How to make AI QR Codes - NEW QR Control Net

Поділитися
Вставка
  • Опубліковано 9 лис 2024

КОМЕНТАРІ • 40

  • @jameelamuhammad4250
    @jameelamuhammad4250 Рік тому

    you made this right when i needed it

  • @deFl0
    @deFl0 Рік тому

    Beginner Questions. On your screen, are you running stable diffusion locally? Do you have a tutorial on how to set that up?

  • @TheCrescentArkProductions
    @TheCrescentArkProductions Рік тому

    want to know the best settings? Don't use the controlnet model at all. set it to none and use some better sd models like rev animated or realistic. and see the results. though you have to enable the controlnet and upload the qr. The images will turn out so awesome that you will not be able to recognize that there is qr in the image until you scan it. but for some reason you want the image to be a perfect blend of qr and the image. you can just control it by the sampling steps. (also a1111 has the qr tab in extensions you can just enable it. dont go to third party websites)

    • @Arddy
      @Arddy Рік тому +1

      Could you elaborate? The start of your comment said not to use ControlNet model "at all." And then later, you said "...though you have to enable the Controlnet and upload the QR." So what exactly do you mean? I.e what are your settings? Could you also provide some recommendations for SD Models for QR Art? I would greatly appreciate it.

  • @calmmarketing
    @calmmarketing Рік тому

    super helpful Russel, thank you! Would the img2img work without a prompt? So basically get a generation “inspired” by the image reference image only?

    • @RussellKlimas
      @RussellKlimas  Рік тому

      I just tried it and the result in my opinion is less than desirable.

    • @calmmarketing
      @calmmarketing Рік тому

      @@RussellKlimas yes same with my results. I was wondering it there was a technique to improve it.

    • @RussellKlimas
      @RussellKlimas  Рік тому

      @@calmmarketing why do you not want to use a prompt with img2img? You could always clip interrogate the image you are using and then use that to help.

    • @calmmarketing
      @calmmarketing Рік тому

      @@RussellKlimas Yes, I'm trying that. I guess it's the best way. THe problem is when the clip predicts the wrong subject, I get a very weird and different image.

  • @FunkTigerisbest
    @FunkTigerisbest Рік тому

    Which checkpoint you are working on?
    Also what about the Models you mentioned in the description where should i put them?

    • @RussellKlimas
      @RussellKlimas  Рік тому

      If running locally you put them in your models folder.

  • @StillBreathing-t3z
    @StillBreathing-t3z Рік тому

    When I open my automatic 1111 there is no extension option, any idea why?

  • @georgematsos4416
    @georgematsos4416 Рік тому

    Is there a way to install the models in thinkdiffusion ?

  • @FunkTigerisbest
    @FunkTigerisbest Рік тому

    can you help me fix this error OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 4.00 GiB total capacity; 3.39 GiB already allocated; 0 bytes free; 3.46 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
    Time taken: 0.45sTorch active/reserved: 3473/3544 MiB, Sys VRAM: 4096/4096 MiB (100.0%)

    • @RussellKlimas
      @RussellKlimas  Рік тому

      Sounds like you don't have enough vram. You'll either need to close programs that take up a lot or upgrade vram. Sometimes restarting your computer helps too.

  • @georgematsos4416
    @georgematsos4416 Рік тому

    Can we just put an image and a qr code without using a prompt ?

    • @RussellKlimas
      @RussellKlimas  Рік тому

      As much as I wished it worked that way it doesn't, not in my tests anyway.

    • @georgematsos4416
      @georgematsos4416 Рік тому

      @@RussellKlimas Thanks.

    • @welkey90
      @welkey90 Рік тому

      I also have a problem with it, we wanna create a qr code with our brand's image, so i dont know what i should put into prompt. I mean i just want the qr code generated into or onto the brand image i provide, is that even possible?

    • @georgematsos4416
      @georgematsos4416 Рік тому

      @@welkey90 Y that would be perfect. I don't think is possible (yet)
      But is AI.
      It will happen.

    • @welkey90
      @welkey90 Рік тому

      why does it call img2img if its gonna completely ignore the image i put

  • @kannakrew
    @kannakrew Рік тому

    I am getting results that have no QR code in them. Control net is up to date and I am only using the QR code control net

    • @kannakrew
      @kannakrew Рік тому

      its like this model does nothing

    • @RussellKlimas
      @RussellKlimas  Рік тому

      Make sure you are putting a qr code that you've already made in the controlnet. Try doing a restart of the program too. I've had a similar issue before.

    • @kannakrew
      @kannakrew Рік тому

      @@RussellKlimas yes I am using a premade QR code. I have tried with and without invert. If I use something like tile or brightness, I get a qrcode, but when I use that qr_code model, my result is unchanged even if I set control weight to 2. I can’t get it to fire off with txt or img

    • @RussellKlimas
      @RussellKlimas  Рік тому

      @@kannakrew Yeah I just did a restart and it worked for me. Since I ran into something similar.

    • @kannakrew
      @kannakrew Рік тому

      @@RussellKlimas here's the problem i ws having: My qr code was only 290x290, when i bumped it up to 500^2, it worked like a charm

  • @sharifmounir2758
    @sharifmounir2758 Рік тому

    lol why r u hiding SD version?

    • @RussellKlimas
      @RussellKlimas  Рік тому +1

      I'm hiding the model name because it's a personal mode of a friend and keep anonymity.

  • @cleocss
    @cleocss Рік тому

    Ok i feel old I don't understand anything 😅

  • @yalexis
    @yalexis Рік тому

    I followed everyone and I got this, RuntimeError: mat1 and mat2 shapes cannot be multiplied (154x768 and 1024x320)

    • @RussellKlimas
      @RussellKlimas  Рік тому +1

      If you go to 3:51 in the video you can see that I explain that it depends on the checkpoint you are using whether it is 1.5 or 2.0 and which to use.

  • @AltayFamily
    @AltayFamily Рік тому

    Is there a way to install the models in thinkdiffusion ?