What AI filmmaking actually looks like.

so let's talk about the future rapidly approaching a new era of film making one where anybody will be able to type in a sentence of what they want to see and out of thin air AI will generate that video so what does that mean for us camera jockeys what does it mean for narrative filmmakers that want to go out and actually film their stories well to Brave New worlds let's hear it straight from the horse's mouth chat GPT it appears to be down right now good omen for us now if you have been following the progress of all this AI stuff you know first there was AI still image generation that was crazy could go on Mid journey and make portraits of your cats that look photorealistic like this which obviously I did then Google announced the first AI generated video where you could just type in a prompt and it would spit out some really janky looking video like this but it was still mindblowing that you could do that but what I and I don't think anybody else expected was that just 14 months after the publication of that research Sora would be announced with which is open ai's new generative video AI which is a total mic drop gamechanging moment for anybody that works in film or any kind of video really just from Simple Text prompts it's able to spit out video like this I mean you're getting photo realistic cinematic shots 3D animated characters up to 60 seconds of video for a single prompt I think anybody who looks at this it certainly gets their mind racing especially if you're a narrative filmmaker even more so because Sora recently started courting Hollywood they came out here started talking talking to production companies agencies directors and trying to get them on board with this new software and Technology what do we get from that well they actually just released some short films that filmmakers who got Early Access to Sor were able to create all of these films were made purely from typing in text prompts into Sora taking that video and then they did traditional post- production on the videos this one with the balloon headed guy clever idea because that plays to the strengths of Sora while minimizing its weaknesses so you get to replicate the same actor very easily because it's just a yellow-headed balloon guy so you don't have to worry about him looking different from shot to shot and you don't have to worry about his lips being in sync with any dialogue both because there is no dialogue and because he doesn't have any lips so that's a great use case for Sora right now obviously as the technology improves even these things aren't going to matter at all now unfortunately Sora isn't available to us right now I wish it was i' love to play with it but I think it's a good time to start making some estimates about what's about to change in film making for everybody I've been thinking about this a lot and I started playing around with the other AI video and photo tools that are available right now just to get a better idea of what we can do and here's what I'm convinced of right now I think screenwriting and acting are both safe for quite a while they will be affected but not in a meaningful way unless maybe you are an actor that does a lot of commercials and branded content not so much for narrative narrative work though and that's really what I'm going to be focused on in this video narrative stuff you just can't get the nuance and level of organic connection to an actor with an AI generated actor no matter what the AI is doing there's always going to be that little bit of uncanny valley the actor is what we really connect with in a film they are the face of the story even in an animated film that's voice acted by real people a lot of times they're using motion capture to capture the emotion and the expressions of real people too so there's a lot to be said for using real people to bring your stories to life and same with screenwriting I've tried many times with jet gbd just playing around seeing if I could actually set this thing up for success to give me a decent script it has failed so miserably every time and I think the level of again nuanced required for a decent story and screenplay is just way too high for chat GPT right now now in the future obviously things are going to get better they're probably going to have better prompting but the problem is always going to be the same it's always going to be really formulaic the charm of a story is that it can surprise the audience it can Delight the audience and when an AI is trained on everything that all the other stories that have worked then you're always going to be kind of one step behind from the human who's able to observe not only those stories but also what AI is creating and just the you know the human emotions and insights that drive every story who knows that better than us right but let's talk about where AI is going to start being effective I mean editing it's already making a little bit of Headway there's some AI tools right now for editing talking head videos like this that can even do some assembly edits for podcast for example it's pretty wild it will just automatically start cutting it up but I think what you're going to see that's even more useful for narrative filmmakers is probably the ability to automatically swap out and audition different takes for certain lines for different shots if the AI is able to just automatically log what each shot is where it is in the screenplay that's an incredibly useful way to speed up and remove a lot of the tedious work that goes into narrative editing do I think that you'll be able to use AI to just edit an entire narrative short film or feature film no way it's the again the level of nuance is way too high you're not going to be able to get anything useful out of that for a very very long time in my opinion and I think editors jobs are very safe for the time being in fact if anything I think there might be more of a demand for editors in the near future because of all the AI generated video that's going to come out that is going to need to be edited to be useful which we'll get into a little bit later let's talk about the big one visual effects now integrating AI tools into a traditional visual effects workflow is going to save so much time doing tedious work I think there's going to be tools that allow you to animate way easier for example so instead of artists having to manually key frame animations of like the arms for a character swinging a bat for example if you're able to just tell the AI hey have this guy swing the bat really really hard and it can do it or at least get it 90% % of the way there and the compounding saving of time and tedious work that that will accomplish is huge so if you extrapolate that across an entire visual effects workflow I think what you'll get is not so much the elimination of visual effects jobs but instead the raising of the bar for what visual effects we think are acceptable and impressive so in a blockbuster film you're going to have even better visual effects and on Lower budget visual effects heavy films you'll be able to accomplish way more with a really small team or even one visual effects artist because of all the AI assistance that's going to be coming available now if we want to completely replace visual effect teams with just AI I think that is going to be much harder and take a lot longer to happen here's the reason right now with any kind of AI generated anything it's really easy to make something impressive right so if you type in I want a dog made out of mirrors on fire jumping across a crater on the moon and that is so cool that you can get that but what you can't get is a ton of control over exactly how that's going to look it's nowhere near the same as having an actual visual effects artist working on creating that shot for you where you get total granular control of every aspect of the shot and on big budget projects on big narrative films that is exactly what you need so despite the speed of all the advancements we're seeing I still think we're going to be a really long way away from being able to just go make Lord of the Rings or Star Wars in your house using AI