good tutorial man, appreciate the great work. was very informative
@Cicadeus9 күн бұрын
It should be noted that The Controlnet demo'ed here does not work on SDXL models, took hours to relize why it wasn't doing anything =w=
@Kinos14110 күн бұрын
Well, this is awesome
@Redgyart15 күн бұрын
Will this work on m1 macs?
@gsestream15 күн бұрын
new definition of sketch board video making.
@kickassets641417 күн бұрын
There isn't control in creativity. Drawing chicken shit with random results isn't practical or appealing to skilled artists. Being able to duplicate myself like a studio would b more practical.
@omg-the-best-crazy13 күн бұрын
Ai is mainly for common people to play around like children do with shiny new toys, feel as if they are artists, and then forget about it after a few days.
@PawFromTheBroons17 күн бұрын
Your vertical bar is called a Pipe
@faredit-cq2xl17 күн бұрын
Thank's ❤
@enriquericossanchis18 күн бұрын
Was the model trained on images stolen from artists? How much energy/pollution did this thing cost us? You guys don't care, right? How about artists losing jobs? Unimportant in comparison, but this tech being in Krita is so ironic that I had to mention it.
@dev_reimu17 күн бұрын
Finally, some common fucking sense in this comment section!
@UltimatePerfection17 күн бұрын
Well, boo, freaking, hoo. For the record, I am an artist and can actually draw and do it myself, but I am not afraid of a roomba with a brush replacing me. The AI will never be able to draw exactly like I do. And as far as "losing jobs" goes, well that's on you for choosing a profession that doesn't pay good to begin with and then pricing your commissions insanely ($200 for a simple 1080p, or less, picture? Gahaha, how about something more appropriate like $15?)
@enriquericossanchis17 күн бұрын
@@UltimatePerfection I wasn't thinking about my job, I do care about people less privileged than me, because empathy. But I'll make an exception for you. And I see that you don't give a shit about climate change either. Good luck.
@UltimatePerfection17 күн бұрын
@@enriquericossanchis I'm not the one doing polluting, why should I care and not e.g. coal powerplant owners or owners of private jets?
@dev_reimu17 күн бұрын
@@UltimatePerfection "Well it's your fault that you wanted to do art!" 🤓
@rameshdevasi672018 күн бұрын
krea is also good
@evilpigskin18 күн бұрын
Wow, great! Thank you!
@steluste18 күн бұрын
Damn. I must admit it's a cool technology.
@dev_reimu18 күн бұрын
most common symptom of a skill issue
@faredit-cq2xl17 күн бұрын
Cry about it
@dev_reimu17 күн бұрын
@@faredit-cq2xl No, you cry about it for having no skills or anything interesting about you.
@omg-the-best-crazy13 күн бұрын
@@faredit-cq2xllol everyone discards ai as trash, wake up
@faredit-cq2xl13 күн бұрын
@@omg-the-best-crazy hahaha so why are you here
@omg-the-best-crazy13 күн бұрын
@@faredit-cq2xl to laugh at you? duh
@MaxyX18 күн бұрын
I thank thee :)
@huson605819 күн бұрын
Great video dude!
@tylerdurden428519 күн бұрын
Great vid
@neoProfessor19 күн бұрын
Thanks
@rayniererasmus925020 күн бұрын
Thank you so much. This was super helpful :)
@neoProfessor20 күн бұрын
Glad it was helpful!
@hosseinmalekian1316Ай бұрын
out of frame !
@kristofszilvasi9021Ай бұрын
Could you please include your negative prompts in the descrption? I've been suffering with all of these uncanny generations lol
@TimeTalesTTАй бұрын
How can we use lora with difution?
@neoProfessorАй бұрын
To do the continuous generations like I did at the start of the video: Start ComfyUI Make sure seed is set to fixed in workflow Select Extra Options ( below Queue Prompt) Select Auto Queue Select Instant Now ComfyUI will generate an image whenever you change you prompt
@tvwithtiffaniАй бұрын
How do you load a checkpoint merge safetensor but without the UI? All tutorials and youtube videos lead to automatic1111 or diffusion web ui instructions.....
@rwarren58Ай бұрын
So that's the extra slow way. Thanks!
@MatthewCrawfordАй бұрын
Code LLM available too!!
@sirk3vАй бұрын
eew windows
@mewtcatАй бұрын
I think the biggest setback of ComfyUI for those just picking it up is the concept of Custom-Nodes. Trying to learn from the workflow of others can be extremely frustrating, because 9/10 you will be missing a node, and not even have any idea "Find missing Custom Nodes" in Node Manager (or even node manager: since it is pretty much a essential part of ComfyUI and does not come with ComfyUI by default) is a thing, AND even if you do know, chances are you will run into deprecated and even niche hard to find nodes not in the Node Manager repo. It's all a bit much to learn and keep track of. Going forward, I think making the workflows more streamlined and a better repo for nodes would make a big difference towards its adoption and retention rate.
