System and methods to facilitate content generation using generative artificial intelligence models
The present disclosure is directed to systems and methods to enhance the process of creating an artificial intelligence (AI) generated content or content items, such as images, text, video, sounds, etc., using a text or other suitable prompt, such as via voice input. The systems and methods disclosed provide streamlined content generation with, e.g., reduced processing power and computing time. In an embodiment the systems and methods receive a prompt for generating a first content item using a generative artificial intelligence (AI) model and retrieve, based on the prompt, a collection of matching content items. The systems and methods may then receive input selecting one of the content items from the collection and identify a prompt used to generate the selected content item. The systems and methods may then merge using a trained natural language processing model, the received prompt with the prompt of the selected content item to create a third prompt. In an embodiment the systems and methods may modify the third prompt based on additional input and, based on the modified third prompt, generate a second content item.
1 . A method comprising:
receiving, via a user interface, a first prompt for generating a first content item using a first generative artificial intelligence (AI) model;
retrieving, based on the first prompt and from a database of stored AI-generated content items, a plurality of content items;
receiving input selecting a content item from the plurality of retrieved content items;
identifying a second prompt used to generate the selected content item;
merging, using a trained natural language processing model, the first prompt with the second prompt to create a third prompt;
modifying the third prompt based on second input received via the user interface; and
generating, using a second generative AI model and based on the modified third prompt, a second content item.
2 . The method of claim 1 , wherein merging, using the trained natural language processing model, the first prompt with the second prompt to create the third prompt further comprises segmenting the first and second prompts, and designating a main description and modifiers.
3 . The method of claim 1 , further comprising suggesting a suggested model based on the first prompt, and wherein generating the second content item is performed using the suggested model.
4 . The method of claim 1 , further comprising suggesting a suggested sampler based on the first prompt, and wherein generating the second content item is performed using the suggested sampler.
5 . The method of claim 1 , further comprising ranking the plurality of content items based on a similarity score.
6 . The method of claim 5 , wherein the similarity score of a respective content item is based on at least a similarity between an embedding vector of the first prompt and an embedding vector of the respective content item, a similarity between the first prompt and prompts used to generate the plurality of content items using their respective embedding vectors, and an image quality of the respective content item.
7 . The method of claim 1 , wherein the trained natural language processing model includes a sentence merging model.
8 . The method of claim 1 , further comprising displaying metadata of the selected content item upon receiving the input selecting the selected content item.
9 . The method of claim 1 , further comprising in response to the merging, determining generation parameters, and wherein the generating the second content item is further based on the generation parameters.
10 . The method of claim 1 , further comprising receiving a negative prompt, and wherein the merging comprises merging the first prompt, the second prompt, and the negative prompt.
11 . The method of claim 1 further comprising:
generating, based at least in part on a text description for the third prompt and one or more generation parameters for the second generative AI model, a fourth prompt, wherein the fourth prompt comprises one or more of:
the one or more generation parameters in-line with the text description and in a format configured for the second generative AI model; and
computer-readable instructions configured to populate one or more parameter fields of the second generative AI model with the one or more generation parameters and the text description.
12 . The method of claim 1 , wherein the first generative AI model and the second generative AI model are the same.
13 . A system comprising:
memory; and
processing circuitry configured to:
receive, via a user interface, a first prompt for generating a first content item using a first generative artificial intelligence (AI) model;
retrieve, based on the first prompt and from a database of stored AI-generated content items stored in memory, a plurality of content items;
receive input selecting a content item from the plurality of retrieved content items;
identify a second prompt used to generate the selected content item;
merge using a trained natural language processing model, the first prompt with the second prompt to create a third prompt;
modify the third prompt based on second input received via the user interface; and
generate, using a second generative AI model and based on the modified third prompt, a second content item.
14 . The system of claim 13 , wherein the processing circuitry is further configured to merge, using the trained natural language processing model, the first prompt with the second prompt to create the third prompt further by segmenting the first and second prompts, and designating a main description and modifiers.
15 . The system of claim 13 , wherein the processing circuitry is further configured to:
suggest a suggested model based on the first prompt, and
generate the second content item by using the suggested model.
16 . The system of claim 13 , wherein the processing circuitry is further configured to:
suggest a suggested sampler based on the first prompt, and
generate the second content item by using the suggested sampler.
17 . The system of claim 13 , wherein the processing circuitry further configured to rank the plurality of content items based on a similarity score.
18 . The system of claim 17 , wherein the similarity score of a respective content item is based on at least a similarity between an embedding vector of the first prompt and an embedding vector of the respective content item, a similarity between the first prompt and prompts used to generate the plurality of content items using their respective embedding vectors, and an image quality of the respective content item.
19 . The system of claim 13 , wherein the trained natural language processing model includes a sentence merging model.
20 . The system of claim 13 , wherein further the processing circuitry is further configured to display metadata of the selected content item upon receiving the input selecting the selected content item.