blocks/text2img.py

import gradio as gr
import torch
from diffusers import StableDiffusionPipeline
from .utils.schedulers import SCHEDULER_LIST, get_scheduler_list
from .utils.prompt2prompt import generate
from .utils.device import get_device
from .download import get_share_js, community_icon_html, loading_icon_html, CSS

#--- create a download button that takes the output image from gradio and downloads it

TEXT2IMG_MODEL_LIST = {
    "OpenJourney v4" : "prompthero/openjourney-v4",
    "StableDiffusion 1.5" : "runwayml/stable-diffusion-v1-5",
    "StableDiffusion 2.1" : "stabilityai/stable-diffusion-2-1",
    "DreamLike 1.0" : "dreamlike-art/dreamlike-diffusion-1.0",
    "DreamLike 2.0" : "dreamlike-art/dreamlike-photoreal-2.0",
    "DreamShaper" : "Lykon/DreamShaper",
    "NeverEnding-Dream" : "Lykon/NeverEnding-Dream"
}

class StableDiffusionText2ImageGenerator:
    def __init__(self):
        self.pipe = None

    def load_model(
        self,
        model_path,
        scheduler
    ):
        model_path = TEXT2IMG_MODEL_LIST[model_path]
        if self.pipe is None:
            self.pipe = StableDiffusionPipeline.from_pretrained(
                model_path, safety_checker=None, torch_dtype=torch.float32
            )

        device = get_device()
        self.pipe = get_scheduler_list(pipe=self.pipe, scheduler=scheduler)
        self.pipe.to(device)
        self.pipe.enable_attention_slicing()

        return self.pipe

    def generate_image(
        self,
        model_path: str,
        prompt: str,
        negative_prompt: str,
        scheduler: str,
        guidance_scale: int,
        num_inference_step: int,
        height: int,
        width: int,
        seed_generator=0,
    ):
        print("model_path", model_path)
        print("prompt", prompt)
        print("negative_prompt", negative_prompt)
        print("num_images_per_prompt", 1)
        print("scheduler", scheduler)
        print("guidance_scale", guidance_scale)
        print("num_inference_step", num_inference_step)
        print("height", height)
        print("width", width)
        print("seed_generator", seed_generator)

        pipe = self.load_model(
            model_path=model_path,
            scheduler=scheduler,
        )
        if seed_generator == 0:
            random_seed = torch.randint(0, 1000000, (1,))
            generator = torch.manual_seed(random_seed)
        else:
            generator = torch.manual_seed(seed_generator)

        images = pipe(
            prompt=prompt,
            height=height,
            width=width,
            negative_prompt=negative_prompt,
            num_images_per_prompt=1,
            num_inference_steps=num_inference_step,
            guidance_scale=guidance_scale,
            generator=generator,
        ).images

        return images


    def app(username : str = "admin"):
        demo = gr.Blocks(css = CSS)
        with demo:
            with gr.Row():
                with gr.Column():
                    text2image_prompt = gr.Textbox(
                        lines=1,
                        show_label=False,
                        elem_id="prompt-text-input",
                        value='',
                        placeholder="Prompt, keywords that describe your image"
                        )

                    text2image_negative_prompt = gr.Textbox(
                        lines=1,
                        show_label=False,
                        elem_id = "negative-prompt-text-input",
                        value='',
                        placeholder="Negative Prompt, keywords that describe what you don't want in your image",
                    )

                    # add button for generating a prompt from the prompt
                    text2image_prompt_generate_button = gr.Button(
                        label="Generate Prompt",
                        type="primary",
                        align="center",
                        value = "Generate Prompt"
                    )

