New GPT makes AI Films Consistent & Automates Midjourney!
ฝัง
- เผยแพร่เมื่อ 22 ก.ย. 2024
- I've created a tailored version of ChatGPT for AI Filmmaking. It automates Midjourney workflows by creating format-ready prompts, helps create consistent look and feel across every scene in your AI movies, and helps you flesh out your story and visuals.
AI FILMMAKING ASSISTANT GPT:
➡️ chat.openai.co...
AI FILM SCHOOL (coming soon)
🎥 www.thereelrob...
TWITTER
🆇 / thereelrobot
#aivideoanimation #midjourney #chatgpt
*Link to the GPT:*
chat.openai.com/g/g-hiKxJNAlp-ai-filmmaking-assistant
I'll be making updates to it based on comments!
Incredible work! The prompt generation is really slick!
Text to video techniques next please!
You got it
I'm a 3D Technical Director, and after seeing this, I know there has to be workflow that can use 3D animation to improve the quality/consistency of the AI video renders
Any thoughts on how you'd approach it? I will typically upscale in Topaz and use their frame interpolation, but these aren't drastic differences. You're a little at the mercy of the AI tools.
@@TheReelRobot You got the post production down to a science tbh. But I was wondering if we could do video2video, or 2-10 imgs2imgs. Does that make sense?
@@tstone9151totally. There’s a tool called AnimatedDiff that’ll build keyframes by just making a ton of images. It’s smoother, but the AI still adds some undesired stuff.
Also, I think the luxury of the type of animation I’m doing is how light the workload is. Video to video and you still need to produce a lot. Though you can do something simple in blender and rotoscope/stylize it easily now.
But an animator like you who can use tools like this or Mootion , that give you the powers of a whole studio without all the grunt work, just the creative work, is the real dream.
@@TheReelRobot The concept of video-to-video might initially seem overwhelming in comparison to your current workflow. However, if we can reduce the workload by 80-90%, the prospect of a single 3D animator independently producing high-quality short films becomes a realistic and attainable goal.
Imagine if we simply block out the environment or background using basic polyCubes to create an authentic depth map. By focusing solely on pose-to-pose animation for 3D characters-just the key poses-and allowing AI to manage the intermediate frames, we can significantly streamline the process. Additionally, the use of 3D cameras provides greater control over camera movements, enhancing the overall quality of the animation.
I genuinely believe that with these approaches, we 3D animators have the potential to create films from the comfort of our own homes. Having seen the videos on your channel, I am confident that this is not just feasible but well within our reach.
Man, your art is rocking beautiful !
Wow ❤ im subscribe
Waiting for updates
What are you looking to see added?
Really cool, thank you. What did you use to animate the shots from the example?
RunwayML Gen 2. They recently did an update that has made it extremely good.
You can try it free. Just drag and drop a picture and try it with a “Motion” setting of about 5
very cool!! is there a video for the next steps?
It’s probably easiest from here to try using it, and click one of the recommended conversation starters so the GPT guides you through what to do. You can try that here:
chat.openai.com/g/g-hiKxJNAlp-ai-filmmaking-assistant
This is just what I have been wanting! I have been toying with doing my own filming GPT, don’t understand enough about how to do so for my film making. So, thank you for this. A question I have is can your GPT be modified to produce Photorealistic out puts?
That is my area if interest in having images for Pika Labs or RunwayML.
Absolutely. I'm serious about taking peoples notes about how to make this better, since it's pretty easy for me to update it.
I mean you could tell the GPT right now "I want this to be photorealistic" and it will likely update things in your prompt right away though.
can this GPT help to make constant characters?
I would say not really. It’s still very hard to get consistent characters. This video is largely how I do it:
th-cam.com/video/69yQjRGFDDU/w-d-xo.html
Love your video. I am using free version chtgpt. So i can not use your GPT. There any free option for try your prompt? I not affort yet 20$ Par month :(
The problem is, I don't have control over it - I want this to be free. I've heard OpenAI might make GPTs available to everyone on basic plans, but for now they're forcing you to upgrade :(
could this be done using Stable Diffusion?
Yep, absolutely. It can even work with APIs. What tools are you using specifically when it comes to Stable Diffusion
i just started getting going with SD and need to learn how to use things like ControlNet and WarpFusion to start getting better at this. I love it though @@TheReelRobot
How dod you change the size of the image in midjourney?
It’s called an aspect ratio. In Midjourney, you write “-ar 16:9” or whatever number your ratio is after the “-ar”. 16:9 is standard widescreen, and if you put in something like “-ar 21:9” it’ll be more like a movie theater format
Thanks man! Subscribed!@@TheReelRobot
I just started messing around with making A.I. shorts. I'll make sure to shout you out brother! thanks again for this cool assistant!
I can not access your gpt from my country :(
Nooooo. I’m sorry to hear that
wohu，try it out