Oh man this is a gold mine! Specifically consistent character on replicate, I've been trying to do this in Flux and ComfyUI for weeks and never got far...
I used this replicate space to generate images for lora training some weeks ago in production! It’s really awsome. Regarding the short, I’m in, you sold me the pitch’s show 100% haha. At first I was really bored by the models aesthetics, but then saw the satire. It really made me laugh, specially the Zeus part. Well done!
As a professional editor for much of my career, it's interesting to watch how people are putting together sequences today in AI. It's basically a reverse workflow. Used to be we'd edit a scene with footage from the set with lots of problems. Actors on a stage with maybe a green screen behind them or nothing at all. We'd edit the scene based on their performance, imagining in our heads how it might eventually look. It would often be months before the special effects team would put them in the appropriate setting, and then things like color timing happen. Now it's like the color timer and visual effects department gets hired first, and then you try to find a place to stick in the actors, and if you are really skilled, you can figure out a way to make them actually talk!.
When I us AI, I often don't have a vision, just some sort of "vibe". It's very iterative where I give generators an idea, then iterate and refine what they give back, sparking new ideas and direction until I'm happy with it. I think it's a great way to work for those of us who may lack a grand vision
I have been editing, doing VFX and everything else for a long time. I have tried AI too. The reverse engineering actually kills the story flow. Just stitching visuals together is not editing. AI doesnt exactly obey you. It breaks all the rules. No rule of 180. The character at first is left facing. In the next its right facing. Thus breaking the context, the story. Very little continuity.
@@phill_ai Thanks for liking. AI today is a chaotic marketplace. Aspiring content creators, established artistes, professional studios and everybody else in-between seek a slice of that pie. Its democratised. Whereas, film production demands novelty. With the rise OTT, theatrical productions are anyways downhill. I am not sure if the big daddys of film production will embrace AI the same way. They will only do that that AI currently cannot do. They may use AI only as a low level tool here and there. Meanwhile, novice content creators will saturate the AI market breaking all rules of filmmaking. That will become the norm. And the world will embrace or get used to it. SD resolution ouput and subsequent upscaling is not the professional way. We all will have to wait for AI to mature. Companies like Runway are not familiar with real production needs. They are learning too. As of now, I fix my Ai problems by reprompting/rephrasing a million times untill I get it perfect. Iinfinite patience is the key. Whatever remains unachievable, I try workarounds, green screens, working on select regions of the frame, using alternate tools or edit stuff manually using traditional compositing software. I guess every professional is using a similar multiple workflow. Its definitely not as simple as joing clips back to back.
It looks great in the same way the Super Nintendo looked "realistic" when the console was released. It will be a joke in 20 years but today it is simply AMAZING!
wonderful tutorial bro, spectacular workaround for consistent characters. Just train a lora for the liked character and continue generating consistent character images in different settings. Bravo !!
The trouble with Kling is that the lip sync is not very good. Runway Act One is excellent, but as I show in my test video (with two simultaneous lip sync characters) and you have shown here, you can't (yet) do whole body movement. It's all rather frustrating at the present time.
Thank you for your invaluable lessons; they have provided us with so much insight! I recently launched my own TH-cam channel where I share AI-generated movies. I'm curious to know if it's possible to achieve consistent character portrayal with Leonardo, or if there's a way to train the model for that purpose. From my experience, using a variety of tools is essential, as relying on just one may not yield the desired results. While Runway has made significant improvements, I've found that Hailuo can sometimes outperform it in terms of movement, despite being slower. When aiming to post twice a week, it's a constant balancing act between perfection and timely output! What tools do you recommend for achieving quick yet decent results?
I Appreciate so much that you share such valuable knowledge, its really helping me to fast-track in to the world of AI film making - my head is bursting with ideas, and can finally start to bring my film to life that I've been storyboarding for a few years now!
I'm an award winning filmmaker. My wife & I have written 4 Season One 6 episodes Limited Series scripts. Unable to submit our scripts to anyone due to the "not accepting unsolicited submissions" Hollywood rule, I am seeking an AI savvy co-producer-partner to produce the shows with AI. Will there be a market? Gonna happen sooner or later. Can you recommend someone? Thanks.
This is a prime example, why I still am not buying it. The example is such a horrifying uncanny valley... Hopefully, given a couple more years, this might improve and "pass". But what I have experienced so far -- it is FAR easier to actually FILM necessary footage, with actual cameras and actual people. Current level of video/AI animation is perfectly fine for hobbyists, and Facebook bots. Kinda similar to the "collages" with roses and pasted faces my great-aunt sends to all family members during holidays.
