Video

I confess, I had fun whipping up those images 😁😁

Thanks, Mike 😁

Cheers, bud! Hey, have you signed up for a paid plan or are you winging it with the freebie version?

@@HaydnRushworth-Filmmaker I'm in the pro plan since the start, its the only way I can keep up with a video every week and shorts almost everyday...

Haven’t heard anything on the grapevine yet.

Hey, thanks very much, great tips! :-)

Ooooh, good question. I've had some positive results with RenderNet.ai as well as Leonardo.ai and, or course, Midjourney. Each tool has it's quirks, and each of them still require quite a bit of tinkering. In reality, my answer to your question is yes and no. It seems that the best consistency comes from characters with their faces in the same direction / expression, which doesn't really solve the problem. The models that you "train" appear to rely on you providing a library of images that you approve of, for the tool to then select from and place in a different costume etc. Not a bad idea, but you still need to create that library of consistent faces somewhere. I'm hoping for the day when AI tools create 3D versions of assets that have all the angles already in the bank, so whichever angle you choose, it already knows what the correct angle looks like time and time again.

This is such an insightful thing to share. Really appreciate it. I actually went and watched a CZcams video of all the Mee6 highlights because I hadn't watched enough Rick & Morty to follow the reference. Good analogy though, it actually did help. From my perspective, it looks more and more likely that the most workable solution to creating the Movie-making tool that I'm dreaming of, would be to blend AI with existing videogame tech. To use AI for the creative parts, but to wrap that creativity around controllable videogame assets (like skins, essentially) and then pick up with the control with the elements that AI helped to create. In fact, what if it turns out to be as "simple" as an AI assistant that uses Unreal Engine? I think the AI skin creation part would still have merit, but I'm longing for a tool that is easy enough for writers to pick up and instinctively use like a video game.

@HaydnRushworth-Filmmaker creating the training data for such a thing would be very challenging but doable. I like this concept and approach. 🤔 I wonder...

Absolutely! For me, the dream of AI Video is definitely the ability for visually rich stories to be told to relatively smaller audiences. Audiences with niche interests and views.

Hey, thanks very much! :-) Btw, I'm aiming to follow your lead and start to create more experimental content that is aimed at just entertainment. Keep it up. Love what you do :-)

@@HaydnRushworth-Filmmaker Awesome, now forget all the other models, the best model in town is out now, go and find Luma Ai Dream Machine, it’s gonna blow your mind!!! Already using it! Its amazing!!!

Actually, the thing I'm most interested in, is the potential of AI to let guys like me with a screenplay that I love, finally see it turned into something visual. I spent more than a decade writing the screenplay for "Telepathy SUCKS!" and I wrote it to WATCH it, not just to read it. For me, the creative process is only half-finished, even though the screenplay is finished. If I'm in this predicament, I'm sure I'm not the only writer with a big-dream story that will never be turned into a finished movie through the traditional filmmaking route.

At the moment, the simplest route to monetization looks like CZcams. My current plan is to upload single episodes to a dedicated CZcams channel (to target the appropriate audience for my story, "Telepathy SUCKS!"). I'll probably upload the movie scene-by-scene as it it completed. Before it gets to a movie, though, I'm aiming to create a "Visual Audiobook", which is essentially, an audiobook with illustrative images or video clips.

@@HaydnRushworth-Filmmaker But still, if we return to your idea, I thought about it. It seems to me that we could really kill CZcams if we made a free service that would have no advertising and no subscription, but for what it would live, I think for donations from users to specific films. That is, if they like the film, they donate, but if not, then they don’t. But I don’t know if people are generous? And the second question that I have is whether the owners of the film will want to publish their film if their material income depends on the preferences of the audience. We all know that the film One Flew Over the Cuckoo's Nest would hardly have gained such popularity at the box office today as The Avengers, and the question arises: how to turn the industry around and force directors to make great movies and not chewing gum?

