Create Consistent Ai Characters from a Single Image!

[@taoprompts]

This method works with any character illustrations you have, whether you made them yourself or created them using Ai!

[@GaladrielMarvelcomics]

You amazing buddy please character prompt best Saitama, one punch man 💪 😊

[@simongus]

Great video, thank you! This process also works for realistic models? Like Instagram AI influencer? Are you able to create NSFW with this tool?

[@TheNadir1983]

You're simply THE BEST.

[@morgangitonga]

​@@simongushe says it works well with non-human characters

[@SufyanBloch-vy7ps]

That's free or not?

[@Tetarean]

you are the first youtube i´ve found that actually made a step by step on how to create consistent models, tyvm

[@Bruno_Borja]

This is incredible! You added audio tracks in multiple languages, including Portuguese (my language)! I can't thank you enough. Truly appreciate it!

[@taoprompts]

KZbin allows audio tracks now! It's a really nice feature they added 👍

[@TerenceKentSEO]

You are a smart man! I have tried so many techniques and could never find one that worked. I would've never thought of this approach, thanks so much for posting this amazing video!

[@gonzo2.0]

You are a true AI master. I have learned quite a bit from you , thank you very much have a great new year.

[@larbibenmhidi3054]

One of the best AI tutorial I ever seen ! Thanks..

[@jasonmccarthy9880]

Brother your video have helped me out lot.. Your vids are actually USEFULL and in depth.. keep up the Great work my man ..

[@taoprompts]

Awesome to hear that these guides are helping you out man!

[@Sean43322]

You're doing great, Tao!❤💥 I've been waiting for this tutorial. It was very helpful and practical. Thank you.

[@taoprompts]

Thanks Sean! it's great to be able to do this with only a single illustration

[@ReveRecordLabels]

와... 선생님 정말 언제나 신기한 신기술을 보여주시는군요!!! 오늘도 매우 매우 고맙습니다 ; )

[@taoprompts]

감사해요! 최근에 친구한테 이 방법을 배웠는데, 공유하고 싶었어요! 이것이 도움이 되기를 바랍니다🙏

[@ArNiFlo-Nio]

Great video ... I appreciate that you share your knowledge ... for those who are still learning

[@taoprompts]

Thanks! I learned this method from my friend also 👍

[@giri.goyo_yt]

I would never have thought if this! And I take still from video and input them as starting points in Hailuo! I shouldve kept going. Thanks for your genius!!!

[@taoprompts]

Thanks, Ai video is a great way to get multiple images for a character! I learned this idea while working on a film with a collaborator recently

[@mysticmaze2268]

This is a FANTASTIC IDEA! .. .Wow ... why didn't I think of this. Thanks a million. Now I can train a character with ONLY AI. Sooooo happy I watched this. "SUBSCRIBED"

[@taoprompts]

Thanks for subscribing! It's amazing to be able to do this with only a single illustration

[@100daysbuildwithAI]

Thanks! I hardly ever comment but this actually worked 😍

[@nickwee258]

Brilliant video. First time I've seen your channel and I'm happy to subscribe.

[@teenola5783]

You did your big one with this bro! I wish you much success! THANKS!!!!

[@teenola5783]

you have to pay for openart to be able to create your own model. just a heads up guys.

[@HaydnRushworth-Filmmaker]

Using Hailuo video for reference images... YOU, my friend, are an absolute GENUIS!!!!!!!!!!!!!!!!!!!! Genius, genius, GENIUS!!!!!!!

[@ShifterBo1]

We all did that before, he just adversited it, useful anyways.

[@nickwee258]

@@ShifterBo1 I been using Hailuo for more than a month and that never crossed my mind. It's an AMAZING video creator. I have paid for it for 3 months and I almost never do that.

[@taoprompts]

Thanks Hadyn! It's nice to be able to get a bunch of training data from just a single image. Ai video is getting really consistent now

[@excelerator]

This is really good! I did something similar before, but taking screenshots of the generated video. Your use of the converter, and a few other tips you had in your process, are excellent!!

[@taoprompts]

Thanks man! Congrats on how well your channel is doing!

[@fabiosanfig1]

Hello from Brazil! Great vídeo! Keep it up!

[@taoprompts]

Thank you! Great to hear from Brazil 👍

[@stargordo]

Very good, my friend. Thank you.

[@ehnews2345]

That's exactly what I've been looking for. Many thanks. The quality is amazing 👍👍👍

[@taoprompts]

Thank you, and glad this was helpful!

[@rhythmfittv4645]

FINALY!! This works so well, THANKYOU👏🏾👏🏾👏🏾👏🏾

[@Yumator]

Thank you! Your findings are great as always! Happy holidays!

