Pages

Showing posts with label GPT-4. Show all posts
Showing posts with label GPT-4. Show all posts

Tuesday, August 13, 2024

In Flux


Been doing some thinking over the past two weeks. I spent much of July focused on the current state of AI video and was trying to determine its limitations and capabilities. That meant researching Gen-3, Dream Machine, and Kling. There are others but those are the main three so far. There is great potential. No doubt. But, to get the best results you need to use Image-to-Video, preferably with one or two images. I can see generating between a series of single images in a longer sequence as a thing. Having one at the start and one at the end of a 10-second clip is wicked cool, but I've heard of these tools being able to do 2-3 minute sequences. Imagine adding 120 single images and next thing you know you have a whole scene created without needing to be so tedious with these shorter clips. Not all the models are doing these beginning and end frames yet but they will soon. Oh, and FLUX is giving Midjourney a run for its money as far as realistic images. 

I love all of that. Imagining how it will progress is just as fascinating to me. It's amazing to watch. New use cases are rolling out every few weeks. For instance, you can shoot a video with your camera and take an image from that video to add a VFX sequence that can be edited into the live shot using editing software. After I saw that I realized I could go back and test on old short films I made years ago. I am discovering these brilliant techniques people are coming up with and testing them to determine possible use cases. 

Like many people who have realized over the past year that we have entered a new era-- one that seems likely to change society and the way we live our lives, I have been trying to determine how to pivot. While I am no longer a young man, I still have dreams and aspirations similar to those of that younger version of me who returned from California with a creative fire burning in his eyes 25 years ago. The stories are the key. They reveal the lessons learned along the way, and the possible futures based on the world as it is perceived.  

Give me an hour at a cafe with a cup of coffee, a good book, my phone, a notebook, and a pen and I couldn't be happier. It's a pocket of time when I am free to let my mind wander. While some of my best story ideas happen while I am out for a walk, so many of those ideas are fleshed out at a cafe. The constant change around me, people coming and going, as I sit there observing while looking inwards, making connections, recalling the journey, and trying to predict and plan for what comes next. 

I have given myself till the end of this week to assess AI Video models to see what I could learn and then determine how that might impact me creatively in the near future. I still have another week, and there are likely to be many new use cases to discover, but I feel I have made my mind up already.

I am not in this space to be the guy who is the first to discover new techniques with these GenAI tools. I do not feel obligated to post content every day to keep my engagement metrics up. No, not yet at least. I want to see how others are using these tools to learn from them so that I can tell my stories in new ways. I think we are in a new frontier-- creatively speaking, and I consider myself one of those pioneers. My goal is to create a multimedia company. Or a "media empire" as a friend recently joked after reading my most recent TV series bible. It lays out big plans for the TV series that involve a few new ways of interacting with the content. 

Simply put, the goal is to work with Gen AI tools to be able to do more. Two things came to mind last spring when I began to immerse myself in the Gen AI space: How do I use these tools to help me creatively? And, how do I use this technology to help others? 

I am by no means an altruistic saint who thinks only of helping his fellow man every second of every day. Far from it. We are a screwy species and it is often best to mind our own damn business. But I do come from a long line of educators so maybe genetically that is where it comes from. Anyway, an APP was one of the first things I thought of last spring after sitting down with GPT-4 for a month. I have been researching ever since. 

While I won't go into detail about the APP at this time, it is interesting that other than creating moving and still images to accompany my written words that I thought of creating an APP to help others. The idea just made sense. Even more so now. Not only can I help others with it but I can also help myself as well.

As I was assessing the current state of AI Video tools last week I realized something. 

If I am serious about starting a multimedia company, I can't expect AI video trailers/ short films, or graphic novels to fund the way forward... yet. AI Video has gotten a whole hell of a lot better than it was this time last year. However, it's not easy to tell a substantial story. And while the trailer I am working on means a lot to me, it cannot be my main focus. These tools need to get a lot better. Right now you need to be a patient and persistent puzzle master to piece together a worthwhile 2-minute trailer. You'll need to pay out your ears for all the tools needed to create something special. But it can be done. Within a few months, folks will start creating longer works where they have pieced together using the same methods from shorts to create something special. We'll learn their process and cringe at how difficult it was. And yet that will be the most difficult it will ever be. By this time next year, it will be so much easier to do all of this.

