Tailored Generation

Description

This route allows you to generate images using a tailored model. tailored models are trained using your own datasets to create unique illustrative visual styles or single-subject illustrations. You can train a tailored model through the API or using the Bria platform.

Overview of Tailored Model API

The Tailored Model API provides a set of endpoints to manage the entire lifecycle of a Tailored model:

1. Dataset Management: Create and manage datasets which will be used to train your tailored models. You can:

   • Create and Retrieve Datasets: Use the /datasets endpoint to create a new dataset or retrieve existing datasets that belong to the client organization.
   • Upload and Retrieve Images: Use the /datasets/{dataset_id}/images endpoint to upload a new image to a dataset or retrieve existing images that belong to the dataset.
   • Manage Individual Images: Use the /datasets/{dataset_id}/images/{image_id} endpoint to get or delete a specific image from a dataset.

2. Model Management: Create and manage models based on your datasets. You can:

   • Create and Retrieve Models: Use the /models endpoint to create a new model or retrieve all models that belong to the client organization.
   • Get and Delete Specific Models: Use the /models/{id} endpoint to retrieve a specific model by its ID or delete a specific model.

3. Training Management: Start and stop training jobs for your models. You can:

   • Start Training: Use the /models/{id}/start_training endpoint to initiate the training process. Note that the dataset must include the minimum required number of images and have a status of 'completed' before training can start.
   • Stop Training: Use the /models/{id}/stop_training endpoint to stop a training job for a model.

4. Image Generation: Generate images using a trained model. You can:

   • Generate Images: Use the /text-to-image/tailored/{model_id} endpoint to generate images.
   • Generate Vector Graphics: Use the /text-to-vector/tailored/{model_id} endpoint to generate vector graphics.

Training a Tailored Model

To train a tailored model via the API:

1. Create a Dataset: Use the /datasets endpoint to create a dataset.
2. Upload Images: Upload your images to this dataset using the /datasets/{dataset_id}/images endpoint.
3. Create a Model: Use the /models endpoint to create a model associated with your dataset.
4. Complete the Dataset: Ensure the dataset meets the minimum image requirements and change its status to 'completed'.
5. Start Training: Initiate the training process using the /models/{id}/start_training endpoint. Training of a tailored model takes between 1-3 hours.
6. Monitor Training: Optionally, you can check the training status to monitor progress.
7. Generate Images: Once the model is trained, use this route to generate images, or use the /text-to-vector/tailored/{model_id} route to generate vector graphics.

Alternatively, you can use the Bria platform to manage and train your tailored models through a user-friendly interface. Access the Bria platform here.

Guidance Methods

This API supports various guidance methods to provide greater control over text-to-image generation. These methods condition the model on additional inputs derived from user-provided images, allowing for more precise and controlled image generation.

Importantly, users can provide any image they want as input. Our pipeline automatically transforms the input image according to the selected guidance method. For example, if you choose the depth guidance method, our pipeline will extract a depth map from your provided image and use it during the generation inference.

Currently, we support the following guidance methods:

1. controlnet_canny: Extracts edge information from the input image to guide generation based on structural outlines.

2. controlnet_depth: Derives depth information from the input image to influence spatial arrangement in generated images.

3. controlnet_recoloring: Uses grayscale version of the input image to guide recoloring while preserving geometry.

4. controlnet_color_grid: Extracts a 16x16 color grid from the input image to guide the overall color scheme of generated images.

Using Guidance Methods

You can specify up to four guidance methods in a single request. Each method requires an accompanying image and a scale parameter to determine its impact on the generation inference. The table below provides detailed information about each guidance method, with an example os use:

Guidance Method	Prompt	Scale	Input Image	Guidance Image	Output Image
ControlNet Canny	An exotic colorful shell on the beach	1.0
ControlNet Depth	A dog, exploring an alien planet	0.8
ControlNet Recoloring	A vibrant photo of a woman	1.00
ControlNet Color Grid	A dynamic fantasy illustration of an erupting volcano	0.7

Note: When using multiple guidance methods, all input images must have the same aspect ratio, which will determine the aspect ratio of the generated results.