[@taoprompts]

Happy Holidays to you too! Glad to know this helps 🎉

[@SamhainBaucogna]

Good idea master! 👏👏👏 Thank you!

[@taoprompts]

I've been using this workflow recently, hope this helps!

[@MrJonathanwolff]

This is such a clever way to get the initial pics. I was using MJ but love this method. At moment Kling 1.6 creates a more dynamic animation for my stylised character over minimax live model in my opinion but will do some more tests. I have learnt lots from you. Thank you and Happy New Year 🎉

[@taoprompts]

Thanks! Kling is also really good, especially with their new update. Although I like Minimax for subtle 2D illustration videos. Happy New Year 🎉!

[@Flixstarstudios]

So clever! i didnt even think to use video to get new angles lol well done bro

[@Pervy0428]

This is by far the best consistent character tutorial out there! Thanks for making this... although is there any way I can get a frame of multiple consistent characters using this method?

[@MyAILife-o9p]

Your tutorials are incredible. I've just gotten started with this AI creation. Uploaded my first video to my channel. It's pretty basic but I'm hoping to improve. Your channel is a great reference!

[@taoprompts]

Great job with your video! I like how many different characters you added in 👍

[@MyAILife-o9p]

@@taoprompts Thanks, I had Chatgpt write the script!

[@amandabergamoantunes5560]

Hello Tao, thank you very much for knowledge, this helps me much.

[@taoprompts]

Great to know this helps you out!

[@prompt.try.repeat]

This method is super smart!

[@YouUsAndAI]

Tao, thank you so very much!!

[@lovegoodmusic2477]

Very helpful ✨👍🏻

[@romerrojo8217]

Excelente, gracias por compartir, son cosas básicas que uno a veces olvida

[@mikrodizels]

Incredible, very useful. Regarding AI Video creation, it's easy to create great looking 6 second clips of AI videos, but the problem is combining them together for a smooth and long video, so it's all a long and consistent sequence/movie. The separately generated ~6 second clips always have a different and unique camera movement/speed/background details etc. even with Img to Vid, if you use the same or adjusted prompt, and use the last frame from clip #1 as the first frame of clip #2, the clips still will always be generated too different, to be combined into a seamless 12 second clip for example.

[@arnoldasmureika]

Thank you and Happy New Year 🎉

[@taoprompts]

Happy New Year to you also !

[@chubbyartstudio]

fantastic idea 🎉

Awesome idea - thank you.

[@taoprompts]

I learned this while working on a film recently!

[@dr_ganz]

Thank you very much for this!

[@JohnTWilliamsFilms]

Your prompt guides for Hailuo have really helped with character motion. Especially, switching to the Live Character model and experimenting with turning the prompt enhancer on or off. BTW, will you be reviewing Google Veo?

[@taoprompts]

Glad to hear those prompt guides helped! I don't have access to Google Veo yet. If it ever gets publicly released I will make some guides for it

[@AbsolutelyJason]

Thank you!

[@AIForWinner]

Viewers learn how to use the image-to-video interface to create different poses and angles of their character by inputting simple motion prompts.

[@zengrath]

This is really impressive. really cool work! I'm hoping one day soon we can do a lot of this locally rather then using online tools.

[@taoprompts]

That may be possible soon! Some of the best image models are open source right now, I'm sure some really good open source video models will come out also

[@gemini9775]

Interesting thanks 😀

[@r.k.8206]

You kill it! Unexpected way)))

[@iluj]

Ficou dá hora, parabénsss pelos seus videos

[@livinglifeinthailand]

Dude You doing the damn thing,,, thanks for taking out the time to create a method and sharing it,, I have 7 different animated properties that need production I cant afford to hire a team for that,,, thanks for the tips, and your time.

[@taoprompts]

Oh wow, sounds like you got a lot of work you're doing. I hope this tutorial makes it easier 👍

[@livinglifeinthailand]

@@taoprompts Yes I already started using the technique still learning but need to transition over from traditional animation to this for affordable production reason's thanks again.

[@F.C.Poet_]

nice, thanks

[@MilanaGaldi]

Thanks a lot! You are the best!

[@YouzerseifYoucef]

This guy is animating his own face for each sentence he says, i don't see why he would do that but it looks good asf

[@RedEyesReels]

Please suggest best tool for creating consistent real person character with same body size in every creations

[@AmsterdamBicycle]

Thanks!

