🚀Crea PELÍCULAS con IA a partir de una sola indicación [personajes y escenas consistentes]

Creating Consistent Stories with AI

Introduction to Story Creation

• The video introduces a method for creating stories with consistent characters and coherent scenes using AI and a single prompt (PR).

• It emphasizes the potential of combining the right prompt with appropriate tools to produce videos that can compete with Hollywood, appealing to filmmakers, artists, and content creators.

Generating Character Images

• The first step involves generating images of characters in a 3x3 grid format, showcasing nine different angles of the same character within the same environment.

• A specific prompt is provided for creating this grid, which may appear complex but is explained in detail later in the video.

Using Tools for Image Generation

• The presenter uses the "nanobanana Pro" model within a tool called Divet to generate images based on text prompts.

• Instructions are given on how to set up the tool: selecting image generation options, changing dimensions to 16x9, setting resolution to 4K, and choosing the number of images.

Resulting Cinematic Grid

• Viewers see an example of a cinematic grid composed of nine images showing various angles of a woman in an abandoned city scene.

• Each frame maintains visual consistency regarding character appearance and clothing due to precise instructions given in the prompt.

Understanding Prompt Structure

• The structure of the prompt is broken down: key sections must remain unchanged while others can be personalized.

• An example describes a woman in an abandoned city; viewers learn how these details reflect directly in generated images.

Customizing Prompts for Different Scenes

• Details about camera angles and shot types are specified within prompts to ensure diverse perspectives across all frames.

• Visual style and lighting can also be modified according to desired outcomes while maintaining subject consistency throughout all images.

Example with Different Characters

• Another example features a Navi woman walking through Pandora's forest, again utilizing a 3x3 grid format while keeping character consistency.

Addressing Common Challenges

• Potential issues such as facial clarity are discussed; solutions include cropping images for better quality before animation.

Final Steps for Image Editing

• Instructions on editing involve resizing problematic images and performing face swaps using clear reference photos.

Improving Video Generation Techniques

Results Comparison

• The speaker presents two images, highlighting a significant improvement in the second image compared to the first, despite acknowledging that it is not perfect.

• Emphasizes the importance of storytelling in video generation, referencing a narrative inspired by "Standersin" to illustrate how to create engaging content.

Narrative Development

• Discusses modifying the description for image creation to enhance storytelling; provides an example involving a young woman entering a house with a portal to another dimension.

• Suggests using ChatGPT to assist in rewriting prompts for generating images that align with narrative sequences instead of just camera angles.

Image Sequencing

• Instructs on copying and executing prompts in ChatGPT, resulting in a clear visual projection of the story through sequential images.

• Describes how each image should depict specific actions within the narrative, such as the woman standing silently outside and approaching the door.

Animation Process

• Advises on animating images one by one after generating them; emphasizes returning to tools for converting images into video format.

• Details how to input prompts for animation, including camera movements and character expressions, while suggesting improvements via ChatGPT.

Final Adjustments and Output

• Recommends refining prompts for better results before finalizing animations; highlights settings adjustments like resolution and duration for optimal output.

• Shares insights on creating dynamic scenes where characters interact with their environment effectively, enhancing viewer engagement.