Transcription
We have talked about various video, audio, and image generators, but today I want to show a tool that combines them. And this is not just a platform where you can choose a model that suits you. This is the next level, a full-fledged agent that maximally simplifies the video creation process. How and in what way does an agent differ from a regular generator? Let's figure out what is usually needed to create a video. A script in ChatGPT, for example, then animation, another tool, voiceover, yet another solution. And then we edit all this in some program. It turns out that you still create everything yourself with your own hands. Focal AI is a different approach. It is a full-fledged AI agent that takes most of the work upon itself. The difference with a generator is huge. You just describe the idea, and then it writes the script itself, breaks it down into frames, generates frame by frame, generates audio, edits, and shows the result. Sounds interesting, but let's go in order. Registration is simple, and after that, we get into a free subscription. Yes, you can use it for free too, with its limitations, of course. To test the capabilities, you are given 110 credits, and 10 will be added every day. This is enough to understand how everything works. Paid plans start from 10 dollars, and this is quite budget-friendly, considering the capabilities. The main menu is the familiar chat window. What capabilities are there? Here you can choose a model for generating images and video. Advanced ones are in paid subscriptions. There is also the popular Sedens, Eveo, and Nanno Ban. I will note that the cost of generation in different models differs, and you can find it out immediately. This is convenient. Next, you can choose the language so that Focal AI accurately understands what you want from it. Russian language is available. Key. There are several ways to communicate with the model. The first is you simply describe your idea. The second and third are if you already have a structured script for the video with frame-by-frame breakdowns, descriptions, and so on. It also understands the JSON format. If you click on the plus, you can manually select models for video, audio, or image generation. There are also templates, or a reference, or other necessary files. You can upload them. Well, let's now look at an example of how it works. This will be clearer. I will show generation from an idea. Prompt. Make a short video. 15 seconds. Samurai fighting with swords. Animation style - cinematic atmosphere. And here is where the most interesting part of the agent's work begins. First, it prepares the script, and we see the entire thought process. You can stop it at any moment if you see that something is going wrong. During script preparation, Focal AI will definitely ask clarifying questions, from the video format, the use of music or voiceover. You can choose ready-made options or write your own as an additional prompt. That's it, the script is ready. Frame-by-frame breakdown with dramaturgy, description of what happens in each frame. And again, it will ask what to do next. It will offer ready-made solutions or we write our own remarks. We send it to generate the video. This is what the final result looks like. Several scenes, a menu on the left, and a working area. First, let's go through the menu on the left. Firstly, all media files are saved in the library, and they can be downloaded separately. They are structured by type. Chat is a saved dialogue that already existed and which can be continued, for example, to correct scenes. Character. You can create your own character for these scenes by uploading a sample photo and describing everything needed in the prompt. Next is a large stock library of what can be used. Photos, videos, music, transitions between scenes, video effects, adding text, subtitles, and general video settings. Working area. Here we can edit the video directly in this window without installing additional applications. No, of course, if it's complex editing, then you will need professional tools. But to trim a scene, swap frames, increase speed, all this can be done here. Moreover, you can regenerate each scene in the chat, add transition effects and video, as well as the musical component. For example, find music in stocks and add it to this project. Let's create some voiceover. We go to media files. New voiceover. ElevenLabs model by default. Here is the voiceover we got. Now let's add some transition effect. The most frequently used ones are here. And a video effect. There are not many here, but the generation possibilities are endless. That's it, now we click export. You can choose quality and resolution. And very importantly, even in the free subscription, there are no watermarks. Well, and one more cool feature must be shown. You can insert references into the chat. Let's create something based on a reference from TikTok. Kittens always work flawlessly, and we will replace them with puppies. You give the agent a video, it analyzes it and reads the structure. How many scenes, what is the frame change tempo, what is the text tone, what is the visual. Then it creates a template, such a structure, such a rhythm, such a presentation style, and applies it to your topic. That is, it does not copy completely. The animation style and so on can be set in the prompt, but the overall structure can be. And here is the result. The overall vibe, in my opinion, is captured very accurately. Focal AI is not just another generator. It is an agent that thinks for you, writes the script itself, generates itself, edits itself. An excellent option to become an AI creator at a time when it is gaining popularity. The link, of course, is in the description. That's all. Perhaps, in the modern world, the most important thing is to remain human. In all senses. Alexander Rybakov, Plytsa Social Technologies.