[@ranedit1]

Continue dont give up bro

[@Nuverotic]

This guy looks like the Asian Tom Cruise. lol

[@happixs]

Great ideas!! Thank you so much ❤

[@Samt2b]

It would be nice if you add multiple consistent characters cuz i dont think you did before, also includ control net

[@taoprompts]

I have some guides for multiple characters using replicate & loras here: kzbin.info/www/bejne/nl6ahHx8gZuahpY although that's not possible in open art yet.

[@FilmSpook]

Brilliant Workflow! Thanks, Brother. Wishing you a Successful and Prosperous 2025 and Beyond. Cheers 😎

[@taoprompts]

Hope you had a great New Years man!

[@cbnewham_ai]

Similar usage of AI video to my recent tests for architecture - using Kling to orbit a scene using a seed image and then placing the resulting video into a photogrammetry program to generate the 3D model.

[@notlessgrossman163]

How well does this handle scale?

[@cbnewham_ai]

@notlessgrossman163 I'm not quite sure what you mean.

[@taoprompts]

It's a great way to visualize things from different angles!

[@cybertuli]

Nice tutorial o)

[@Realidyne]

Great video! Thank you so much. I'm assuming this works well with photorealistic characters as well?

[@taoprompts]

Yeah, I have a version of this guide that shows some more realistic characters inside: kzbin.info/www/bejne/a4ubl6p7mryZiMk

[@nifty7585]

That is so OP

[@roberthintz4017]

This is perfect!

[@JSnyder1946]

Great video! QUESTION: Comparing Kling with Hailuo, which one do you think is better? What are its strengths and weaknesses?

[@taoprompts]

here's a comparison video I made: kzbin.info/www/bejne/nmXQdHukgNl6r7M

[@thewebstylist]

Amazing Wooo hooo! 🎉🎄✨

[@taoprompts]

Happy New Year man 🎉🎄!

[@Caryson122]

Thanks Tao great video! Can I also use Luma to generate different poses with a first and final frame?

[@taoprompts]

I haven't tried Luma's image generator myself. Based on what I've seen, I think you can get different poses with it

[@LynneHand]

Thanks. I am going to try this with some woodland critters I created. :)

[@taoprompts]

I'm excited to see what you come up with 👍

[@crookedfilms4190]

What if we take the video shots that we generated of the character from single image and use it to train a Kling model. Then use Text to video in Kling using that model to generate scenes. Will that work?

[@agnieszkaczachor7181]

u are a pro

[@promptsdcom]

Soon we will be making own films on a prompt ))

[@taoprompts]

There's some Ai's that already generate super long videos!

[@ART-ificial]

Great video. Did you use Eleven labs for audio? How did you prompt for the girl's accent? :)

[@taoprompts]

yeah I used elevenlabs, the voice name is Nadya

[@ART-ificial]

@@taoprompts oh haha, thanks, I wonder if it's the same actress that stars in What we do in the shadows lol

[@morpheus2573]

Very clever process. Q: Why wouldn't you lip sync the character in Runway as a final step? Also... Could you just use a screenshot instead of extracting images via Ezgif?

[@taoprompts]

sure you can also lip sync in runway if you want. Screenshots work, ezgif is just easier for me to see all the frames

[@ChiragR-007]

U da best

[@jazzampa]

Hi! Does it work for realistic human faces? Also, which one is the most realistic for creating my own face consistent character? Thanks

[@taoprompts]

Yeah it works for human faces, this will work with your own face as well 👍

[@Fearflix252]

How could I do this for multiple consistent characters in one frame? I want to create a comic book

[@taoprompts]

There's no real way of doing that in OpenArt I have a guide for doing that using lora+flux: kzbin.info/www/bejne/nl6ahHx8gZuahpY Here's an old guide for comics in Leonardo Ai: kzbin.info/www/bejne/qHrWpqZqi7GJitk

[@jenniee_jq]

If during model training, I use images where the character is smiling, can I generate images of the character with different expressions after training?

[@playthisnote]

Have you went over runway custom characters and or items. You train with multiple images. And also there’s Luma Dream machine reference custom character with @ symbol to call them. After creation.

[@taoprompts]

Hey, I haven't tried runway's custom characters. I'll have to test out Luma's new feature sometime 👍

[@playthisnote]

@ Thx! That would be great to see! 💯✅

[@nicolasmarnic399]

Hi Tao! The trained LORAs in OpenArt can be downloaded to be used with other apps? Thanks!

[@taoprompts]

Hey, They can be downloaded. I'm not familiar with using them in other apps tho

[@JadeEmperorTales]

For openart what plan are you using?

[@taoprompts]

I have the Hobbyist plan

[@MabelYolanda-c9i]

The holsters are not consistent. How can you deal with that? Adding closeups to the Lora?