There is a window that has opened for AI Video creators and those like myself who are gradually learning more about it every day. The familiarity with these current tools and the proven results of using them may help during the big run to create content that will likely arise next year once AI Video takes its next big leap forward. That leap should be to provide the ability for these models to take a script from a scene, ask you questions about it to make sure it understands what you are wanting and then generate the scene. Once these models can communicate with us like LLMs do using chain of thought then we will see a massive explosion of AI-empowered storytelling. 

For the time being, AI Video is still too unstable, both with its outputs and the overall process. These tools have only become worth my time since June. Sora was in February, but that doesn't count because we still haven't gotten our hands on it. Again, I am not here to discover all the techniques and share those. The people doing that are amazing and I thank them for what they are doing. Their work will be a road map for all of us. They are the OG pioneers, charting the path forward for the rest of us. 

As far as the APP, I can help people with it while working in the background on the more creative side of things. I want to avoid going the clickbait route where I create disposable content to feed a metric. I prefer substance not only with my creative output but also with the APP. The goal is to provide a service that people need.  I want to create value for others and I fully plan on doing a free version of the APP, which may be all people ever need. And that is great! But, I also see charging a monthly fee for premium services for those who need more than the basic service. 

The decision to focus on the APP is not the one that I wanted to make. If I was calling all the shots, then I would have access to all AI video tools that are being held back for the election. That would mean I might be able to go full-steam ahead on making movies and TV with AI tools. Something that I may be able to do now with animation, which, as I have said before, has more room for error than the life-like AI content. But I am not in the animation mindset yet. Once I transition to the comic book series/ graphic novel then I might be more open to focusing on AI animation. Thinking about that now maybe I should focus on the comic book sooner rather than later. Food for thought. 

My evaluation with one week to go in my AI video assessment period is that with the publicly available tools you can make comics, illustrated novels, commercials, trailers, music videos, short live-action films, and longer animated projects that most people would never know are largely AI-generated. The VFX part of this can't be overlooked. That means those who have been filming live-action sequences but have been strapped financially, can actually do some amazing things right now with AI tools. That is all great but these are not my main creative focus. While this company I am creating will include illustrated novels and comics, these current capabilities are still short of where I would need them to be to create realistic AI TV series and movies. That said, in the meantime, I can focus on all the other things I can create using AI video and audio tools, which is a lot. For me, it is all training for TV and movies, though.  

If I had access to all the tools that are being worked on behind the scenes I would likely have a different take on things in this moment. I like to think I have some idea of what may be in the pipeline, but you never know for sure. The good thing is that it is highly likely that these tools will only get better, and fast. So, it makes sense to focus more of my attention on creating the APP for the next few months. Once the dust settles after the election, it will be the perfect time to shift my main focus back to AI Video. Not that I won't be working on AI Video at all between now and then. No, I just need to prioritize the APP for now to try an make some headway before Fall. 

This time last year, I was thinking that we would be right about where we are with video. A short scene is not a performance, though. Not yet, at least. Consistency and stability are nearly solved and performance will be the next big hurdle. Or at least I think it should be. I believe we may have an AI-animated movie out by the end of the year that will be indistinguishable from a traditionally animated movie.

While I want to be able to do all of these things, I am not attempting to be the first. I want to keep learning about all of it because my goals are more intricate than just trying to be the first to create a proper AI movie. That said, I have thought about what that might be like-- the first AI movie that most people cannot tell was created using only AI tools. It could be a hybrid that actually has some live-action. That seems likely to happen soon, which will raise a lot of eyebrows. And that may open the door for the first all-AI movie that generates enough buzz to create some acceptance and appreciation from the public. The Blair With Project always comes to mind when I think of this. 

That said, these AI tools will continue to improve with each passing week. While my main focus will be on the APP, I will keep working on the trailer for the TV series. I won't be sharing a lot of details about the APP until it is ready for public testing. My initial goal is to have it ready for initial testing by November. Rolling out the APP after the election is a good target. It seems likely that even better tools will become available then. This will allow me to adapt the APP based on any updates before final testing and release. 

I'll keep pushing with the trailer and the illustrated novel series in the background. If I am to create a proper multimedia company I need to have a lot of content. I am also open to doing more with these tools in ways that may not be top of mind at this moment. I may get adept enough with these tools that others may want my assistance with their projects. I could grow fond of creating commercials or fall in love with AI animation. Maybe I decide to create a video game. Who knows? 