video technology on the other hand I do think if you have the patience and a little bit of compositing skills a little bit of after effect skills to manipulate what the AI spits out then you will be able to open up a ton of doors to telling lots of different types of stories here's some use cases I keep thinking about creating crowds of people background Talent battlefields that type of thing is going to be much easier using AI generated video and that's exactly the type of thing that lets you tell bigger scale stories way way cheaper cuz extras are really expensive and difficult to manage another thing is already you're able to do things like set extensions with still AI generated art or you can do individual elements that you can then composite into the video for example if I wanted you know a plant back here I could just type in on Mid Journey so if you have a little bit of post-production knowledge a lot of patience you're going to be able to create individual assets way way easier and customize them and put them into your video however you want another big game changer for lowbudget filmmakers is being able to create creatures or different types of characters entirely just by doing motion capture just filming yourself acting out the scene and then using something like what Wonder Dynamics is trying to do right now although it's very janky right now to replace an actor with a creature like this and you can make that whatever creature you want using AI video or how about taking any film that you make and essentially putting an AI Snapchat filter over the whole thing to make it look more more cinematic with more dramatic lighting make it look like 1990s film stock even though you shot it on your iPhone and now suddenly your iPhone film looks like its speed sir I need to take your phone take the phone or taking your film and say make this whole thing underwater now you have your Atlantis epic or make this whole thing rotoscoped like a scanner dark L the possibilities are kind of endless with this lane and then the process of making this video I wanted to learn about where the state of all this technology is so I tried a few tools I shot these two shots just using the free Blackmagic app on my iPhone and I was like okay maybe I can do a AI generated still backgrounds on Mid Journey so I did that I created this Arctic futuristic ice planet with like a rocket launch pad out there and I tried plugging that background into some other tools like hyper and runwayml and animating those images into something cool so that I would have a actual AI video going on but the results I got were so janky and useless that I said we're not there yet until sor comes out in my opinion this stuff is basically useless but the still AI art is actually quite impressive so what I did was just TI the backgrounds and then composited some stuff over it using the sponsor of this video motion array motion array is kind of a big One-Stop shop for video assets whether you're talking about templates Motion Graphics sound effects music stock video all that stuff is super useful when you're creating something like this little sci-fi scene that I'm doing here so I was able to find this stock video of a rocket ship taking off throw that into the frame feather out some of the edges of the smoke make it a little bit blurry to match do the color correction on it but since this is a still image background what really gives it away is the lack of motion so let's add some more Motion in using a overlay of some snow particles falling put that behind the subject in front of the subject so it looks more immersive of course motion Ray also has us covered on music so we can grab some music throw that in they also just release their own plug-in Hub so that gives you access to over 50 plug SK that you can use for your favorite editing software like Premiere After Effects resolve and the cool thing about the Hub is each effect has its own little tutorial on how to use it which is really handy so by adding some of these effects I can spice up this footage even more like this and to cap it off let's use a cool sci-fi animated title for motion array too this one should do the job motion array Hub lets you add all kinds of visual effects color grading moving green screens all kinds of stuff to your video so easily so big thanks to motion Ray for sponsoring this video they're my favorite ones web shop for editing assets and you can get $50 off a year of motion Ray using the link in my description now let's see how this little video turned out there's also some big issues with AI things that have made it very controversial for one there's like no transparency of where all of the data for these AI generated videos and images are coming from we all kind of know they just scraped the entire internet for all the data so everything on Google image search got snatched up and logged and categorized and that's what they used to know hey this is what a duck looks like hey this is what Lois Griffin getting railed on Deviant Art looks like and if anybody types in something on Mid Journey they'll be able to recreate something that is pulling from those Source images problem with that is these artists didn't give consent for their original work to be used for this new work and if you're using music as an example when artists use samples of older songs in their songs they usually have to pay like a royalty or a license fee to that original artist to reuse their work but AI art doesn't have to do that and that's kind of a annoying thing I mean it kind of De incentivizes humans from creating new cool stuff when we all know that some d futuristic machine is going to suck it all up and then any [ __ ] with an internet connection can spit it back out and say that it's theirs at the very least there should be a way to opt out of your original work getting pulled into this machine's data set so look the ethical concerns are valid but what's not worth pushing for is just sticking your head in the sand and pretending that none of the stuff exists or just shunning it entirely this industry is going to be turned upside down soon enough why help it why not be prepared you can't close Pandora's Box these tools are out there that means that anybody who refuses to use them is just going to be handicapping thems and eventually they'll be outcompeted by those who adapted to the new tools at our disposal to me it's not that much different from when we went from film to digital there was a lot of change but in the end it ultimately created more opportunity for more stories to be told and I would argue more jobs created than lost so I think the best way to approach these new tools is Pivot adapt and use this as an opportunity to collectively raise the bar for our work rather then just Embrace AI induced laziness or mediocrity and cynicism and sure there's going to be some of that but that makes it all the easier for the rest of us to Stand Out by putting in the same amount of effort and will as before while working with just an improved tool set I don't believe AI video will ever devalue human film making just like every technological advancement in the past didn't even though it seemed that way at first particularly when it was abused by the early adopters no no we we just keep shooting this is video where we we shoot and shoot and then we deal with it later humans intuitively want to support the efforts of other humans over soulless machines and computers we love watching Magnus Carlson play chess even though we all know he'd get his ass kicked if he played a computer but we don't care in the slightest about watching the far superior computer play chess so despite all the issues with this stuff I for one am beyond excited to be able to bring some otherwise Impossible film ideas I have to life right here in my office and take the films I can go out and make traditionally to a much higher standard by using these new tools we will or will not continue this conversation [Music]

Channel: Standard Story Company

Views: 16,362

Keywords: generative ai, sora filmmaking, ai filmmaking

Id: uhTCLS2bnis

Length: 15min 16sec (916 seconds)

Published: Wed Apr 24 2024