Good idea, it might work :-)

That's a really thought-provoking point. I suspect we will never truly remove humans from the filmmaking process, and I suspect actors, writers and directors will actually be the ones who are hardest to replace, just as you mentioned.

Thank you, really appreciate it 😁

Thanks very much 😁

Ooooh, good one, I'll test that and add it to the list of suggestions from the comments here. Thanks very much. :-)

Hey! Brilliant, glad you found it helpful… next step, I need to start actioning my own ideas 😁

Hey, no problem, thanks for watching 😁😁

That’s a REALLY interesting point. Ultimately, the micro expressions that make up a huge amount of non-verbal communication will all need to be possible and controllable before meaningful stories can be told with video AI, and so when that level of subtlety is viable, tackling the non-symmetrical nature of the human face would make sense during the same development phase. I guess the risk is that it could go SO wrong. I’ve been having a nightmare of a time recently trying to get Midjourney to produce a photo-realistic image of a beautiful girl who DOESN’T have freckles. It’s like some kind of “freckle-equality” campaign group has worked their way into all the Midjourney board meetings to push for freckles on every girl possible.

@@HaydnRushworth-Filmmaker 🩷

Fantastic, thanks so much for the tip-off. I've added those to the document I started based on the suggestions from folks right here in the comments for this video. I'll try those out :-)

Oooooh, good tip. Thanks very much. I’m gathering a list of all the hints and tips from all the comments into a document that I’ll probably feature in future CZcams content.

@@HaydnRushworth-Filmmaker Very cool! Can’t wait to see them!

Oooooohhhhh, now, THERE’S a thought! 🧐🤔🤔

Quite a few people have shared the same idea in the comments here. I really like the way you put it though. Thanks very much :-)

Thanks, you’re right, of course. Several people have commented on the same point. 🙂

Ahhhhh, brilliant. Thanks for the tip, I’ll add it to my growing list of additional tips, tricks and techniques from the all the comments in this video 😄

😂😂😂😂 Same time next week? 😉😆

Thanks very much, that’s what I’m aiming for :-)

Hi, thanks very much :-) Sponsorships? Sure, let’s talk: haydnrushworth@me.com

Absolutely, it's another major problem, and as it happens, that's what I'm working on right now, testing Midjourney, Leonard and RenderNet to see which one is the best for consistent characters (and costumes, hair, makeup, jewellery, hair colour etc.)

It's a valid point, although this discovery really was a breakthrough for me, and the good thing is that it works with the free AI tools as well as the paid plans :-)

@@HaydnRushworth-Filmmaker Well whatever you do don't make a video about a white guy who uses ai like a slave to make videos for him. That would be too ironic. It's not like it can tell you no.

Scientists have break throughs before it is turned into a product, weapon, or pharmaceutical all the time.

Oooooh, now there’s a great idea for a story! 😃

Absolutely.

Hey, thanks very much :-)

NOOOOOOO!!!!!! 😂😂😂😂 it’s SO obvious now!!! I’ve had several great suggestion posts about this in the comments but this one is the winner!! I think I need to make a follow up video with all the techniques. Thank you so much for sharing that 😁😁

@@HaydnRushworth-Filmmaker You can also control the strength of that "--no". You can prompt like this: all the stuff you do want:: all the stuff you don't::-.3 A --no is -.5 but is often too heavy so you can vary that with a multiprompt negative. But ya, tell the character what to look at. Short prompts aren't the answer - good prompting techniques are. V6 prompts can be HUGE and detailed. Be specific, be succinct. If you say no to something in the prompt proper, MJ will ignore the no (or the "without", "remove" or "not") and instead include that thing. Use either the --no command or a multiprompt negative.

Good tips, I'll try those out. Don't know about LangChain or DSPy.

That seems to be the common thread here. Thanks for the tip-off, I think you're right.

Hey, mate, thanks very much 😁, you’re totally welcome!

Good one, I'll give it a try. Has it worked for you?