The one thing I would stop everything to work on is a new form of storytelling entertainment. If the tools get good enough that I can do all I laid out in the bible for my recent TV series, then that will be my top focus. In reality, I am building towards that anyway. So it is best that I take this step-by-step approach toward the likely inevitability of a more immersive entertainment experience. 

It is a process. A process guided by imagination and fueled by rapid technological change. Embracing it was much easier than expected. Some are vehemently against any use of AI for anything involving creativity. Again, I get it. However, the dreams I have had for over 20 years were stifled in the pre-AI era. My creative visions had remained only partially realized through the written word. The chance to create more with these stories may allow me to fulfill the creative goals that I began setting out for myself when I returned from LA at the turn of the century.

I am an independent artist, and through the years I have grown to value my artistic freedom more than I felt a need to sacrifice it all for someone else's idea of success. I just love to create stories. And with AI I will be able to create all the worlds I've ever imagined while maintaining my artistic autonomy. And that's all that matters. Thanks for reading. 

Tuesday, July 23, 2024

All At Once





I love Time Bandits. A TV series based on the movie is coming out tomorrow on AppleTV, which I didn't realize until I started looking for the GIF. Hope it's not what caused AppleTV to essentially shut down. Anyway, I love the original. So, what does this have to do with anything? 

Today, I got up early and went to the store for a few things. That was when I saw an odd wreck. Some guy ran his car over a curb, between two poles, through a flower bed, and into the stone wall/ sign for an apartment complex. 


Went right between two poles and smashed right into the wall, which you can see in the picture cracked upon impact. The guy was talking to the cops. I'm guessing one of his flip-flops got stuck and he couldn't stop or he was intoxicated. A peculiar accident to see at the beginning of the day. 

When I got home I took out the trash. As I was putting a new liner in the can, a story idea came to me. Ideas sometimes develop over days, weeks, or months. This one came all at once. I normally don't Tweet about ideas, but this one was one of those amazing ones that comes almost fully developed. 


Is it too late to be thinking in terms of Classic Hollywood? 

For a decade, with the rise of streaming, I rarely thought in terms of movies, but stories just decide on their own what they want to be. This story is either an anthology movie, a mini-series, or a holodeck experience.

A what?

By the time I get around to actually writing this story Gen AI tools may make it possible. Nothing wrong with thinking ahead. If this is to be a new frontier we are entering, and it sure seems like it, then we need to think beyond what has been possible up until now. A holodeck or a truly amazing immersive video game-like XR headset experience are what seem the most likely next big steps.

As a storyteller of fictional worlds, I have long wondered what it would be like to tell stories in a gaming format. The worldbuilding on my recent TV series got me thinking about the creation of a gaming world. This was before the era of AI started last year. 

Again, I write a lot of anthologies. I do this because I have a lot of stories brewing at any given time. Most of which have been on the backburners for years. One has been sitting there on simmer for 2 decades. It too is an anthology tale but on the grandest of scales; it feels like the time has to be right, like I have to earn the privilege of writing that story. And I haven't yet. 

Many of the other stories that are waiting their turn upon the stage are stand-alone stories. As time passes while I am working on other projects, these stories will sometimes magically coalesce into something greater than their parts. An Anthology is born. What once could have been three to upwards of a dozen or more, otherwise stand-alone, stories come together like a pod of killer whales ready for whatever the ocean might bring. 

Good grief, I'm mixing metaphors here. What it means is there are certain tales within the fictional world of my current TV series that a gamer may want to explore, especially one of the three interwoven stories in the first season. The same could be said for this new idea I'm so excited about, but also the one that's been sitting on the back burner for 20 years. 

Today's new story harkens back to the early 2000s once again. An era my mind gravitates towards. So many of my beliefs about the world were carved during that period. It was a time when I started to think more like a writer and less like an actor. The new millennium began with so much drama, much more than any of us could have expected. This is likely why the first decade of the 21st century so deeply resonates with me. 

The reason this new story idea triggers my recollection of that time is because it has similarities to both Monarch and Psykosis. The Monarch similarities are because of the story Cipher, which actually echoes back to the 90s when I was studying philosophy and poetry. Therefore, this new story echoes back to those days when I was kicking around Hollywood and Los Feliz. 

Why do so many of these stories always seem to be anthology-type tales? I am not sure I can pin that down. With this current idea, could it be a movie? Yes, of course. Could it be a series? 100%. It really is a collection of stories that make up one story. Sound familiar? An anthology. 