@neoProfessorАй бұрын
Yea it can be a hard to deal with at times, think it would be nice if comfyui came with prebuilt workflows for popular workflows like controlnet and img2img which you could access from tabs, like in automatic1111.
@wakegary3 күн бұрын
This is what got me, and I'm a diehard Blender Geometry / Shader Node guy. It's messy and I already hate conda/pip work with Windows package management. So this takes that concept and makes it more modular so the custom nodes bash in the creative flow - that said i came here to get a nduget to jump back in. crazy that the vanilla download with the 'basic's is 50g. I I'm gradio'd out anyways
@FrostyDelightsАй бұрын
very clear explanations, gj
@daruthebeastАй бұрын
you should've kept the pixel art, used a pixel art lora+cloud lora on the model and img2img with half denoise, probably 1 sprite at a time is way better too, Issue is that the sprites now don't align correctly and are too noisy, so they'll never be of enough quality to be used. I've been researching how to use AI for assets and the noise kills animations.
@neoProfessorАй бұрын
I tried batching sprites to help keep consistency but I see what you are saying, noise is definitely an issue Hopefully new tools will come out which will fix these issue
@SommervillleАй бұрын
These videos are an invaluble resource!
@neoProfessorАй бұрын
A slightly different video from the regular tutorials, let me know if you like this challenge format
@steinscamus8037Ай бұрын
No way to do something similar on a1111?
@neoProfessorАй бұрын
Should be, try searching for A1111 ipAdapter on youtube, you will most likely find something
@lanoi3dАй бұрын
I always get a ton of extra hands, glitched limbs, floating spare heads, etc. With any models I download it's the same.
@seto007Ай бұрын
Whilst ComfyUI is definitely harder to grasp initially as a beginner, I've found it a lot better with regards to actually teaching you what stuff does since everything is exposed and editable under the hood, and thus examination of and modifications to the workflows of others that you download demonstrate more tangibly how the individual settings work off another to produce the final image
@puppetmaster442Ай бұрын
uhh a little correction to the title. Why everyone else's Stable Diffusion IMAGES are better than yours
@user-un1iv3tg5t2 ай бұрын
Hi, I was wondering if there is any way to bridge a LORA model that is trained on SDXL into a LORA that can run on SD 1.x?
@neoProfessorАй бұрын
This method works for using SDXL loras on SD1.5 models as well Besides that there is no real way to bridge a lora model, that I know of
@somyeol8042 ай бұрын
Interesting, thanks!
@ewenchan12392 ай бұрын
Thank you for this video. I just deployed Automatic 1111 last night, so I'm just getting into it and there is a LOT more to learn.
@siruno60242 ай бұрын
Why my comfyui generates images in seconds but my automatic 1111 takes minutes, even hours.
@captainpike34902 ай бұрын
ComfyUI I only use if it is really necessary, If you use Forge WebUI over Automatic1111 that is huge performance boost especially if you have Low VRAM. Personally I use Fooocus WebUI it is so practical image quality is better with foocus V2 expansion and styles, in painting very good and handy face swap image to image more than enough. If you have low VRAM ist is so fast and esay to use SDXL lightning models. Performance presets amazing.
@neoProfessor2 ай бұрын
Will have to check both Forge and Fooocus out when I can
@11_amankumarmall14Күн бұрын
Hey do you know how to run ipadapter in forge or you know any other image generator in automatic 1111 image quality but run faster also gives a lot of controls like comfyui like ipqadapter faceswap consiatency and others
@captainpike349021 сағат бұрын
@@11_amankumarmall14 Try Fooocus so handy find this channel on youtube there are so many tutorials "Kleebz Tech AI" I can not give you link if you use link your comments automaticly deleted by youtube
@MadMaxwell72 ай бұрын
Neither, just use Forge. It's A1111 that runs on 4gb of graphics memory easily. It's better and faster and wont freeze up your pc.
@OutOfWards2 ай бұрын
Out of all three, Comfyui reign's supreme. Not only for the speed of generation but also for the mere ability to be more in control. :)
@renlow2 ай бұрын
Yeah Forge is a lot faster than 1111. And with Comfyui u will vaste hours and hours with errors from missing notes that it cant find automaticlay. i liked the idea of comfy, but have given up.
@IOSALive2 ай бұрын
Neo Professor, You're fantastic! Let's be friends and have fun!
@bigglyguy84292 ай бұрын
I'm getting the general hang of it but how to write the prompts? My attempts so far have been nightmare fuel...
@DeanCassady2 ай бұрын
Pls continue the series to explain the basic or next level.
@DeanCassady2 ай бұрын
I'm so tired to Thumbs up whole series😁😁😁😁 (I learn much from you, many thx)
@albertogarciagomez74602 ай бұрын
Awesome! your way of explaining things is on another level. Thank you very much for these videos, I hope you continue in the future! You are amazing at making tutorials damn!
@WaseemOnlines2 ай бұрын
I have a question, 2:47 why the image is not 100% same as you copied all the same prompt and settings?