Hey, thanks for the feedback. I actually agree with you. I've joined a local actors workshop so that I can find real human actors to work with because from what I can see of AI tools, the best version to get authentic, human-like results will be focusing heavily on video-to-video instead of text-to-video or image-to-video. I actually made a video about the camera and filming gear that you'll still need in order to get the best results with AI as the technology continues to evolve. In terms of not making money from something that anybody can do, I agree 1000%. From where I stand, I see a bucketload of AI video content on social media that is really only getting traction because of the novelty factor of the AI tools, and you're right, anybody can use the avalanche of "free" Beta version tools to create that kind of content. I believe the most enduring content will always be based around a solid, core idea or story, where the AI tool becomes invisible, and just sits in the background. As with a great story, nobody cares what computer you used to type it on or what word-processor you used because it's just a creative tool. AI video needs to become the same thing, a creative tool for human creatives to use to tell stories and share big visions and ideas that they would otherwise struggle to bring to life. In terms of the low income challenges that most of the world faces, I couldn't agree more. In fact, that's the reason I love the possibilities of AI, because it allows a guy like me, struggling to get by with a family of 5 living in a 2-bedroom apartment and living in Yorkshire (not Hollywood, and not even London) to have half a chance to bring a big-vision story to life and share it with the world. I love the democratisation of these kinds of tools precisely because of how empowering they are for little guys in obscure areas like me. As it happens, I also share your distaste for using CZcams as a platform for selling online courses. I've talked about this with my wife at some length, and even though the various CZcams "gurus" would be telling me that's what I need to do to make some decent money (so that I can move my family into a family home with a garden for my children with at least three bedrooms), it just doesn't feel like the right thing for me. Instead, I'm going to figure out ways to sell my story ("Telepathy SUCKS!"... you can buy the ebook on Amazon... though I haven't even promoted that yet). I'm going to try a visual e-book, an audio book, a visual audio book (stills and then video clips played whilst the story is being read) and I'll eventually start uploading scenes from the movie to CZcams and try to monetise that (in a few years from now, with the AI really catches up). The one thing that I definitely agree with you is that if something is easy to master, it will never make me money, but as I'm discovering with AI tools, its easy to create one, single shot, but it's incredibly difficult to create a coherent, connected narrative with fluid continuity and authentic, human moment, facial expressions and voice intonation. For that level of professionalism, I believe it will always be a highly skilled filmmaking endeavour, and so that's why I'm learning as much as I can as quickly as I can. I'm not at all a highly-skilled user, but then, at this stage, nobody is. So, the real skill is to be willing to learn as fast as possible. I love your passion and enthusiasm. Keep it up :-)

I agree, actually. In time, when the tools make real cinematic results possible (which is FAR more than SORA-level image quality) I'm going to aim to upload my movie to a dedicated CZcams channel, scene by scene, and hope to build a big enough audience that the ad revenue from a monetised channel will make me some money. Either that, or, if the tools lead to great results, maybe a dedicated streaming service will pop up that specialises in independently-made AI films.

Hey, mate! thanks very much 🙂

FANTASTIC tips, thanks very much :-)

@@HaydnRushworth-Filmmaker theres a guy called future tech pilot, he knows so much about MJ, ive learned so so much from him.

Great tips, did those work for you? I'll check them out today and give them a try 🙂

Cheers, bud! Loved your latest short too 🙂

Thanks a lot, have a great week yourselves :-)

It's a good question and I'm still not sure. I'm about to delve deeper into that whole area, so watch this space. The next step for several projects I'm working on is to find the best tool for creating consistent character reference images to use as foundation references in a project. I'm comparing Midjourney, Leonardo and RenderNet... watch this space for the results :-)

@@HaydnRushworth-Filmmaker awesome, i’ve been looking into ways to add more movement like you depicted here for ai films that i make. I tried viggle and a couple others but the export quality was too low for what i needed. I’ll check out the tools you showed here and keep looking for alternatives to improve how much movement i can add to films like this: The Story of Pentecost | 4k Ai Film Official Trailer czcams.com/video/ct5T9zmAjZU/video.html