Reference Only - ControlNet Method - WOW! You NEED this NOW!

The new Reference Only ControlNet Method is very Powerful. It can create similar images from just a single input image. You don't need to train a Model or a Lora. You don't even need a complex prompt. While other ControlNet Methods rely of composition or image Details, this method give you artistic freedom.

#### Links from the Video ####

#### Join and Support me ####

Support my Channel:

[Lansolot]

Wow, we are moving so fast. This channel is always the first of the bunch to put out videos on new tech.

[moe_joe_man]

Loved the bloopers! Thanks again for your amazing videos. They help me out immensely.

[wojtekmazur]

It doesnt realy work for me. Every result I get is just completely different.

[RoboMagician]

i have controlnet 1.1.170, and followed the instruction, the generation came out completely different from the original image. anything wrong?

[aldiergreen]

even if i don't learn as much as i thought I would learn (my bad, not yours... lack of attention & retention) I always end your videos with an involuntary smile stamped on my face. a very wide smile. I know (singing in falseto), you have cool vibes! :))

[OlivioSarikas]

#### Links from the Video ####
#### Join and Support me ####

[Saimsboy]

Thx for your amazing work =D
This is getting better to fast <3

[LeKhang]

Thank you for sharing. I kinda like the bloopers at the end hehe.

[youngbroadway]

Cant lie.. You give us all the GOODIES bro! Appreciate your work! 💯

[clamhammer]

I can't seem to get this to work. just takes my prompt and creates a new image.

[rbscli]

Would be nice to see a comparison between this and img2img, ip2p and maybe other controlnets like tile, maybe. Also the explanation behind this is nice to understand what problem this controlnet was tackling. From what I read, people were using inpanting when they wanted a second subject with similar characteristic to another in a image generated, they would use inpanting with mask so the model would reference what was in the image but would create another one, but it needed more prompting like "2 dogs", etc. But with this it is an easy way to do it.

[siltoruz]

Great video as always Olivio! This system is exactly what i needed for the current project i m working on. My boss will be delighted! Thank you very much for bringing those news out the moment these tools launch or update! Super useful content. I hope your channel keeps growing fast. You deserve this!

Opa from Greece! Na eisai kala!

[RoguishlyHandsome]

I tried with photos and it failed miserably. It creates an *completely* different face. It even removes glasses. If I put no prompt, using RealisticVision, it almost turns the man in the photo into a woman.

[CoconutPete]

Once I get this mastered I'm going to make a detailed tutorial on some of the stuff I struggled to understand. The biggest problem with AI is the technology is advancing so quickly, existing tutorials go out of date a week after they are uploaded 😮

[chibieisu]

i found with this that increasing the steps to something greater than 30-40 steps improved the initial image quality alot.

[DJVARAO]

Wow Olivio, you are on fire!
This is fantastic, thanks!

[CleomokaAIArt]

OMG, was this reference only there in control net 1.1 this whole time or is it a recent addition? This is seriously the most awesome option in control net by far! You can place ANY image as your reference, and your prompt will use that image for style, look, everything. I am playing with it and the results are mind blowing. Put an anime photo with referend pose, put a realistic completely unrelated prompt, all of a sudden your prompt will be build using that as a reference, it's so incredibly powerful and a total game changer, with so many possibilities. Seriously this has so many crazy use cases, many more than the ones you described!

[oaahmed]

Hi Oli, it's great to see you again ♥

[StSpoiler]

why for me not working with refrence only? fellow each step you did. still not getting same result. completely different. 😭

[blackvx]

Keep it up, thanks for your help. Staying up to pace is hard work! 😊