The Most Realistic AI Video Tool Yet!

it's been a historic week for AI Film Production welcome to AI film news first off I'd like to kick things off by saying thank you so much to the Swedish Film Institute and the national Film School of Denmark which we are in right now we had the chance to talk about artificial intelligence with artists from around the world at their AI film making events it was incredible to meet so many talented artists and a few curious Refuge students we'll let you know when the talks are online a few months ago at CRA the idea of gaussian splatting popped up as a white paper and while it seemed revolutionary at the time tools over the last few weeks have finally started unveiling the technology to the masses now you literally have the ability to scan any environment in the world and get photo realistic scans in return it's a huge update from a typical Nerf workflow this has two huge implications for filmmakers number one you can use these scans for a virtual production environment so if you are using the LED screens as your background you can actually transport your characters into a world that you have scanned it's incredibly powerful and the second is you could actually shoot your characters on a green screen and composite them into an environment that you've scanned so essentially this has dramatically changed the process of compositing subjects into a scene instead of having to shoot a video plate you can simply scan your environment and get a realistic environment in return since Dolly 3 came out last week people have been playing around with the new model from open Ai and the images it creates are truly mesmerizing dolly is really good at taking Direction and creating artistic style now I really think I should note though that dolly is not quite as is good at creating realistic cinematic imagery so while I think it's very helpful for art direction for marketing projects and anything that has an artistic and stylized design you still probably want to use mid Journey whenever you're trying to create a cinematic project and one hack that I saw from Nicholas newer this week was essentially you can upload your Dolly 3 image as a reference image inside of mid Journey so if you're having trouble adjusting your compositions and getting things right exactly inside of mid Journey you could go to Dolly 3 get your composition and use that inside of mid Journey it's a cool workaround that I think really highlights the best of both tools over on X able art put together a really interesting concept where they used Runway gen 1 in combination with a cardboard layout of their cinematic environment to essentially create multiple stylized iterations of the scene they were trying to create this really hits home on a really interesting point inside of AI Film Production and that is these tools will very soon be able to Simply upres and stylize on top of footage that you're feeding inside that has huge implications because essentially you can upload what would be basically a video game cinematic and get a very cinematic result in return while this technology isn't quite there just yet it's very easy to see how this technology May progress to where all you need is a lowres version of your film and then you can upres and stylize into a hyper stylized and cinematic result the team at likewise released a new chatbot called pick that essentially allows audiences to type in the type of film that they want to see and it uses artificial intelligence to give recommendations while we've had AI algorithms like the Netflix Auto recommendations for a little while now this new chatbot experience may be a helpful Innovation for helping audiences find new films like perhaps the ones that you create canva also released a ton of new AI features this week and we tested each and every one of them now I have to say that for most of them I don't see a very quick film making application but there are a few innovations that I really want to show you today so it all starts with magic grab so with magic grab essentially all you have to do is upload your artwork select the image go to edit photo and select magic grab it will use its AI algorithm to essentially grab the main subject from your image and then you can move them around the frame so now we've isolated our subject and we can move them around to wherever makes sense this has huge implications because as you know it can be quite difficult to Art direct directly inside of tools like mid Journey so now you can move your subjects around your composition with ease and it does an amazing job at cutting out your subject including their hair strands and it really just looks very professional the other tool that looks really interesting is Magic expand so essentially I have this image from mid Journey that I've imported into canva and I'm going to go ahead and select that image go to edit photo and we will go to magic expand now there's a lot of different settings that you can select here I'm just going to select the whole page and select magic expand so now just like most AI tools you get four different iterations that you can select from now I should note that this outcropping feature utilizes a technology that is not quite as good as photoshop's generative fill but it is a bit faster and sometimes working inside of canva is a little bit better experience uh if you're working on a project that just quickly needs to get done and so you can see that all four of these images here they look pretty good it's not quite the most photo realistic thing you've ever seen uh but I honestly think at a quick glance you would never know the difference so I definitely recommend checking out this tool they also announced a new text to video integration with Runway now whenever I was doing my quick test it seems like it's not quite available just yet but it will be incredibly powerful to have runways text to video platform directly inside of canva and I would love for us to be able to have image to video Direct IR L inside of canva as well it would be amazing to directly animate your story boards that you lay out inside of canva inside of the tool so if you're listening Runway or canva team please create that feature it would be amazing now those three updates are the biggest ones I think you should know about as an AI filmmaker but there are a few other tools that could be helpful for marketing projects or if you work in a professional environment so I'm going to go through a few of those features really quick so the first one is Magic rewrite essentially anytime you have text inside of canva you can go to magic right and you can click apply brand voice now canva will use AI language models with the information that you've pre-trained it on in your Brand's voice canva will then change the text to match your specific brand voice it's really powerful and it works incredibly well they've also introduced an automatic translation feature into what looks like over a 100 different languages all you have to do is go to magic Switch and Go to translate and then from here all you