"RealCartoon Pixar" is an AI Cartoon Generator
Creating cartoon images is a classic endeavor for AI, but achieving Pixar-like cartoon images from users' text descriptions is still challenging. While generating cartoon images from text is feasible, the synthesized images are confined to a limited number of objects. To address the challenge of generating a full set of Pixar-like cartoon images, we propose RealCartoon-Pixar, a new image generation model that leverages the multimodal capability of our core engine and novel Deep Transformer Networks as our Cartoon-Conditioned Computer Vision models. We transform a complex image generation task into two generative models: Text-to-Cartoon and Cartoon-to-Image. The new architecture can directly manipulate cartoon images and facilitate innovation while minimizing time and effort. The deployment of the new model is demonstrated through highly visible communication of geometric remand shapes.
Modeling cartoon translations from sketches to fully colored images in the pixel domain—driven by users' text descriptions—is a well-settled task. However, generating the full set of Pixar-like cartoon images from text descriptions is still a hard nut to crack. Previous works rely on generative images with no structures, i.e., blobby shapes. Designing geometric shapes from scratch imposes a big challenge on laymen, and the processing time entails lots of computation. Focused on semantic hierarchical modeling, RealCartoon-Pixar becomes a powerful tool to facilitate the layout of cartoon pictures through cartoon images that are easily generated from users' texts. With innovative geometric designs, a new set of cartoon images is created rapidly to complete the designated tasks.
Use RealCartoon Pixar with API request
RealCartoon is an AI model to generate cartoon images from text. It accepts a caption and then generates a relative cartoon image. The cartoon can be an object, scene, animal, or a mixture of backgrounds with non-existing characters or objects that do not exist in nature.
This API is pre-released and only supports business customers! If you are an individual, stay tuned for the open version!
Request Parameters
This API only supports POST methods.
Request Body
{ "caption": "A cartoon about a skier who is skiing.", "model": "Pixar"}
caption: The text input to generate an image. This is an English sentence/dialogue that contains potential cartoon content. Text in Chinese is supported, but English is recommended as the first choice.Maximum Length: 500
model: Model NamePixar: The Pixar model is trained on combinations of cartoons with various features including objects, scenes, persons, animals, and backgrounds. When you enter an English sentence that contains a specific object, scene, person, animal, or background, the model then searches for similar real images and generates the final cartoon according to the sentence. If the entered sentence does not contain this valid feature, the model is mixed with the specified caption object and other features for cartoon image generation.
Use RealCartoon Pixar with an interface
RealCartoon can generate animated images from any text. You will get a cartoon, no matter in which language. The user interface does not require you to know any computer programming language. For example, you can write, "an old castle near the green everlasting forest," and get a colorful picture in an animated style. Also, RealCartoon is capable of generating cartoon characters and objects such as houses or railroads from descriptions. It can be used as both professional graphic design software and entertainment. The main feature of the model is interpretability. This means that the model works quickly.
With the model we provide a service to create an illustration or animation in which objects will be placed according to the style of the user. The final product can be easily revised: the user can choose the aspect and size of the objects, the general style, the color scheme of the background, and whether there will be clouds in the sky. The model accepts the signed contract and starts working. A possible case of the model's functioning is represented in the figures, in which we offer an amateur professional-looking for the desk of a person who has just finished a much-awaited job. In summits, trade displays, and other relevant events, the model will look at these displays.
