News from the AI & ML world
DeeperML
David Crookes@Latest from Tom's Guide
//
Midjourney has officially launched its first image-to-video generation model, named V1, marking its entry into the competitive AI video market. This new model enables users to transform static images, whether generated within Midjourney or uploaded, into short, dynamic video clips. Unlike some competitors that rely on text-to-video generation, Midjourney's V1 focuses on animating existing visuals, building upon the platform's established expertise in AI-generated imagery. The model supports features such as dynamic motion, basic scene transitions, and various camera moves, with aspect ratios of 16:9, 1:1, and 9:16, catering to diverse creative needs.
The V1 model generates four variations of each video, each approximately five seconds in length at 24 frames per second. Users can extend these videos in four-second increments, up to a maximum of 21 seconds, allowing for greater control over the final output. Midjourney offers two primary motion dynamics settings: "Low Motion" for subtle animations and atmospheric visuals, and "High Motion" for dynamic movements and lively subject animations. Users can choose automatic prompting, where Midjourney determines motions based on the image context, or manual prompting, where they explicitly instruct the desired animation style via text prompts. However, its founder, David Holz, said the goal is aesthetic control, not realism.
Priced starting at $10 per month, the Basic plan grants access to the V1 model, making it available to a wide range of users. However, generating videos consumes significantly more GPU resources compared to image generation, approximately eight times as much, which will eat in to monthly credits faster. The launch of Midjourney’s V1 positions it alongside industry leaders like Google and OpenAI, although each company approaches video generation with different focuses and strengths. While V1 is currently accessible via the Midjourney website and Discord, the company acknowledges that the costs of running the model are still hard to predict.
ImgSrc: cdn.mos.cms.fut
References :
The V1 model generates four variations of each video, each approximately five seconds in length at 24 frames per second. Users can extend these videos in four-second increments, up to a maximum of 21 seconds, allowing for greater control over the final output. Midjourney offers two primary motion dynamics settings: "Low Motion" for subtle animations and atmospheric visuals, and "High Motion" for dynamic movements and lively subject animations. Users can choose automatic prompting, where Midjourney determines motions based on the image context, or manual prompting, where they explicitly instruct the desired animation style via text prompts. However, its founder, David Holz, said the goal is aesthetic control, not realism.
Priced starting at $10 per month, the Basic plan grants access to the V1 model, making it available to a wide range of users. However, generating videos consumes significantly more GPU resources compared to image generation, approximately eight times as much, which will eat in to monthly credits faster. The launch of Midjourney’s V1 positions it alongside industry leaders like Google and OpenAI, although each company approaches video generation with different focuses and strengths. While V1 is currently accessible via the Midjourney website and Discord, the company acknowledges that the costs of running the model are still hard to predict.
ImgSrc: cdn.mos.cms.fut
Share:
References :
- AI GPT Journal: Midjourney Introduces Image-to-Video Generation Model: What You Need to Know
- Fello AI: Midjourney Video V1 Is Here! How Does It Compare to Google Veo 3 & OpenAI Sora?
- Shelly Palmer: Midjourney Set to Release its First Video Model
Classification:
- HashTags: #Midjourney #AIVideo #ImageToVideo
- Company: Midjourney
- Target: AI Video Creators
- Product: Midjourney
- Feature: Image-to-Video
- Malware: V1
- Type: AI
- Severity: Informative