To use a guidance method, include the following parameters in your request:

• guidance_method_X: Specify the guidance method (where X is 1, 2). If the paramter guidance_method_2 is used, so does guidance_method_1 has to be used, and so on. If you would like to use only one method, use the paratmer guidance_method_1

• guidance_method_X_scale: Set the impact of the guidance (0.0 to 1.0)

• guidance_method_X_image_file: Provide the base64-encoded input image

By leveraging these guidance methods, you can achieve more precise control over the generated images, tailoring them to your specific needs and creative vision.

Description

This route allows you to generate vector graphics using a tailored model. Tailored models are trained using your own datasets to create unique illustrative visual styles or single-subject illustrations. To see a detailed description of the tailored models' functionalities, please refer to the /text-to-image/tailored/{model_id} route documentation. *Text-to-vector is compatible with tailored models in the illustrative domain.

Guidance Methods

This API supports various guidance methods to provide greater control over text-to-image generation. These methods condition the model on additional inputs derived from user-provided images, allowing for more precise and controlled image generation.

Importantly, users can provide any image they want as input. Our pipeline automatically transforms the input image according to the selected guidance method. For example, if you choose the depth guidance method, our pipeline will extract a depth map from your provided image and use it during the generation inference.

Currently, we support the following guidance methods:

1. controlnet_canny: Extracts edge information from the input image to guide generation based on structural outlines.
2. controlnet_depth: Derives depth information from the input image to influence spatial arrangement in generated images.
3. controlnet_recoloring: Uses grayscale version of the input image to guide recoloring while preserving geometry.
4. controlnet_color_grid: Extracts a 16x16 color grid from the input image to guide the overall color scheme of generated images.

Using Guidance Methods You can specify up to four guidance methods in a single request. Each method requires an accompanying image and a scale parameter to determine its impact on the generation inference. Note: When using multiple guidance methods, all input images must have the same aspect ratio, which will determine the aspect ratio of the generated results. To use a guidance method, include the following parameters in your request:

• guidance_method_X: Specify the guidance method (where X is 1, 2). If the paramter guidance_method_2 is used, so does guidance_method_1 has to be used, and so on. If you would like to use only one method, use the paratmer guidance_method_1
• guidance_method_X_scale: Set the impact of the guidance (0.0 to 1.0)
• guidance_method_X_image_file: Provide the base64-encoded input image

By leveraging these guidance methods, you can achieve more precise control over the generated images, tailoring them to your specific needs and creative vision.

Description This endpoint allows the creation of a new dataset that will belong to the organization associated with the used API token. The dataset will be created with a default status of "draft".

Description Retrieve a list of datasets that belong to the organization associated with the used API token. If no datasets are found, an empty array is returned.

Description Retrieve a specific dataset by its ID. The response includes all the details of the dataset and an array of associated images.

Description Update the details of a specific dataset by its ID. If the status of the dataset is completed, the style_description and use_case cannot be updated. Additionally, if the status is completed, it cannot be returned to draft if a model has already been trained based on it.

Description Delete a specific dataset. Changes status to deleted.

Description Upload a new image to a specific dataset. One of the following parameters, file or image_url, must be provided.

Upload a new image using the image file

In our documentation request example, we demonstrate how to upload a new image using the image URL.

If you want to use an image from a file, follow the example below:

  import requests

  url = "/tailored-gen/datasets/{dataset_id}/images"

  payload = {}
  files=[
    ('file',('image_name.jpeg',open('/path_to_local_file/image_name.jpeg','rb'),'image/jpeg'))
  ]
  headers = {
    'api_token': 'xxxx'
  }

  response = requests.request("POST", url, headers=headers, data=payload, files=files)

  print(response.text)

Description Retrieve a list of images in a specific dataset.

Description Retrieve a specific image by its ID.

Description Delete a specific image.

Description Create a new model.

Description Retrieve a list of models. If there are no models, an empty array is being returned.

Description Retrieve a specific model by its ID.

Description Delete a specific model.

Description Start a new training job for a specific model. Training of a tailored model takes between 1-3 hours.

Description Stop a specific training job for a model.