Create CONSISTENT CHARACTERS in AI SCENES Comfyui Tutorial

[@GiancarloBombardieri]

great!!! so cool explanation!

[@goshniiAI]

I appreciate your feedback. thank you

[@cr_cryptic]

You was damn right. I didn’t know where to begin. I wanted to give up. You literally just did like two master class videos on it. Thank you so much! Finally! I’m so eager! 🙏 Subscribed!

[@goshniiAI]

Thank you for being here and taking the time to share your thoughts! We welcome you to the family.

[@crippsverse]

A really fantastic tutorial. I need a beginning and end image for a series of AI videos I am making. This is perfect. Thanks 🙂

[@goshniiAI]

You're very welcome!I'm glad it was helpful.

[@HestinPF]

brooo wtf im rly bad at math but with your explanation im sure i will be expert in quantum physics, your job is amazing keep doing it

[@goshniiAI]

Lol. I'm really glad the explanation helped things click for you. Your support means a lot

[@kyle-bensnyders3147]

Not sure about the consistency factor being very high. But great video. I hate these other ones that just tell you to download a workflow without explaining it fully or building it from scratch like this one

[@goshniiAI]

I’m glad you appreciated the step-by-step approach, and I totally get what you mean about the consistency. It can be a bit tricky, but with some fine-tuning it definitely improves.

[@97BuckeyeGuy]

This is an amazing workflow. Very well done and well explained. It works really well with your previous character video.

[@goshniiAI]

I'm really glad the workflow was helpful and worked well with the character creation video. Thank you for sharing and for your support. I appreciate it!

[@ToddDouglas1]

Awesome! I love that you build it step by step for us. Picked up quite a few shortcuts. The only part I couldn't get to work right was at the end - My Upscale Latent By. No matter what settings I use, the image looks garbled and bad. But skipping that part, everything is perfect!

[@goshniiAI]

Awesome, I'm really glad you found some handy shortcuts! I'll work on making the upscale process easier to explain since I've dealt with it a few times too. Thank you for the feedback!

[@devnull_]

Probably not going to touch SDXL, but enjoyable to watch anyway - was going to only watch a short clip to see what is going on, and ended up watching the whole video lol, color choices in both images used and video elements are very nice, no messy node layouts either, and nice editing too ✨

[@goshniiAI]

Thank you for sticking around for the whole video. I appreciate your kind words about the visuals and the editing-it’s great to hear the colour choices and clean layout stood out for you.

[@saadabbas]

well done beautifully said.

[@goshniiAI]

Thank you, and I appreciate the feedback.

[@squallseeker-i2i]

Around 5:08, shouldn't the model from the ipadapter for face have fed the input to the body's ipadapter loader? As is, it seems the first ipadapter advanced node is being bypassed, since its (model) output goes nowhere.

[@goshniiAI]

Yes you are right and a good observation, the model input can be linked too, however i realised through the process that different connections created different results, therefore I went with the option I used.

[@chuncheon99]

So cool explanation. Thanks

[@goshniiAI]

You're very welcome. I appreciate your feedback.

[@onepgone5879]

还是喜欢你极简的工作流 ，希望博主大人多做几期，简单且好用的工作流

[@goshniiAI]

Thank you for mentioning that. I will keep it in mind.

[@richardrispoli4508]

Nice video, You could use a pre processor to extract your pose without troubling yourself.

[@richardrispoli4508]

And even though it's one more custom tool to install, I'd suggest you start using the connectAnywhere nodes, they are easy to find, they would really clean out your type of nodes layout, the vaes for example are always taking a useless path that aren't helping reading through your workflows

[@goshniiAI]

You are absolutely correct about using a preprocessor to extract the pose, especially for quick workflows. I used a different approach to introduce the pose editor, so viewers could choose what best suits their style. Thank you for the tip and your observations!

