Great stuff and thanks very much for taking the time to do this. I've played around with local Flux and Loras, but didn't know how to do this.. BTW, one small tip, "Euler" is pronounced "oiler" (just as "Gauss" is pronounced "G-ou-s," ou as in ouch).
Lol that apple vs android comparison 😂 Problem is that most apple users don't know what they are missing. I like Forge better because I get things done faster. I run into config and errors with Comfyui more often. There are no such problems with my S23! 😂
Really comprehensive tutorial. Thanks. Is there a way do this without using lora if we dont care what the person exactly look like as long as it is consistent? For example, i want multiple pictures of a couple with their son. They should look the same with the same clothes across pictures.
Hi i downloaded your workflow from patreon, but i cannot solve this error: Any help please? Thanks for amazing content! Failed to validate prompt for output 221: * BasicGuider 234: - Required input is missing: model
What happens when you want to add two people into an image but you don't have the two people you want to use a picture together. You have to photoshop them in
Hey man thank you so much for your help.. Can I ask you what kind of version of comfy are you using for these consistent characters workflow to work? I heard that these wont work with specific versions of comfy... I m currently using the pinokio version do you think its good? Should I download it directly from github or maybe stability matrix? What do you suggest me to do? Thank you so much for your help bro
Is it possible to use two different LoRAs for two people? For example, if I have a LoRA of myself and one of my wife, can I apply both separately rather than combining them into one? I'd like to use my LoRA with a strength of 1.0, my wife’s LoRA with a strength of 1.0, and add a realistic LoRA with a strength of 0.2. Would this setup work?
The best results I've found have been with two people trained in one LoRA. I've tried the separate loras but ran into some character bleeding. Regional Prompting helps but at the moment it is experimental. We posted our findings, workflows, and results on our Patreon.
Can I do this online instead? This is a lot more complex and I don't want to download a lot onto my PC I saw that there's some downloading involved from other videos and all
This would have taken less time with photoshop 🤔it's just taking the heads and putting them on different bodies. If it was a complete remodel of yourself ok cool. But copying your head on LeBron body🤷🏽♂️
It's a complete remodel. You can generate yourself as anything. So you could keep the same clothes, body type, skin, etc. If you check out the later example with the Paris photo, you'll see it even generates my backpack and our heights accurately. The fact it looks like a head replacement is impressive because technically everything is made from scratch.
![Shoot EPIC MOVIES with this FREE AI Tool! [ComfyUI Tutorial + Free Workflow]](http://i.ytimg.com/vi/gHI6PjTkBF4/mqdefault.jpg)
![Shoot EPIC MOVIES with this FREE AI Tool! [ComfyUI Tutorial + Free Workflow]](/img/tr.png)
thanks for the video. good to know sam detector. didnt know theres such function
Great stuff and thanks very much for taking the time to do this. I've played around with local Flux and Loras, but didn't know how to do this.. BTW, one small tip, "Euler" is pronounced "oiler" (just as "Gauss" is pronounced "G-ou-s," ou as in ouch).
Glad to help and thank you! I've been getting slammed on my pronunciations in our previous video too haha thanks for the easy tip
Lol that apple vs android comparison 😂
Problem is that most apple users don't know what they are missing.
I like Forge better because I get things done faster. I run into config and errors with Comfyui more often. There are no such problems with my S23! 😂
Amazing!
Really comprehensive tutorial. Thanks. Is there a way do this without using lora if we dont care what the person exactly look like as long as it is consistent? For example, i want multiple pictures of a couple with their son. They should look the same with the same clothes across pictures.
🔥 thank you
Hi i downloaded your workflow from patreon, but i cannot solve this error: Any help please? Thanks for amazing content!
Failed to validate prompt for output 221:
* BasicGuider 234:
- Required input is missing: model
What happens when you want to add two people into an image but you don't have the two people you want to use a picture together. You have to photoshop them in
WHere is the FORGE link? Do you provide private coaching ?
Yes, email us at hello@blackmixture.com
And here's the link a to book a private coaching with me: cal.com/blackmixture
Hey man thank you so much for your help.. Can I ask you what kind of version of comfy are you using for these consistent characters workflow to work? I heard that these wont work with specific versions of comfy... I m currently using the pinokio version do you think its good? Should I download it directly from github or maybe stability matrix? What do you suggest me to do? Thank you so much for your help bro
Hey I think getting it from Comfy github is the best place. The comfy github will have the latest builds which I'm using to generate these.
Is it possible to use two different LoRAs for two people? For example, if I have a LoRA of myself and one of my wife, can I apply both separately rather than combining them into one? I'd like to use my LoRA with a strength of 1.0, my wife’s LoRA with a strength of 1.0, and add a realistic LoRA with a strength of 0.2. Would this setup work?
The best results I've found have been with two people trained in one LoRA. I've tried the separate loras but ran into some character bleeding. Regional Prompting helps but at the moment it is experimental. We posted our findings, workflows, and results on our Patreon.
@@BlackMixturethanks for you reply! ❤
Great tutorial 🙏🏼
1st!!!
Can I do this online instead? This is a lot more complex and I don't want to download a lot onto my PC I saw that there's some downloading involved from other videos and all
freepik
Is this free?
100% free
2nd 🥲😪
What’s with the AI dubbed voice?
Lol that's my real voice, though I think the mic had a pretty strong noise gate to block out my desktop gpu fan running at max
ahaha voice just didnt match
Bruh you're literally the A.i Lv 40 boss
Lmao thanks!
This would have taken less time with photoshop 🤔it's just taking the heads and putting them on different bodies. If it was a complete remodel of yourself ok cool. But copying your head on LeBron body🤷🏽♂️
It's a complete remodel. You can generate yourself as anything. So you could keep the same clothes, body type, skin, etc. If you check out the later example with the Paris photo, you'll see it even generates my backpack and our heights accurately. The fact it looks like a head replacement is impressive because technically everything is made from scratch.