The TV series I am creating a trailer for is the same. However, this new story might be better served as a movie. In fact, even though there are a ton of storylines it might be better to put it all into this one overarching story. I think people will relate to it. A full series might dilute the multi-story potency if not done right, but a mini-series might be another option. I've been surprised by how many movies I once wrote have evolved into miniseries. The specific nature of this new story's overall tale makes it perfect for a future holodeck expedition or a video game. It will be very personal.

The stories I have been comparing it to since this morning are all movies. If it were to be a series, I think it would be better as a limited series than one with multiple seasons. The multiple-story aspect would likely be better served in a more contained format (a movie or a handful of episodes) instead of a sprawling 8 to 14-episode season.

I will probably develop it as a movie and expand if needed depending on how it feels once completed. 

And that's kind of how these things develop. An Idea pops and the next thing you know you're making plans on how to write it. I keep wondering if I will have a normal story idea stir my soul like this one day. But then what is a normal story? A Norman Rockwell painting comes to mind. Hallmark movies and Romantic Comedies too. I'm not sure any of that is normal anymore nearly a quarter of the way into the 21st century.

Nowadays, everything seems to be perceived through a superhero filter, but that's not normal. Normalcy is no longer one view of the world. The most normal story I created over the past decade was not received well. My writing partner thought it was boring, and then I used up one of my last contacts from my time in California with it as well. I thought it was something that it wasn't or had a hard time making it come across. I can't help but think it was too much like an episode in a telenovela. Which was too bad because it was meant to tie into the Monarch universe. Oh well. We go again. 

Okay, it's time to get back to working on the Trailer for the TV series. Need to get some traction on that by the end of the week. I finished the script over the weekend and started working on visuals yesterday. 


I probably should have paid for unlimited generations when I signed up for Runway Gen-3 at the beginning of the month. That way I could just keep generating and generating without concerns for running out of tokens. However, by having spent most of the month focused on getting the script for the trailer right I have also been reviewing the strengths and weaknesses of the model based on the outputs of others. 

We'll see how this week and next week go. Apparently Kling will become available worldwide very soon. I have been reluctant to jump through hoops to get access. It's a Chinese company which gives me a bit of pause but it was the technical hoops I would have to jump through just to use it that has really kept me from using it so far. 

If I can achieve all I want with the trailer using Gen-3, Luma, and Hedra then I will. But Kling may be better than all of those models. We'll see in just over a week what I've got and go from there. And I can't help but wonder if all my new ideas will be geared toward a holodeck or a cinematic videogame. Thanks for reading. 

Friday, July 12, 2024

Summer: Let's Do It!

The past year has been so interesting. A year ago, I believed I would continue on with just writing books and screenplays. Nothing wrong with that. Nothing at all. I had been doing it for fifteen years. Not going to lie, things had become... stagnant. Why? I knew the outcome before starting: excited about the story, mild response, and repeat. I lacked options and the willingness to change my routine.  

GPT-4 drew my attention in February of 2023. And then I dove into research mode about AI. I haven't come up for air since. There is so much to learn and the landscape seems to change every few weeks, often within only a few days or hours. It took me a few months to gain a broader view and see how things were unfolding.

You can see what is possible in the short and long term with just a little research, but you have to be willing to dig. And I love research. A large part of writing is research. It never really stops. 

There are different fields that I have an interest in when it comes to the emergence of AI. Not only do I have an interest in how these tools can creatively help me as a storyteller, but I also think about how these advancements may affect the world. 

Creative Path: AI images, AI Video, AI Voice, and AI Music

World Path: AGI, ASI, Education, Health, and Security

The Creative Path is self-explanatory. I am a storyteller and these are the tools that I need to tell my stories differently than by just writing them, which is why I started writing in the first place. Not to be read but to have a story fully experienced as a viewer.                        

Beyond the worlds within my head that motivate my pen, the World Path is more about us as humans being aware of how computer intelligence will affect us more broadly. My own angle will likely focus on Education.

There is a chance we may all become much smarter as a result of these advancements. We may even be able to live longer and figure out ways to have clean energy to meet the world's needs. It would be nice if we could also figure out a way to keep a handful of greedy people from profiting from the destruction of the planet. 

I am not saying computer intelligence will change human nature. It may, but it might take some doing. And we can be a stubborn lot. But there may be a chance we can clean up our act and become better stewards of this rock. I'm not sure we can do it without evolving in some way. Maybe if we can better educate all people and extend lifespans, allowing more people to become at least wiser if not smarter, maybe we can keep up with computers. Maybe. 

