ComfyUI - Live Stream! Let's make some amazing art with Stable Diffusion!

Тәжірибелік нұсқаулар және стиль

Let's make something together! Come along and let's talk ComfyUI as well as other things related to generative AI. It's always fun! Note that the playback of this video will be available to channel patrons after the show is over. Catch it live! #comfy #stablediffusion
Become a member to get exclusive access to perks!
/ @sedetweiler
Gigabyte 17X Laptop is doing the inference today! Grab one here:
amzn.to/3thtfpR
Link: www.amazon.com/Foldable-Compr... (Ortur Master 3 and rotary)
For painting, I prefer Rebelle to Photoshop. Grab that here: tinyurl.com/2bk6b4oq
Join us on Discord! / discord (we have quiet rooms for MidJourney work)
Print and sell your own artwork! www.printful.com/print-on-dem...
Music by share.epidemicsound.com/vwe5mn
Backup your images! All you need is here: a.co/1uwXqv2
Enjoy this video? Consider buying me a coffee! ko-fi.com/sedetweiler

Пікірлер: 26

  • @Fiveash-Art
    @Fiveash-Art2 ай бұрын

    I like the soothing NPR style jazz stuff in the background 😁👍🏻

  • @gimperita3035
    @gimperita30353 ай бұрын

    This was a very fun streaming, laughed out load with the steamroll witch. Let's put that node to work! Have a great weekend, everyone!

  • @GamingDaveUK

    @GamingDaveUK

    3 ай бұрын

    Just got to that part and startled a customer walking past as I laughed

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    Thanks, you too! I sure had a laugh at that! Have a great weekend!

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    Perfect timing! lol!

  • @korinlifshits8780
    @korinlifshits87803 ай бұрын

    hi . great content. where I can get the workflow jsons from the videos? thank you

  • @AlexKellyArtUK
    @AlexKellyArtUK3 ай бұрын

    As a traditional artist, I'm still amazed by some of the images that vanilla 1.5 AI could produce-not photorealistic, but very "painterly" and atmospheric. Juggernaut is quite reliable yet somewhat predictable in its output style, or maybe I just suck at prompting! The 1.5 models have a unique quality that goes beyond the pursuit of hyper-realism, which seems to be the focus of many models post 1.5. It's difficult to define, but SDXL appears more geared towards photorealism rather than painterly abstraction.

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    I think you and I are on the same page. I still really enjoy many of my 1.5 images and can't wait to show you what we can do with SD3!

  • @rawkeh
    @rawkeh3 ай бұрын

    Missed the stream, dang, cause I wanted to ask: how can you force a certain facial expression? Like I want to build a set of phonemes, so I want things like "half open mouth" or "saying the sound 'ee'" sort of thing

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    That is a hard one with SDXL. There are some pose controlnets for mouth that might help.

  • @murphylanga
    @murphylanga3 ай бұрын

    hi, check the JSP nodes and the prompt helper if you are too comfortable with Clip gG and Clip L. Copy to both if empty ;-) Greetings Murphy

  • @Gabriecielo
    @Gabriecielo3 ай бұрын

    Talking about messy nodes menu, before someone think about a nodes management, is there a way to at least sort by name? It's extremely hard to find a new installed node. Search function sometime doesn't work or listing too many too.

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    No, it is a total train wreck. Nothing at all in the way of management, not even by order.

  • @Gabriecielo

    @Gabriecielo

    3 ай бұрын

    And just like your "O" folder, I have some nodes under "test", just illogical at all. LOL@@sedetweiler

  • @Ymirheim
    @YmirheimАй бұрын

    The safety balance cuts two ways due to how hard it is to define something so gradient. I as a professional canceled my Adobe subscription of 6 years because, despite not trying to make anything nsfw the safety training of Photoshops AI tools keeps interfering with my job when I retouch photos. I am not the only one, not sure that is the product they would be proud of either. I understand the PR questions but sometimes the position of the censorship advocates also seem to be that the only reason anyone would ever want parts of a breast in an image is to make revenge porn. Yet some of the worlds greatest works of art of all time would not be acceptable to our image generators. This is not a problem for as long as people can train their own models and specialize. But the pressure from both companies and governments is clearly on reigning this in further.

  • @GamingDaveUK
    @GamingDaveUK3 ай бұрын

    I am watching this late (as per usual due to work) I just got up to the zero shot comment. I was joking the other day as the AI was having trouble making images of Knights in yellow armour, riding squirrels.... then it had difficulty with squirrels of burdan, was going for an old farmer walking down the road with a giant squirrel ladened with sacks of spuds and wheat... it wouldnt do it for ages... finally I got it, I was happy... then we tried to have squirrels carrying people around a medieval city (think tiny furry porters holding people aloft as they carry them about..... this the AI drew a line at! it said no more.... I got squirrels holding squirrels, I got people holding squirrels but it refused to do squirrels holding people! Its almost like the training data used to make these models, has no medieval images of squirrels carrying people. Long post made short, my suggestion for a prompt: Larger than normal squirrels carying people around a medieval city or town, paws shaking under the burdan as they hold the humans aloft.

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    This is a great example! Yes, there are some topics that just don't work. I will have to try this prompt!

  • @GamingDaveUK

    @GamingDaveUK

    3 ай бұрын

    @@sedetweilerI will be keen to see if you manage to get it to do it lol.

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    @@GamingDaveUK I will have to give it a good solid attempt.

  • @TKs3DPrints
    @TKs3DPrints3 ай бұрын

    think you might be better using the prompt like flame texture or pizza texture or texture the dress with pizza no idea if it will work your the wizard i am just a broom :)

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    Yeah, but it was pretty fun to watch it fail! :-)

  • @thomasgoodwin2648
    @thomasgoodwin26483 ай бұрын

    "Jinn, also romanized as djinn or anglicized as genies, are invisible creatures in early religion in pre-Islamic Arabia and later in Islamic culture and beliefs. Like humans, they are accountable for their deeds and can be either believers or disbelievers, depending on whether they accept God's guidance."-Wikipedia 🖖😁👍

  • @sedetweiler

    @sedetweiler

    3 ай бұрын

    Yes, that is more in-line with what I was expecting, not a lizard for sure. Thanks for checking into that though, as I am sure another model might have a better and more accurate image.

Келесі