have to do is select any language you want or you could search so I'll say Japanese I'll select and now I'll hit translate so this is translating an entire presentation in just a few seconds now all you have to do is open your presentation and you can see that it has been translated into Japanese now you still may need to go in and change the font sizes because obviously VI ly different languages may take up more or less space but it's an incredible feature if you are working on a film pitch that needs to be in a different language now if you're working on a very quick marketing video you may want to have some motion design in there but you don't want to have to use an After Effects template or even create the animation yourself well canva actually released a new magic animate feature that may be super helpful essentially all you have to do is select your background go to animate and click magic animate canva will then interpret your entire scene and then give you some animation options for example when I select this professional option here you can see how the text animates on and it's all staggered a little bit so it doesn't animate all on at the exact same time it's just a very quick way to do animation if you're in a pinch and the last AI feature that I want to showcase inside of canva is actually their own image generation tool so whenever you are inside of canva essentially all you have to do is go to elements and go to generate your own so now I want to generate an image so this was an image I generated inside of mid journey I want to see what canva can do to create an image in the same cinematic style so I'm going to say a cinematic still of a woman reading the news paper in a Victorian library and I'm going to select to the filmic style and let's take this one step further and say dramatic lighting and create the image okay so let's take a look at our results so inside the thumbnail view here the images look pretty darn good and I would say this is comparable to about mid Journey 4 which is the version that we had back in February and in March now if we click in here and drag this up kind of move this over to the side you can begin to see how there's quite a bit of Distortion in the face and uh the newspaper here has quite a bit of distortion as well but generally it looks fairly cinematic so I would still say that mid journey is a much much better tool to use if you're trying to get the most cinematic image as possible but if you're in a pinch or if you really want to get a very quick image directly from canva you can do that using this tool not to be outdone this week Adobe also announced that they will be unveiling a few new tools at their Adobe Max event next week a few of these features includes the ability to have a magic grab similar to canva and also a magic edit which is also very similar to the functionality directly inside of canva but these tools will be using Adobe Firefly which is a much more powerful AI generation tool so I would assume that the results would look much better they also have this button that essentially allows you to select the frame and remove all distractions from the background which is incredible especially if you've ever been at Disneyland like we have and there's a million people in the background that you want to get rid of well instead of having to generative fill one by one you can slick get rid of distractions and everyone in the background will magically disappear they also unveiled a new feature that will essentially allow you to Art direct and select different aspects of your frame so in their demo they selected these subject shirt and completely changed it to a new jacket they selected the pants and art directed it to something different you can do that using generative fill currently but it seems like this workflow may work a little faster now the demo seemed a little suspect because I did notice that even in the marketing video there were some issues with the generated images but I'm sure we'll get a better idea of what they're unveiling very soon there's also this amazing art generation tool called Leonardo that has kind of been flying under the radar but it is actually a very powerful competitor to both mid journey and Dolly so I went to the four most popular image generation tools and prompted the exact same scene I essentially asked for a woman in a sci-fi film with green color grading so the first place I went was Dolly 3 and you can see we have these images here which don't look that bad but I'm not getting a lot of sci-fi or uh very much cinematic quality the next one was stable diffusion and we created some very interesting images but this woman has what looks like a nose mouth and this person has a very long neck but the frame does look pretty interesting now the next one is Leonardo and you can see that it looks really really interesting it has a filmic Vibe and it seems like it stuck to the prompt pretty well I think it could be an interesting alternative to Mid Journey if you don't like using the Discord experience but but finally I do want to show you mid Journey which really knocked it out of the park the quality from mid Journey was definitely better but I do recommend checking out Leonardo and seeing if it might have some interesting use cases for you on your video projects the team at genmo also released a brand new feature that essentially allows you to do really intricate looping effects using their tool I was playing around with this earlier today and it seems like it could be a very interesting tool for doing concert visuals or just any visuals that need to have a very interesting looping style I should also note that last week we announced our AI horror contest in collaboration with 11 Labs we've already seen some submissions coming through but you have until October 20th to submit your work the winner will get $11,000 a seat inside of the Curious Refuge AI film making course and over 2 million credits to 11 l not to mention a very beautiful tortilla blanket there's also a new AI video generation tool that looks very promising it's called Moon Valley so the tool is still inside of beta and it works inside of Discord but it's very interesting because the movements are very very smooth compared to some of the AI video generation tools that we've seen up to this point so in this video here we have Mark Zuckerberg punching Elon Musk with boxing gloves and you can see that it's looks more realistic than most of the other AI videos out there now I wouldn't say that either one of these people are Elon Musk or Mark Zuckerberg but it is very interesting to say the least and the motion is also a lot more stable so I saw this example from the community earlier today and it really showcases the potential of using AI for motion design and UI visualizations the quality of the render looked very very very interesting but the quality isn't 100% perfect just yet Dan inside the Curious Refuge Community prompted cat eating pizza and the result is this cat eating a pizza made out of other cats and it's kind of horrifying so I guess