I know my current limitations. I am not the greatest writer in the world. I do not crave the limelight and often move on to a new project before I have exhausted all efforts to sell something. These are flaws I constantly work on. The act of creation is my main purpose. Each evolving story and her menagerie of characters are a mystery for me to solve and the source of my inspiration. The process is the point. Even if I do create a brilliant movie or TV show with the help of AI tools, I will need to be able to entice people into watching, and I loathe shaking my ass. Shake it, Pitters, shake it. Gross. 

I try to keep it simple and do no harm to others. Writing is a way of dealing with the world as I have come to know it. And my path has completely changed since the spring of 2023. The simple, well-worn writing path I had been treading for a quarter century has forever been changed.

AI destroyed that path and I couldn't be happier. 

I know these tools will update within months, but I am committed to learning what I can about them before they do. The tools I have needed to allow me to realize all the creative dreams I have had for decades are here. I have been released from the shackles of my own creative limitations. I cannot draw the images I need to create the graphic novel I've always wanted to make, and I can't create an entire "REAL" movie or TV show without a ton of help. World-building is one of my favorite parts of writing, and I've always wanted to use that passion to create a video game but lacked the wherewithal to attempt it. The tools now exist to help me compensate for those deficiencies. 

For years, I have had images in my head that I have wanted to share, to express in a way that I could be satisfied with and that might allow others to take something away from the experience. The act of writing has had to suffice for a long time. Words are one thing. Images another. Adding images, video, and audio allows me to present stories in ways I have long dreamt of. And they will only improve, maybe even create some new form of entertainment. Would I have loved to have had these tools 25 years ago? Of course. But they are here now and only getting better. 

They will soon be so good that I hope to be on the frontline of a new form of creative storytelling. Ever since I started packaging the TV series this year, I've been imagining exactly what that will be like. There is still so much to learn but the tools are here and a path is clear.

While I try to get up to speed with all of these tools, I fear I may have to push out work on the next illustrated novel series and the graphic novel. Since I already knew how to write a book before AI and I now know how to create a graphic novel with the help of AI -- something I learned over the past year, I can no longer just write books and screenplays when I can also get more involved in creating movies and TV shows. In case you weren't aware, I started writing to create the kind of stories I wanted to be cast in as an actor. An actor? I know, right. 

However, thinking about the story from a character's perspective has taught me a lot. By the mid-2000s I came to think of myself as a method writer. What the hell does that mean? I was never a "method" actor, my teaching was grounded in the work of Stella Adler who preached personal experience and imagination over emotional memory. Over time, that not only helped me understand the motivations of my characters, but also opened my mind to imagine all kinds of stories. 

The sheer volume of writing work I have cranked out over the years has gone largely unread. With 75% of that material meant for the screen, it is not surprising that I always visualize a story for the screen, even if I am writing a book. I see the movie play out before me, all around me. I live those moments with my characters as I write their stories, even to this day. Therefore, to now have the ability to visualize stories for the screen is like turning back the clock to Day 1 of my writing journey. Not that I want to act in anything ever again, but to have control of the sound and images of a story is both exciting and terrifying. 

The terror comes from knowing that it's all on me now. I can't just toss a new story onto the dust heap and say, "I tried to convince people to help me create the movie or TV show but there wasn't enough interest. Oh well, I guess I'll try again with something new." No longer. I'm breaking the cycle. The pile is too tall and I have new tools to work with. 

While my first objective is to create a teaser and a longer trailer for my current TV series, I would like to try and make the series the old-fashioned way while we still can. The long-term goal is to revisit some of these other stories using AI tools. Whether as movies, TV series, or this new hybrid storytelling format that is emerging. Very exciting! 

I mentioned the dusty pile of stories from the past 25 years. That is not a joke. Maybe that's just me laughing so as not to cry. Either way, I have several dozen stories that I can use to build a video library, with new tales waiting in the wings. That is why I have completely changed the path ahead. I can't see myself fully focused on an illustrated novel series while learning how to use the tools needed to create a teaser and a trailer, which may be more like a short film. 

Once I have a handle on these tools, I can start to divvy out more time for the illustrated novel series. My hope is that it will only take until August to get up to speed. When considering that I have written, directed, produced, edited, and arranged the music for several short films, maybe that will help me learn on a bit of a curve. We'll see in August. Until then it's time to accelerate my AI video research. Next stop? Teaser.