ChatGPT Astra + Flux 3 Max = Insane AI Videos

0:00Videos like these used to take hours to create, but now you can make them in seconds without filming anything or showing your face. Plus, it costs less than a dollar for each video. I'll show you how with Flex 3 Max, which just dropped. First, we'll create an animated story from a prompt, and I'll share how to keep your character consistent. Then, we'll turn images into advertisements and videos with music, sound effects, and voice-overs. Finally, we'll schedule everything to social media. Let's get into it. Let's start in the file playground. I'm going to show you this setup first, and then we'll make our animated story. Here, I have selected

0:34Flex 3 Max and text to video. That cowboy you see on the right side is just their default sample, okay? We're going to replace this prompt with our own. First, we're going to make an animated character, Pixar style, escaping an ordinary day at the office for a roller coaster ride. And all we need to do is start with a prompt. Okay, so I'm going to give it a character, an action, a setting, and sound, so that there's an actual scene to work with. And here's what I had Gemini write. We're the office scene first, then the escape to the roller coaster. And for this first attempt, I'm going to generate these

1:08separately, okay? Now, you select how long you want your video to be, anything from 5 to 20 seconds. Generation time just means how long it takes for the AI model to create the video, okay? And here I've set the video length to 5 seconds, and here the resolution 1080, then hit run. Here's another generation using the same prompt, and it's a different character, different vibe, similar bedroom. Now, I'm going to use a second prompt for the roller coaster, and watch what happens

1:37to the character, completely different. She's changed between the two clips. So, next, I tried putting the whole story into a single 20-second generation. Back in Gemini, I'm asking for one continuous sequence, starting at the desk, show the transformation, then arrive on the roller coaster. Her description stays consistent all the way through. Okay, now I'm copying that whole sequence together so the model has the full story in one prompt. Okay, then we're going to go back to file and I'm going to replace the old prompt with this new prompt. The transition is part of the action now instead of asking for two separate clips and hoping that they match. This time,

2:18I'm going to choose for duration 20 seconds. I'm going to keep it at 1080 resolution and then hit run again. That will give the scene time to move from the office into the roller coaster ride. Okay, this works much better for keeping your character consistent because then you can watch her as the whole scene changes around her and everything is

2:38consistent.

3:01Now let's make a product advertisement. We're going to start with one perfume image and give it movement, music, and a voice. So here's my starting image. We have a bottle, label, and the lighting is already good. Now I need to describe what happens next. I'm using Codex for this one. The bottle stays still while the camera moves around it. We have petals and mist that add movement and we want the light to catch the glass. So I gave it this guidance from file. Write a clear sequence of visible actions. Think about what we see first and how the camera moves and how the shot ends. And

3:36I've included music, atmosphere, and a soft voice introducing the perfume. Okay, now I'm going to switch to image to video and paste the prompt. Okay, the upload the perfume image as the starting frame. Okay, here we are uploading it and then I want to run it for 10 seconds duration at 720p then hit run. And listen to how the sound finishes the

4:01advertisement. Nocturne. Leave a little mystery. The next one is pretty fun. I'm going to make a video inspired by Space Jam starting with this image of Michael Jordan on an outdoor basketball court. I want to keep him recognizable while everything around him turns into a cosmic universe just like in Space Jam. So, here's our prompt. Okay. He's going to bounce the ball then a glowing ripple spreads across the floor just like you see there and then the

4:34court should transform into Space Jam. So, we want to every action to seamlessly lead into the next so everything feels like one connected scene. Okay, so here's our prompt. Now, let's go back to Fal AI. I'm going to replace the perfume prompts with our basketball prompt. Okay, I've described his face, proportions, and his jersey and again we want the transformation to happen around him including the sound of the basketball dribbling. I'm going to leave it at 10 seconds duration, 720p, let's

5:03run it.

5:11All right, now let's get these videos posted to social media. I'm going to use the app I built Blpotato that easily posts everywhere. Okay, here's the upload process in Blpotato. Open a post, add your video, then choose where it's going. I'm showing this with another clip but the steps are the exact same. Okay, you can add your caption, you can select your social accounts, and you can publish straight away or schedule for later. Okay, so once the videos are ready you can line them up easily across Instagram, TikTok, and YouTube. So, how much does it cost to generate these videos from Flux 3 Max? So, here I've

5:46shown the rough pricing rates. 10 seconds is about $0.43 at 720 or $0.73 at 1080. Then, 20 seconds is $0.85 at 720 or $1.45 at 1080p. And that's the entire workflow. Turn an idea or image into a video, add movement and sound, and then get it posted with Blotato. So, a few seconds under a dollar and AI for

6:08the prompting.