【無料で使える】動画生成AI★Haiperの使い方、意外と知らない同時動画生成方法

The video introduces the viewer to the world of AI-generated videos, highlighting the realistic and high-quality visuals produced by AI, specifically mentioning a video of clouds reflected on water and a woman with realistic skin texture. The presenter, Misuzu from the Beginner's AI Lab, discusses the anticipation for OpenAI's video creation tool, 'Sky,' which is not yet publicly available. However, she introduces 'Hyper,' a video generation AI that has been generating buzz on social media for its comparable quality. Misuzu invites viewers interested in video creation to try out Hyper, which is in beta and free to use at the time of the video (May 2024). She also mentions that the channel is aimed at beginners and provides methods to utilize image editing and AI, encouraging viewers to learn and grow with AI.

The presenter outlines the process of starting with Hyper, beginning with the necessity of account registration. She guides viewers through the sign-up process using a Google account and navigating through the initial setup, including answering an optional survey from Hyper. Misuzu then takes viewers through the home screen of Hyper, showcasing the variety of sample videos available, ranging from live-action to anime and oil paintings. She explains that Hyper can create videos from 2 to 4 seconds long and demonstrates how to create a video from text, using a provided prompt and seed value. The seed value is crucial as it controls the randomness of the generated video, allowing for different outcomes with the same parameters or reproducing the same video with the same seed value. The presenter also discusses the possibility of changing the video's playtime and mentions that the method may vary or become a paid feature in the future.

The video continues with a demonstration of creating a video from an image using Hyper. Misuzu selects an image that she finds powerful and instructs viewers on how to upload it to Hyper for video creation. She emphasizes the importance of confirming the image size is 16:9, suitable for YouTube, before proceeding. The presenter then explains how to use the 'Create' function without adding any text prompts to generate a video from the uploaded image. She also touches on the possibility of re-generating the video if the initial result is not satisfactory and mentions the download option once a satisfactory video is created. Misuzu then explores the concept of using an image generation AI, like Cアート (C Art), to find suitable images for video creation, especially for those who find it challenging to come up with their own prompts.

Misuzu introduces Cアート (C Art), an AI-powered image generation platform that allows users to create a wide range of high-quality images from realistic photos to animations and illustrations, simply by inputting prompts in Japanese. She provides a brief guide on how to access and use Cアート, including how to log in using a Google account and change the language settings to Japanese. The presenter then demonstrates how to search for images on Cアート that align with one's creative vision, either by browsing through categories or using the search function. She selects an image that she believes would be suitable for creating a video with a powerful and dynamic feel, and shows how to use the 'Search' and 'Paper airplane' buttons to find and generate images, respectively.

The presenter explains how to use an image from Cアート to create a video in Hyper. After downloading an image from Cアート, she returns to Hyper and selects the option to create a video from an image. Misuzu uploads the image and chooses to create a 4-second video without adding any text prompts. She emphasizes the convenience of Hyper's video generation process and the option to download the generated video. The video also touches on the possibility of不满意 (not satisfied) with the initial video and the option to re-generate it. Misuzu then experiments with adding prompts to influence the video generation process, such as simulating dust flying in accordance with the movement in the image. However, she notes that the results are somewhat subtle and that the 'Repaint' feature, which allows for style changes, does not seem to be functioning as expected yet. She expresses hope for future improvements in these features.

Misuzu demonstrates a method for simultaneously generating videos from multiple images using Cアート's free version. She guides viewers through selecting multiple images and setting the video duration to 4 seconds for each before initiating the video creation process. The presenter shows how this method can significantly save time, especially when creating multiple videos. She reviews each of the generated videos, commenting on the dynamic transformations and the AI's ability to recognize and animate elements within the images. Misuzu also discusses the potential challenges of editing these videos due to their complexity and the impressive nature of the transformations. She concludes by mentioning the 'Update' button for video regeneration and the importance of considering commercial use guidelines when using Hyper, which she summarizes from the provided terms of service.

The video concludes with a discussion on the terms of use for Hyper, emphasizing the requirement to be at least 18 years old to use the service. Misuzu highlights that Hyper transfers all intellectual property rights of the output data to the user, allowing for peace of mind when using the platform. However, she cautions against generating videos featuring famous anime characters or celebrities, as sharing such content on social media could be prohibited. The presenter encourages viewers to experiment with Hyper while it is still free and to enjoy the process of video creation. She invites viewers to leave comments, likes, and subscribe to the channel for more content and potential explanations of unclear points in future videos.