Luma Dream Machine's NEW UPDATE Makes Hollywood a Thing of the Past!

[@theaifilmmakingadvantage]

🕐 TIMESTAMPS 0:00 The Update That Changes Everything 1:42 Navigating the New Dream Machine Interface 3:15 Understanding Boards: Your Living Storyboard 4:38 Building a Sci-Fi World (Live Demo) 7:23 Character Creation & Customization 9:45 Weapon Design & Props Generation 12:30 Vehicle & Transportation Design 15:20 Starting a New Project From Scratch 17:45 Advanced Tips & Creative Possibilities 19:38 Integration with Other AI Tools

[@QuizmasterLaw]

Your voice is good. You speak well. Good narration. Quality.

[@theaifilmmakingadvantage]

@ appreciate the awesome feedback! Looks like that investment in my Shure SM7b many years ago continues to pay dividends! 😁

[@FuturisticAgent]

Don't change a thing including the thumbnails, your awesome! 😎

[@tobiasfalk9112]

Excellent video, really looking forward to your upcoming coverage of Luma.

[@CodexPermutatio]

Realism in everything concerning the human figure is becoming better and better. Look at that beauty walking on the beach! (at 0:10) However, even if tools like Luma or Kling are great, the real revolution will come when we can locally generate video with this quality (just as we can already generate images locally with incredible quality). As artists it is crucial to have control over our means of production. That is why I say that local video generation, using our own machines and programs that we can modify, is very important. Anyway, this is unstoppable.

[@DanTheVeterinarian]

Thanks for your help! I have been struggling with the update.

[@bobicus]

Great vid! Just subbed, i tried Luma ages ago, but looks like they have upgraded. Thanks for sharing, hard to keep up with everything going on. it is exciting times

[@theaifilmmakingadvantage]

Thanks for watching, subscribing, and for the AWESOME comment!

[@CameraGodz]

I can’t believe I watched that whole thing just for you not to even animate the pictures into video 😂

[@theaifilmmakingadvantage]

@@CameraGodz dude, it was just a dry run of the Luma AI update. I even said multiple times that the project was just an example of what you can now do in Dream Machine. I’m not going to waste credits to generate videos I never plan to do anything with. But I AM going to create a separate video for a project I DO plan to use for AI filmmaking

[@CameraGodz]

@ it’s all good brother I love your videos, just sayin 😂

[@theaifilmmakingadvantage]

@CameraGodz thanks my man! I’m still learning as I go! If people want to see videos generated in Luma, then I am going to give it to them! 👌

[@leonwinkel6084]

It would be way more interesting if you show more of what comes out if you change stuff etc. just showing the menu is not that entertaining. Esp converting stuff to videos. That’s why I came here 😅 and there is barely video stuff to see

[@BeautifulDigitalArtChannel]

Right where is the video.? I came for that as well. good job tho. but still..

[@theaifilmmakingadvantage]

That’s why you should have watched the video from beginning to end, because you would have seen me do exactly that multiple times 👀 Welcome to KZbin! We are NOT TikTok buddy 😂

[@theaifilmmakingadvantage]

Thanks for the feedback! I guess I can make a separate video on that. But generating videos is such a basic thing at this point that I didn't believe it would add anything to the talk about the new features (you could always generate videos with Luma Dream Machine. So nothing changed in that regard). All of the video clips at the beginning of the video were generated by me with Luma Dream Machine, FYI.

[@IntuitiveGoat]

Do the characters stay consistent from scene to scene?

[@theaifilmmakingadvantage]

If you need consistent characters, watch this tutorial I did the other day: kzbin.info/www/bejne/q5zEfIWMo8p7eaM

[@johndiggs8227]

I didnt see anything move in motion. Can you show us?

[@theaifilmmakingadvantage]

@@johndiggs8227 Hey John! All of the videos at the beginning of this tutorial were from Luna AI. But I think I will also record another video walking through the image-to-video tool and upload it either tomorrow or Saturday

[@FilmSpook]

👍🏾👍🏾 You've got one of the BEST A.I. channels, being that you're actually a Real Experienced Filmmaker. Thanks, and keep up the Great Work, Bro. Carlos

[@theaifilmmakingadvantage]

Appreciate that Carlos! 🙏👊

[@FilmSpook]

@@theaifilmmakingadvantage 🙏🏾 Very Welcome, Brother, and Happy Thanksgiving to you and your Family. 😎

[@theaifilmmakingadvantage]

Happy Thanksgiving to you and yours as well! 🦃

[@bcraigcraig4796]

Hi I like Your Channel really great stuff maybe you can help me I have a spy character I created in MJ and I did a training with her over at Krea AI now I see Luma AI now has a Character Reference is it any good? here my problem I took 5 single screen shot from a movie trailer so I have them act like a storyboard now I want to regenerator each of this 5 images that make up one scene sequence but with my character instead do You know of a good workflow for this? Thanks

[@theaifilmmakingadvantage]

Thanks for the comment! The only thing I can think of at the moment is you can take your character into something like Canva, remove the background in her photo, download her image with the background removed as a transparent PNG, then upload your 5 single screenshots into Canva, upload the transparent PNG of your character and add her to those screenshots, then download the photos and do image-to-video on them. As for Luma AI's character reference, it is good, but if you are looking to make a consistent character, then character reference is not what you want. Character Reference will only generate an image with the "likeness" of your character. If you are trying to make a consistent character of your spy character, then watch the video I created on how to make consistent characters here: kzbin.info/www/bejne/q5zEfIWMo8p7eaM Hope that helps!

