File size: 6,428 Bytes
dbc5b70 9827f59 dbc5b70 9827f59 dbc5b70 9827f59 dbc5b70 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 |
import gradio as gr
import numpy as np
import random
#from diffusers import DiffusionPipeline
from diffusers import AutoPipelineForText2Image, AutoPipelineForImage2Image
#from diffusers.utils import load_image
import torch
from PIL import Image
from huggingface_hub import hf_hub_download
from eSeNTranslate import TranslateFromAny2XModel
# Load translation model(s) (Pipeline)
fasttextModelPath = hf_hub_download(repo_id="facebook/fasttext-language-identification", filename="model.bin")
translatePipe = TranslateFromAny2XModel(nllb_model_path="facebook/nllb-200-distilled-600M", fasttext_model_path=fasttextModelPath)
modelPath = "stabilityai/sdxl-turbo"
if torch.cuda.is_available():
device = "cuda"
torch.cuda.max_memory_allocated(device=device)
#pipe = DiffusionPipeline.from_pretrained(modelPath, torch_dtype=torch.float16, variant="fp16", use_safetensors=True)
pipeTex2Image = AutoPipelineForText2Image.from_pretrained(modelPath, torch_dtype=torch.float16, variant="fp16")
pipeImage2Image = AutoPipelineForImage2Image.from_pretrained(modelPath, torch_dtype=torch.float16, variant="fp16")
#pipe.enable_xformers_memory_efficient_attention()
pipeTex2Image.enable_xformers_memory_efficient_attention()
pipeImage2Image.enable_xformers_memory_efficient_attention()
#pipe = pipe.to(device)
else:
device = "cpu"
#pipe = DiffusionPipeline.from_pretrained(modelPath, use_safetensors=True)
pipeTex2Image = AutoPipelineForText2Image.from_pretrained(modelPath, use_safetensors=True)
pipeImage2Image = AutoPipelineForImage2Image.from_pretrained(modelPath, use_safetensors=True)
#pipe = pipe.to(device)
#pipe = pipe.to(device)
pipeTex2Image.to(device)
pipeImage2Image.to(device)
MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024
def infer(prompt, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, use_as_input, strength, image):
if randomize_seed:
seed = random.randint(0, MAX_SEED)
generator = torch.Generator().manual_seed(seed)
prompt = translatePipe.generate(prompt)
if use_as_input:
print("Image to Image:")
pipe = pipeImage2Image
init_image = Image.fromarray(np.uint8(image)).resize((width, height)).convert("RGB")
init_image.save("input.png", format="PNG")
print(type(init_image), init_image.size)
image = pipe(
prompt = prompt,
negative_prompt = negative_prompt,
guidance_scale = guidance_scale,
num_inference_steps = num_inference_steps,
width = width,
height = height,
generator = generator,
strength=strength,
image=init_image
).images[0]
else:
print("Text to Image:")
pipe = pipeTex2Image
image = pipe(
prompt = prompt,
negative_prompt = negative_prompt,
guidance_scale = guidance_scale,
num_inference_steps = num_inference_steps,
width = width,
height = height,
generator = generator
).images[0]
return image
examples = [
"Face of a modern woman of Balkan descent 25 years old",
"Blue car sandero stepway on dirt road",
"Cow in the skin of a dog of dalmatian breed",
]
css="""
#col-container {
margin: 0 auto;
max-width: auto;
}
"""
with gr.Blocks(css=css) as app:
with gr.Column(elem_id="col-container"):
gr.Markdown(f"""
# Text-to-Image, Image-to-Image by Slavko Novak
Currently running on {device}.
""")
with gr.Row():
prompt = gr.Text(
label="Prompt",
show_label=False,
max_lines=1,
placeholder="Enter your prompt",
container=False,
)
run_button = gr.Button("Generate", scale=0)
result = gr.Image(label="Result", show_label=False)
use_as_input = gr.Checkbox(label="Use image as input", value=False)
with gr.Accordion("Advanced Settings", open=False):
negative_prompt = gr.Text(
label="Negative prompt",
max_lines=1,
placeholder="Enter a negative prompt",
visible=True,
)
seed = gr.Slider(
label="Seed",
minimum=0,
maximum=MAX_SEED,
step=1,
value=0,
)
randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
with gr.Row():
width = gr.Slider(
label="Width",
minimum=256,
maximum=MAX_IMAGE_SIZE,
step=32,
value=512,
)
height = gr.Slider(
label="Height",
minimum=256,
maximum=MAX_IMAGE_SIZE,
step=32,
value=512,
)
with gr.Row():
guidance_scale = gr.Slider(
label="Guidance scale",
minimum=0.0,
maximum=10.0,
step=0.1,
value=0.0,
)
strength = gr.Slider(
label="Strength scale",
minimum=0.0,
maximum=1.0,
step=0.1,
value=0.5,
)
num_inference_steps = gr.Slider(
label="Number of inference steps",
minimum=1,
maximum=12,
step=1,
value=2,
)
gr.Examples(
examples = examples,
inputs = [prompt]
)
run_button.click(
fn = infer,
inputs = [prompt, negative_prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps, use_as_input, strength, result],
outputs = [result]
)
#app.queue().launch(server_name="0.0.0.0", server_port=8080, share=True)
app.queue().launch()
|