1. Thrilled to get a message from you 😍 2. Noted on that, I was getting a few errors in my testing, which is why I opted for it 3. Is my Unet explanation suitable? It’s very basic, but I know you covered it in one of your videos too right? Thanks again Mateo @latentvision!

@@EndangeredAI yeah for the purpose of these youtube videos is totally fine... and you also pronounce my name correctly which is rare 😄 . Good luck with you videos hope to see you on my discord!

@latentvision I should hope so. From your accent were geographical neighbours 😂

Well said! 😂

after a certain point, it is that simple though. im an artist who is learning comfy in order to have superpowers as an artist. Im a proponent for fellow artists to get into localized AI and incorporate the multiple years of experience to make super artists. but at the end of the day, 2 weeks of learning how to set up comfyUI and the ipadapters, controlnets, etc VS 4 years + of learning how to draw, paint, learn anatomy to a mechanical degree, 3d, light and color theory, composition, etc. is a huge difference also how long it takes to paint an image vs generating one. 2 weeks for comfyUI to be in place 2 weeks to paint a good piece, give or take. (after the 4+ years of experience) then another 2 weeks to paint another piece, (or somewhere in between.) compared to spamming generate button and pushing art after art after art in a matter of seconds to minutes depending the gpu once I have this set up, ill be able to do hundreds of iterations a day, and be a super power as an artist. I look at this art AI and AI in general as a superpower. yes it takes time to set up comfyui but once its ready to go and you have each workflow in place for whatever purpose. its clockwork

@@skrollreaper Yeah. It almost becomes an axe VS chainsaw issue. The chainsaw is actually the more complicated tool to build and maintain, but the axe (to be used effectively) requires more time, fitness, and practice. At the end of the day, those trees are coming down. Reminds me of John Henry VS the steam engine.

​@@virtualalias It's not quite like John Henry nailing nails to a railroad track. Art is a science. just because the program can generate art doesn't mean it's creating compositionally sound art or art with good anatomy. It's passable, sure, but even if it gets close to nailing good anatomy, the composition might be off, etc. among other hinderances for less skilled artists. I see it more as the swordsmith creating pristine blades and a robot making pseudo-pristine blades. An untrained eye can't see why the robot's blades aren't pristine, but a swordsmith can see the flaws, and sometimes major flaws that would break the fake pristine sword if in use. the same goes artist who learns the skills it takes to make a masterpiece with lighting, composition, and painting techniques, we can see the major flaws hidden underneath good rendering, etc. flaws it would take years of anatomical understanding, compositional, and other theories to see. for the artists who are 20+ years of seriously developed knowledge, they live in a realm where AI would slow them down actually. it would be quicker to just hand draw everything because the level of understanding is so vast. waiting for an ai generator to prompt it instead of just drawing 50 poses, or 50 designs would be quicker. im more of a 6+ year sworsmith. so for me it will just fit right in with my workflow and actually speed up a lot of skill acquisition. so in my case its more like the swordsmith who makes only pristine swords, now uses the robot that if, by itself, would only make pseudo-pristine swords, weak and fragile in areas. But with the swordsmith's craftsmanship, they can now produce 10 truly pristine swords in a month or more. So for an artist who can match any style, using the AI is more like if the swordsmith equipped the exosuit. Someone with a superpower that was handcrafted, combined with the technological superpower. If everyone has superpowers, then noone has. but if someone with a superpower combines with another superpower. they have a super power again.

@@skrollreaper Yeah AI is still a ways off. And I always think there'll be the "human element" which AI can never replicate. At the end of the day, we're explaining what we want. And you may or may not get it, and even if one thing is off, and you want it to be PERFECT, getting an AI to do that is next to impossible. It's just randomness thrown together to give you kinda sorta what you're looking for. And I 100% agree that if a REAL artist used AI as a tool and not see it as a threat, it could increase their own work by a lot, especially with poses, concepts, compositions, etc.

Glad it was helpful!

Glad it was helpful!

Wow, thank you! I’m very happy the video has been helpful for you!

Glad to be helpful! I found other guides had this long complicated process, but… this is much simpler and handles 90% of situations haha

Glad you liked it!

Glad it was helpful!

It means a lot to be compared to mateo! He’s one of my favourites too! It means a lot, as I try and be circumspect with the content, even if it is sponsored, like the scenario one, which is a tool I actually use I try and bring value!

Hmmmmm that’s an interesting question. I’m guessing it would depend on what else is in the image. If you have a person holding the shield, that makes the scene/prompt more complex resulting in greater room for error. You might be able to use a generalist workflow but it would be complex, require multiple passes and may need an image detection node which can dynamically detect the area to manipulate and inject noise accordingly. The downside is that you may struggle with even a 4090 as I imagine you’ll need to load a bunch of things into memory. For something like this you might be better off with a lora trained on the icon, which you can then easily reference in the prompt “shield with icon printer” where icon is your token

You're right. I don't think I can avoid the effort

You need to reinstall it using comfy manager. So best practice is uninstall with comfy manager then install it again :)

@@EndangeredAI thank you and the discord link is invalid

There’s a load insightface node, you likely need to use it and link it

I’m a Linux user, so that’s the Linux command terminal. If you’re on Mac or windows your regular terminal should work, just make sure you cd into your comfyui folder. The same one where the requirements.txt folder is

@@EndangeredAI Thanks EndangeredAi for the quick response. Install -r won't work does it need to be a ? I can't get it to work :( It seemed to install ok using this method kzfaq.info/get/bejne/rKlzidt42andmKM.htmlsi=QIXBL8kUT1chjppr , but comes up with "insightface required" if I try to use a faceid ipadapter model. I have uninstalled and reinstalled the ipadapter nodes. On day 2 of trying to get this one module to work now:(

@@EndangeredAI Thanks, neither or works from me, this one module is giving me a lot of trouble :(

The second one should have worked pip install -r requirements.txt what error are you getting?

@@EndangeredAI no such command, probably user error. Think I need to run virtual environment first? but can't remember how. I'm a bit out of my wheelhouse tbh, I'm a 3D artist not an IT guy lol :D I have had partial success with this method - kzfaq.info/get/bejne/rKlzidt42andmKM.htmlsi=uYbRQwSqFdGbi07t, it seemed to install ok, but using any faceiD ipadapter models throws up errors with missing insightface. nodes module was uninstalled and reinstalled.

It’s in the next video ipadspter -> facial expressions

Crl c on the terminal

Crl+c 😃

What prompt are you using? Come by the discord and drop a screenshot

Here’s the link, come on over and we’ll help you out discord.gg/yH7KfsYm

Ah! Ok so couple of things to check! I suggest you come by the discord and ask, but off the top of my head; 1. If you are using an advanced IpAdapter node, add a load clipvision and connect it. 2. If you are using a unified loader, be sure to check that you’ve downloaded and properly renamed the clipvision models. That’s usually the two main causes. Anything else come by the discord