Oh man this is a gold mine! Specifically consistent character on replicate, I've been trying to do this in Flux and ComfyUI for weeks and never got far...
💚
I used this replicate space to generate images for lora training some weeks ago in production! It’s really awsome.
Regarding the short, I’m in, you sold me the pitch’s show 100% haha. At first I was really bored by the models aesthetics, but then saw the satire. It really made me laugh, specially the Zeus part. Well done!
As a professional editor for much of my career, it's interesting to watch how people are putting together sequences today in AI. It's basically a reverse workflow. Used to be we'd edit a scene with footage from the set with lots of problems. Actors on a stage with maybe a green screen behind them or nothing at all. We'd edit the scene based on their performance, imagining in our heads how it might eventually look. It would often be months before the special effects team would put them in the appropriate setting, and then things like color timing happen.
Now it's like the color timer and visual effects department gets hired first, and then you try to find a place to stick in the actors, and if you are really skilled, you can figure out a way to make them actually talk!.
That is a really interesting take ! A reverse workflow ! Awesome thought
When I us AI, I often don't have a vision, just some sort of "vibe". It's very iterative where I give generators an idea, then iterate and refine what they give back, sparking new ideas and direction until I'm happy with it. I think it's a great way to work for those of us who may lack a grand vision
I have been editing, doing VFX and everything else for a long time. I have tried AI too. The reverse engineering actually kills the story flow. Just stitching visuals together is not editing. AI doesnt exactly obey you. It breaks all the rules. No rule of 180. The character at first is left facing. In the next its right facing. Thus breaking the context, the story. Very little continuity.
@@SantanuProductions Very interesting I love this conversation. So how would you fix it?
@@phill_ai Thanks for liking. AI today is a chaotic marketplace. Aspiring content creators, established artistes, professional studios and everybody else in-between seek a slice of that pie. Its democratised. Whereas, film production demands novelty. With the rise OTT, theatrical productions are anyways downhill. I am not sure if the big daddys of film production will embrace AI the same way. They will only do that that AI currently cannot do. They may use AI only as a low level tool here and there.
Meanwhile, novice content creators will saturate the AI market breaking all rules of filmmaking. That will become the norm. And the world will embrace or get used to it. SD resolution ouput and subsequent upscaling is not the professional way. We all will have to wait for AI to mature. Companies like Runway are not familiar with real production needs. They are learning too.
As of now, I fix my Ai problems by reprompting/rephrasing a million times untill I get it perfect. Iinfinite patience is the key. Whatever remains unachievable, I try workarounds, green screens, working on select regions of the frame, using alternate tools or edit stuff manually using traditional compositing software. I guess every professional is using a similar multiple workflow. Its definitely not as simple as joing clips back to back.
This is the first AI video I am genuinely enjoying (and not for the wrong reasons). Expert use of our current technology and limitations!
I'm genuinely impressed
Totally mind blowing ! Impressive work 👏
It looks great in the same way the Super Nintendo looked "realistic" when the console was released. It will be a joke in 20 years but today it is simply AMAZING!
Great video. You make sense of things that, as an absolute novice, are at times overwhelming. Thank you. God bless.
Thank you 🙏🏻
The video is awesome, my friend! Congratulations.
💚
wonderful tutorial bro, spectacular workaround for consistent characters. Just train a lora for the liked character and continue generating consistent character images in different settings. Bravo !!
Your AI videos is the best I've seen so far. Great job!
💚
Very cool! Thanks for the detailed video and workflow, I shall follow it closely!
💚
this turned out freaking amazing man ! I only guess the patience of all the generations ! Good work man!
💚
That's actually pretty dang slick! 👍👏👏
The Short film is just incredible 💯 and funny 🤣
Wow pretty good. I would watch that Netflix show.
💚
Wow amazing job, thanks for all the tips and ai platforms!!
Thanks for always making awesome AI tutorials! Has helped me so much!
Thank you for generous tip 💚
🙏🏾 Great film and tutorial, Many Thanks, Brother!!
thank you for posting this. it is greatly appreciated!
Great work, both the movie and the turtorial
💚
Great video and great shorts.👏👏👍❤
AI is improving
Brilliant, great work!
Thanks a lot, this will help a lot of AI Artist and Content Creator out there
Awesome !!! Well done
Nice work. How do you deal with music for the whole movie and the Copyrights?