[@richardrispoli4508]

Makes sense. I also saw another video where you used it. 🎉 Your layouts are very similar to mine :) Not just the work flows, the way you layout your nodes

[@Alivangs]

It s look good... Waiting for workflow

[@goshniiAI]

Thank you! You'll be able to access the workflow in a few hours.

[@thegtlab]

Can we do this with a flux model instead?

[@goshniiAI]

Yeah, Flux is still possible but using the Ipdapter is inconsistent, therefore using SDXL. But the workflow and process remain the same

[@m.ai.chi.22.]

Well explained, can you do one for animal characters ? Any suggestions

[@goshniiAI]

Thank you for your feedback. For animal characters, i haven't given that a try yet, but you can experiment with "huchenlei/animal_openpose" for controlnet. that’s an awesome idea

[@samlavi]

Amazing! Thanks

[@goshniiAI]

Thank you for being here, and you are welcome.

[@EpicNarrationTales-i3j]

Hi can i do this with multiple consistent characters? And lastly i want to make the characters from a photo, eg i want my face to be on a scene, can i create a character following your last video, but using my image?

[@goshniiAI]

Hello there, Absolutely, you can work with multiple consistent characters using this workflow! It's all about setting up the ControlNet nodes properly for each character. As for using your own face in a scene, yes, you can definitely achieve that! You’d follow the same steps from my last video, but instead of a generated face, you’d feed in your photoreference features to guide the character's appearance.

[@lordmo3416]

error: IPAdapterUnifiedLoader ClipVision model not found.

[@goshniiAI]

Hello Lord, the initial workflow did not include a clipvision node and model, which you can do without using the IPADAPTER; however, if you want to get the models, you can follow this video guide to get all of the models for the IP Adapter. kzbin.info/www/bejne/j4qsep6dq7SNaa8si=K9wgcIWphnyoROA_

[@lordmo3416]

@@goshniiAI thanks a lot... it works... and I enjoyed the video

[@thunpax]

🎉🎉🎉

[@goshniiAI]

💚💚💜

[@abhinavaserkar1055]

Nice

[@goshniiAI]

Thank you kindly

[@KoganeNoYubi]

how to get this comfy ui bar in the browser?

[@goshniiAI]

Hello there, i have explained how to change the UI bar in this video towards the ending. kzbin.info/www/bejne/hoGzgmSJdrOGma8si=Af8g0oOOW3r2eB7I

[@technobabble77]

What's the name of the extension that creates the top toolbar?

[@goshniiAI]

The extension is called Crystools, and you can find the installation guide here. kzbin.info/www/bejne/qnrJo5KaqryskLMsi=3lqvI39VblZDfy0-

[@qsguo]

太棒了！

[@goshniiAI]

Thank you very much.

[@chochlik223]

Would you be able to create a workflow on comfyui in which at the very beginning you would add a photo of the character (let's say it would be a photo in just underwear), then you create a so-called bra and panties mask. From this mask, the bra and panties are created on a white background. And at the very end add an upscale to these photos on a white background. I think it wouldn't be a problem for you and you would really help me a lot.

[@goshniiAI]

Thanks so much for your suggestion! I’ve actually come across a few tutorials online that are quite similar to what you’re looking to achieve. I’ll share the link with you in my next message-hopefully, it’ll be helpful!

[@chochlik223]

@@goshniiAI Ok, will be waiting for the video. I was looking for it myself but didn't come across just the kind of tutorial I need. I know you made a video on how to create consistent characters. I respect this tutorial very much. I have a question, however, if it would be possible to do so that instead of generating characters in comfyui it would be possible to insert a picture of the character directly into comfyui and based on that picture of the character create all those poses that are originally in the workflow you created? Probably it can be done that way and without any problem, it's just that I don't know what to do and how to do it. I hope you understand what I mean.

[@chochlik223]

@@goshniiAIHi brother, I think you forgot to write me back on the comment, but nothing is happening, however, we hope you will write me back now.