[@taoprompts]

For very small details like that, you can use image inpainting afterwards to try and fix it, or image editing like photoshop

[@fabiogiacometti380]

so grateful to have stumbled upon your channel! so inspiring, keep going!🫶🏼

[@taoprompts]

Thanks for the support man!

[@JadeEmperorTales]

Wow amazing

[@talesoftheheart8625]

Is there any free training model? Openart lets you tey it for free but you can't train the model :(. Great video❤❤

[@taoprompts]

The only way to do model training for free, is to run it on your own computer system

[@dwainmorris7854]

This needs to be in an all-in-one app you haveto go to four or five different AI companies no just get one consistent Character . The workflow is too long and you wind up spending too much money with all these different apps in order to create one image

[@taoprompts]

That would be convenient, although it's pretty amazing this is possible at all

[@AI.Finance.Tradecity]

i love the workflow. But, why not use midjourney to create characters? and is Hailuoai better than Kling's 1.6?

[@taoprompts]

Thanks! Midjourney's consistent character feature doesn't work as well as training a custom model. I think Kling's better overall, for this specific use case Hailuo worked better for me

[@mrdorich6996]

Please tell me, does Leonardo have the ability to make portraits of real people in certain poses and emotions (uploaded from your computer) in photos in different styles (comics, graphics, pop art...)? And is MIDJOURNEY capable of such an operation?

[@burskyozbekov]

and how would you do that with multiple characters?

[@factfusion4268]

Make image to video tutorial for vector graphics or cartoons.

[@jenniee_jq]

Hi Tao, if i don't wanna train a model with OpenArt, can i use all of the character's photos to make Cref in Midjourney?

[@taoprompts]

It won't work as well in Midjourney, although it is possible. You'll get the best results training your own model, regardless of what platform you use to do it

[@jenniee_jq]

@taoprompts thank you 🫶✨

[@gerrysohal5730]

what if we want to create realistic character, is it possible to make in I2V life or I2V 01. I want to get realistic characters with different poses. please reply

[@epheleo4128]

nice bro

[@Nader.subaih]

How to have multiple consistent characters in the same scene?

[@taoprompts]

here's a guide for using lora + flux kzbin.info/www/bejne/nl6ahHx8gZuahpY

[@julieplavnik1486]

Thanks for the tutorial! Does the size of the input image matter here? Should it be PNG or JPEG? I found that AI models struggle when it comes to high resolution images to train them. Sorry for the silly questions in advance, I'm a newbie in this whole AI generation thing

[@taoprompts]

I think both png or jpeg work. As long as the size is larger than 384x384 it should work, the AI probably downsizes the images before training anyways

[@JoeTigger]

Sorry to bother you Tao but with your grinch how did you make him lip sync. I want to create a music video and have the song but unsure of lip syncing. Did you just write lip syncing: and the words you wanted?

[@BobGrantTM]

Do you have a tutorial on how to make consistent characters for an Instagram influencer?

[@taoprompts]

Not specifically for instagram, although these techniques will work for different types of characters

[@timhendson2550]

Hi do you have an email I am trying to use the same different topic but I want Pixar style but I am not getting the same output? Could you please help?

[@taoprompts]

Hey, you can always dm me on instagram. To get the pixar style, make sure your original character image is clearly in pixar style

[@pdjay8912]

일괄적인 캐릭터 만들기 진짜 어려운 것 같아요. ㅜ

[@taoprompts]

it will get easier over time!

[@damens-xn9pu]

Why open art and not MJ?

[@taoprompts]

MJ's consistent character feature doesn't actually train a "custom model", which means it will not work as well as these types of platforms.

[@SufyanBloch-vy7ps]

Is this free?

[@ANDREJESUS985]

Yes bro

[@jerrytrump8394]

But the openart are paid😢

[@FCTouhou]

Now it's time to have a tool to unify all the process

[@taoprompts]

yeah, that would make everything a lot easier!

[@FilmCartoonist]

Thanks I made a new channel a week ago and now I have 100 subs

[@pdjay8912]

kzbin.info/www/bejne/o5TKamCjrq6XoJIsi=hXInyt9clLMsAgvW 늘 좋은 정보 감사합니다. 제가 AI로 뮤직비디오 직접 만들었어요. 감상해 주셔요. Thank you always for the great information! I created a music video using AI myself. Please take a look!

[@fernandoferreira2273]

Your video is really great! Congratulations! I'd like to know how you made it.

[@taoprompts]

뮤직비디오 정말 잘했어요! 정말 공을 많이 들였나봐요, 정말 잘 만들었네요 👍!