Amazing, you are a genius
Why did your Zeus character keep changing looks yet you others stayed perfectly on point?
Amazing!
we have come a long way already with AI.. but there is still a WAY LONGER way to go still until this will look convincing
25 dakika nasıl geçti anlamadım. harikasın 🙏😇
The trouble with Kling is that the lip sync is not very good. Runway Act One is excellent, but as I show in my test video (with two simultaneous lip sync characters) and you have shown here, you can't (yet) do whole body movement. It's all rather frustrating at the present time.
Thank you for your invaluable lessons; they have provided us with so much insight! I recently launched my own TH-cam channel where I share AI-generated movies. I'm curious to know if it's possible to achieve consistent character portrayal with Leonardo, or if there's a way to train the model for that purpose.
From my experience, using a variety of tools is essential, as relying on just one may not yield the desired results. While Runway has made significant improvements, I've found that Hailuo can sometimes outperform it in terms of movement, despite being slower. When aiming to post twice a week, it's a constant balancing act between perfection and timely output!
What tools do you recommend for achieving quick yet decent results?
Very soon Hollywood is going to be a dinosaur looking for the nearest tar pit, game design is going to be revolutionised, a lot happening!
Another question. While the character is talking, the rest of the scene is static. How do you solve the movements in the background?
amazing
do you use your own voice and then choose a voice on runway?
3:28 those hands would give you nightmares 😅
Amazingly beautiful work❤
💚
Awesome work cyber jungle!
💚
Hi, Is the Macbook pro M4 10 core enough to do this quality of AI Videos locally? Or would I need the M4 pro, 20 core GPU?
This just reminded me Netflix cancelled KAOS😩
Thanks!
I Appreciate so much that you share such valuable knowledge, its really helping me to fast-track in to the world of AI film making - my head is bursting with ideas, and can finally start to bring my film to life that I've been storyboarding for a few years now!
bu are kling allowed for commercial use?
ERES E MEJOR THE BEST!!!!!
Congratulation My Friend, Winer Winer
Was it expensive to create it? I mean each rendered video costs credits, so every failed try costs money as my understanding.
❤❤🎉🎉WOW 🎉🎉❤❤
Love it lol
Noice! 😍
Thanks:)
💚
The blonde character looks exactly like a coleague of mine
even with this professional production.. there are is no character consistency.. ridiculous
I'm an award winning filmmaker. My wife & I have written 4 Season One 6 episodes Limited Series scripts. Unable to submit our scripts to anyone due to the "not accepting unsolicited submissions" Hollywood rule, I am seeking an AI savvy co-producer-partner to produce the shows with AI. Will there be a market? Gonna happen sooner or later. Can you recommend someone? Thanks.
latina soundtrack for a greek mythology movie ?
haah that also he should have let Ai decide
Which application?
Name of the application?
This is a prime example, why I still am not buying it. The example is such a horrifying uncanny valley... Hopefully, given a couple more years, this might improve and "pass". But what I have experienced so far -- it is FAR easier to actually FILM necessary footage, with actual cameras and actual people. Current level of video/AI animation is perfectly fine for hobbyists, and Facebook bots. Kinda similar to the "collages" with roses and pasted faces my great-aunt sends to all family members during holidays.
I really want to know what is a cumbia doing in the middle of this greek gods corporate setting 😂😂😂
Where are you from man? Your accent sounds familiar 😅
I’m from Türkiye
🐺🐺🇹🇷🇹🇷😻😻
@@cyberjungle I knew it 😂👍👍
The original superheros
bro, make a new Terminator or another season of the Simpsons 😛
I wish these platforms were cheaper.
7:18 what to name that transition ?
footagecrate.com/video-effects/footagecrate-leaf-fan-out-transition
❤
lets be honest.. it looks amazing.. but it also still looks shit... AI still cant draw fingers and doesnt know human anatomy
Ok they missed up fingers, but you'll get a satisfyingly long video of 5 seconds for just $20/month. We are living in the age of AGI
@@Rumble2024injungle
we're not, but we will
Athena looks like Amber Heard
Ares looks like Leonardo Di Caprio
👋 hi
Hi Louis 👋🏼
3:31 ... cancer de tetilla !!!
It's all Greek to me
ahahha olm çok iyi lan
ahh, that's paid method replicate isn't working this model in free plan
Hollywood is sooooooo death!!! 😂😂😂
That was cool. Thanks for sharing. Much love. God bless.
💚
thing is the mouth movement still look off which makes it still not useable for high quality stuff
muito legal da hora .