                    # show a text box with the generated prompt
                    text2image_prompt_generated_prompt = gr.Textbox(
                        lines=1,
                        placeholder="Generated Prompt",
                        show_label=False,
                        info="Auto generated prompts for inspiration.",
                    )

                    text2image_model_path = gr.Dropdown(
                        choices=list(TEXT2IMG_MODEL_LIST.keys()),
                        value=list(TEXT2IMG_MODEL_LIST.keys())[0],
                        label="Text2Image Model Selection",
                        elem_id="model-dropdown",
                        info="Select the model you want to use for text2image generation."
                    )

                    text2image_scheduler = gr.Dropdown(
                        choices=SCHEDULER_LIST,
                        value=SCHEDULER_LIST[0],
                        label="Scheduler",
                        elem_id="scheduler-dropdown",
                        info="Scheduler list for models. Different schdulers result in different outputs."
                    )



                    text2image_size = gr.Slider(
                        minimum=128,
                        maximum=1280,
                        step=32,
                        value=768,
                        label="Image Size",
                        elem_id="image-size-slider",
                        info = "Image size determines the height and width of the generated image. Higher the value, better the quality however slower the computation."
                    )
                    text2image_seed_generator = gr.Slider(
                        label="Seed(0 for random)",
                        minimum=0,
                        maximum=1000000,
                        value=0,
                        elem_id="seed-slider",
                        info="Set the seed to a specific value to reproduce the results."
                    )


                    text2image_guidance_scale = gr.Slider(
                        minimum=0.1,
                        maximum=15,
                        step=0.1,
                        value=7.5,
                        label="Guidance Scale",
                        elem_id = "guidance-scale-slider",
                        info = "Guidance scale determines how much the prompt will affect the image. Higher the value, more the effect."
                    )

                    text2image_num_inference_step = gr.Slider(
                        minimum=1,
                        maximum=100,
                        step=1,
                        value=50,
                        label="Num Inference Step",
                        elem_id = "num-inference-step-slider",
                        info = "Number of inference step determines the quality of the image. Higher the number, better the quality."
                    )
                    text2image_predict = gr.Button(value="Generate image") 

                with gr.Column():
                    output_image = gr.Gallery(
                        label="Generated images",
                        show_label=False,
                        elem_id="gallery",
                    ).style(grid=(1, 2), height='auto')

                    with gr.Group(elem_id="container-advanced-btns"):
                        with gr.Group(elem_id="share-btn-container"):
                            community_icon = gr.HTML(community_icon_html)
                            loading_icon = gr.HTML(loading_icon_html)
                            share_button = gr.Button("Save artwork", elem_id="share-btn")

                    gr.HTML(
                        """
                        <div id="model-description-text2img">
                            <h3>Text2Image Models</h3>
                            <p>Text to image models will generate an image guided by the prompt that is provided</p>
                            <p>A prompt should be specified with keywords that describe the image you want to generate.</p>
                            <p>Negative prompt can be used to specify keywords that you don't want in your image such as "blood" or "violence".</p>
                            <p>Example prompt: "A painting of a cat sitting on a chair, fantasy themed, starry background"</p>
                            <hr>
                            <p>Stable Diffusion 1.5 & 2.1: Default model for many tasks. </p>
                            <p>OpenJourney v4: Generates fantasy themed images similar to the Midjourney model. </p>
                            <p>Dreamlike Photoreal 1.0 & 2.0 is SD 1.5 that generates realistic images. </p>
                        </div>
                        """
                    )
            text2image_predict.click(
                fn=StableDiffusionText2ImageGenerator().generate_image,
                inputs=[
                    text2image_model_path,
                    text2image_prompt,
                    text2image_negative_prompt,
                    text2image_scheduler,
                    text2image_guidance_scale,
                    text2image_num_inference_step,
                    text2image_size,
                    text2image_size,
                    text2image_seed_generator,
                ],
                outputs=output_image,
            )

            text2image_prompt_generate_button.click(
                fn=generate,
                inputs=[text2image_prompt],
                outputs=[text2image_prompt_generated_prompt],
            )

            # share_button.click(
            # None,
            # [],
            # [],
            # _js=get_share_js(),
            # )

            # autoclik the share button
            
            

        return demo