if you need some great content for our AI horror competition you might want to check out Moon Valley before they update their algorithm I should also note that the presentation that Shelby and I gave for outside the frame is now available on VI's website you'll find a link in the description of this video I also came across a very interesting tool this last week that I think could have a big impact for folks who work on documentaries it's called kits AI now the tool is typically used for AI music generation but what I think is really interesting for filmmakers is the vocal separator picture this someone gives you archival video footage of a person talking but it has music underneath now if you're working on a documentary having music on top of the music bed inside of your film can be a problem but if you could separate the voice and the music layer you could then comp in your own music and preserve the quality of your overall film and it's really easy to use this tool all you have to do is go to the vocal separator and simply upload the wave file or the MP3 or you can upload the YouTube track when you're done it gives you a file that essentially separates the vocals from the background music layer it's very powerful and hypothetically you could take that vocal track and train a voice clone algorithm inside 11 labs and generate a brand new voice there's also some interesting news from mid Journey this last week Ali Jones who keeps track of the meeting minutes for Mid Journey essentially said that pretty soon we will be able to save personality profiles inside of the tool for recalling essenti the styles that you use on a regular basis there will also be an image upsizer directly inside of mid journey to give you over 2K quality these features will be shipped more than likely by the end of October but what I'm really excited about is mid Journey 6 which will release in 4 to 8 weeks inside of mid Journey 6 will'll have better image quality better language understanding more resolution and bigger batch sizes that will be 2 to four times larger than the current batch sizes that we get inside of mid Journey which is 40 so we're looking at 80 to 160 image Generations inside of mid journey and remember each one of those image Generations gives you four Images so pretty soon you might be able to generate close to 500 images in just a couple of minutes using their brand new features Getty Images also announced that they will be releasing their own AI image generation tool very soon I'd like to take this time to thank every single person who joined us for our October session of our AI film making course this is our biggest session ever and we have artists from around the world who are looking to use AI on their film productions for the very first time we've updated the course with a ton of new content and we can't wait to see all of the incredible work you create a dental company put together in advertisement featuring Tom Hanks inside of their production that used artificial intelligence but here's the catch they didn't ask Tom Hanks to use him to endorse their product Tom Hanks was admittedly not very happy about it but he's Tom Hanks and so he wasn't super mad either this is just a lesson that if you're using someone's likeness to promote your product make sure you get permission chat GPT also brought back web search inside of their tool so now when you're inside of chat GPT you can search the internet and pull relevant data for example earlier I asked chat gbt to pull a list of some of the big films coming out this year and predict the box office performance it used chat gbt's algorithm with the data that it's getting from the internet and predicted how much each one of these films is going to make I think there's a ton of research potential in using this tool I also used it to scan the contemp orary news stories and try to figure out if any stories might make for an interesting film concept and it did just that it scanned the news and put together a list of ideas that might make for an interesting feature film there's a ton of use cases out there so if you found some clever ways to use chat GPT on your film projects be sure to let me know in the comments now it's time for AI films of the week so I want to kick things off with this film created by big films called blueberry pie essentially they used w Studio to replace one of their subjects with a robot and there's a lot of really expressive movement and I think they did a fantastic job with the film the second film I want to showcase is a music video called Earth by a band called The New Horizons they essentially curated this really interesting sci-fi film that takes place in space they essentially used mid journey and Runway Gen 2 to curate a very interesting sci-fi film that takes place on what looks like to be somewhere between Earth and an alien planet the shots look super cinematic and I think it's really a Showcase of what is cinematically possible when you use these tools I also want to showcase this incredibly creepy weird video that Shelby sent me this week of essentially a sushi experiment gun completely wrong it's just these really weird people who were like wearing Sushi and made out of sushi and I don't think I want to eat sushi for a while after watching this video but it's super super strange and for this week's student Spotlight I want to shout out ktia who put together a really interesting film concept for a film called exoskin she did a good job of curating her shots and telling a compelling story through the use of her curation inside of mid Journey fantastic work Katia I should also note that ktia was in the very first session of our AI film making course but actually just recently turned in her assignment number four which is the film that you're watching right now so if you're ever feeling stressed or like you can't follow along week by week inside the course no sweat you have plenty of time to turn in your assignment and get feedback from our filmmakers and our final bit of news today comes from the Department of Homeland Security who is actually using AI to help solve child abuse cases essentially they use AI with large data sets to help them scrub through a large amount of data and solve cases and that's it for this week's AI film news be sure to subscribe to our newsletter over on Curious Refuge if you want to get this news and more delivered directly to your inbox each week thank you so much for watching we'll see you next week

Channel: Curious Refuge

Views: 100,629

Keywords: AI News, AI Film News, Filmmaking News, AI Filmmaking, Artificial Intellegence, AI, Curious Refuge, AI Horror Film, YouTube AI News, Midjourney, Pika Labs, Pika AI, ai filmmaking tools, ai filmmaking tutorial, ai film, ai film news, ai, ai video, ai news, midjourney ai, runway gen2, midjourney, ai art generator, ai art news, ai filmmaking course, ai short film, ai tools, stable diffusion, caleb ward, curious refuge, Moonvalley, Canva AI, Adobe News, Adobe AI, AI Art

Id: boq2If6BT1s

Length: 22min 26sec (1346 seconds)

Published: Sat Oct 07 2023