[@PassionBrains]

Looks like Google Labs will open their tool very soon

[@theaifilmmakingadvantage]

@@PassionBrains Veo 2 is text-to-video only. It might be good for some great looking b-roll, but completely useless to narrative filmmakers

[@mrhmakes]

very interesting! I came across luma labs a while back (well about 6 months ago = eternity in gen AI) as i was looking at some txt23d with "Genie". This storyboarding brainstorming technique is super useful. From your replies below I see they haven't yet cracked the character consistency thing and you have to do a bit of photobashing externally to get custom (face) characters in. OK my main questions are: 1) from your experience how does their video gen compare to say Kling and minimax? 2) How is Luma with custom styles - can you just prompt them in "in the style of an oil painting"? At the end of the day the storyboard feature for me is worth the price of admission! A solid addition to the workflow!

[@theaifilmmakingadvantage]

Appreciate the comment! 1) As of this comment I would say it is ahead of Minimax/Hailuo and definitely below Kling. Luma also generates video the fastest out of any on the market. Hailuo has really been hit or miss as of late. It was better back when I started AI filmmaking in early October. 2) I have not used the custom styles feature with Luma just yet. Honestly, I have all of my scenes looking exactly how I want them (via Freepik, Krea, or Midjourney) before I bring them into someplace like Luma. Their new Photo feature seems promising, but I am not going to spend a lot of time generating photos in Luma other than for storyboarding purposes

[@mrhmakes]

Thanks for that! Will have to check out krea and freepik as well

[@DjGolucky]

you are very informative, but most of things seem to be coming soon. its not at the moment making Hollywood a thing of the past. But I appreciate the video, as you make it so we know exactly what we are dealing with.

[@theaifilmmakingadvantage]

The thing about AI is that it used to be that this stuff was evolving by the month. Now it seems that things evolve by the WEEK! Those who are early adopters will be light years ahead of those still patiently sitting on the sidelines waiting for the tech to be perfect enough for them to finally jump into the fray.

[@dwainmorris7854]

yeah but you still cannot tell consistent stories with AI still long way to go for it and put Hollywood out of business right now as this you can't even Tell a consistent Marvel or DC style comic book story with Ai

[@theaifilmmakingadvantage]

Hollywood will put themselves out of business. Since they are handcuffed by the trade unions and guilds, they will fail to innovate while tech and AI make MASSIVE strides literally every single WEEK! And you'd better believe Hollywood is afraid of the rise of the AI Filmmaker. Just take a look at the Ben Affleck interview I posted on my X account. And we now have consistent characters almost down to a science. So the only thing keeping you from telling consistent stories are your limiting beliefs that it cannot be done!

[@kbcproductions]

​@@theaifilmmakingadvantage what about consistent backgrounds and areas? If you have a character in a room you need the furniture to stay the same and remain in the same spot. I've struggled with this myself. It seems best to just keep driving your story to different places so it's less of an issue. I've also noticed when you put a lot of details in your prompt the AI struggles with processing that much information. You'll get some of what you asked for, but not everything or it will apply one thing you said to the wrong object. Maybe I just need to get better at writing prompts. Overall I believe you're right though. I believe that AI will get there sooner or later.

[@theaifilmmakingadvantage]

@ have you ever thought about getting all of your dialogue from one consistent frame? If you are having a two-person dialogue scene, there shouldn’t be any reason why you can’t reroll the video with all of the dialogue you need without having the background change. I’ll be demonstrating this in my own examples shortly. But it will be no different shooting a two-person dialogue scene with real people. You get all of Actor 1’s dialogue with camera A, and Actor 2’s dialogue with camera B

[@ielohim2423]

​@@kbcproductionsmy current workaround is that i use stills from unreal engine environments for the backgrounds.

[@INVICTUSSOLIS]

They still need to work out AIs way of manipulating HANDS

[@theaifilmmakingadvantage]

@@INVICTUSSOLIS it’s getting better! I’ve found that it may take a few rerolls to get your video just right.

[@INVICTUSSOLIS]

@@theaifilmmakingadvantage Very true, this is what Ive noticed as well but overall. Luma has been very innovative in the space. One of the best out there.

[@theaifilmmakingadvantage]

@INVICTUSSOLIS yes! They’ve gotten much better. When I first started AI filmmaking last month, they were at the bottom of the list for me in terms of quality generations. But as of the last few weeks, their generations have all turned out better than my Hailuo AI generations, which I used to rank above Luma Dream Machine