Nice tutorial!!! When We are gonna have more tutorial talking about Multi Controlnet Stack?

    • @Urban_Decoders
      @Urban_Decoders  2 месяца назад +1

      Thanks! I will do a few more videos covering some other core nodes/ principles in comfyUI and then plan to create some intermediate/ advanced complete workflows which combine them all. Multi-controlnets will be part if of it for sure!

  • @abiodunshonibare832
    @abiodunshonibare832 4 месяца назад

    Nice tutorial, why did we have to copy the reference image and connect one to the vae encode and another to the preprocessor, it would mean for every new image we change we need to have 2 copies?

    • @Urban_Decoders
      @Urban_Decoders  3 месяца назад

      You are correct, you can just use the same reference image and connect that to a vae and preprocessor. Was just separating it out here to try and make the process clearer.

  • @andresz1606
    @andresz1606 3 месяца назад

    The creativity level is set by CFG, while denoise will determine how much of the original image can be redrawn. Using the Advanced ControlNet instead of the regular one seems pointless unless you are fine-tunning the advanced node's values. I didn't quite get the purpose of the Scribble Lines node, is it just to resize the source image? I believe that's done automatically depending on the output size anyway.

    • @Urban_Decoders
      @Urban_Decoders  3 месяца назад +1

      the advanced control net gives additional settings to adjust the control net, particularly on the strength of the effect. here is some good documentation on it: github.com/Kosinkadink/ComfyUI-Advanced-ControlNet. The scribble lines nodes and preview is there for you to see what the effect of each control net is. You can compare the various styles before running through the k-sampler. But you can omit this node and preview if you want.

  • @bimpf8570
    @bimpf8570 2 месяца назад +1

    Good viedeo i watched basically all of them but i have a problem with accassing my controlnetloader in comfyui can you maybe help me reugh discord or something

    • @Urban_Decoders
      @Urban_Decoders  2 месяца назад

      you need to install the control net nodes from the comfyui manager

    • @bimpf8570
      @bimpf8570 2 месяца назад

      @@Urban_Decoders i did but maybe i need to try again thanks

    • @bimpf8570
      @bimpf8570 2 месяца назад

      @@Urban_Decoders it still has problems to find the data. The contorlnet loader does not find my data but it is in the right folder.

  • @joep7162
    @joep7162 4 месяца назад

    Where is part 2 and 1?

    • @Urban_Decoders
      @Urban_Decoders  4 месяца назад

      You can find the playlist here: ruclips.net/p/PLBTaSym_yghIFjCRf92VfmAQ0uwKz9SEM