Commit
β’
e300c6e
1
Parent(s):
1500e0d
Add i2i
Browse files
app.py
CHANGED
@@ -5,9 +5,9 @@ import logging
|
|
5 |
import torch
|
6 |
from PIL import Image
|
7 |
import spaces
|
8 |
-
from diffusers import DiffusionPipeline, AutoencoderTiny, AutoencoderKL
|
9 |
from live_preview_helpers import calculate_shift, retrieve_timesteps, flux_pipe_call_that_returns_an_iterable_of_images
|
10 |
-
|
11 |
from huggingface_hub import hf_hub_download, HfFileSystem, ModelCard, snapshot_download
|
12 |
import copy
|
13 |
import random
|
@@ -25,6 +25,15 @@ base_model = "black-forest-labs/FLUX.1-dev"
|
|
25 |
taef1 = AutoencoderTiny.from_pretrained("madebyollin/taef1", torch_dtype=dtype).to(device)
|
26 |
good_vae = AutoencoderKL.from_pretrained(base_model, subfolder="vae", torch_dtype=dtype).to(device)
|
27 |
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=dtype, vae=taef1).to(device)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
MAX_SEED = 2**32-1
|
30 |
|
@@ -88,7 +97,26 @@ def generate_image(prompt_mash, steps, seed, cfg_scale, width, height, lora_scal
|
|
88 |
):
|
89 |
yield img
|
90 |
|
91 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
92 |
if selected_index is None:
|
93 |
raise gr.Error("You must select a LoRA before proceeding.")
|
94 |
selected_lora = loras[selected_index]
|
@@ -107,32 +135,44 @@ def run_lora(prompt, cfg_scale, steps, selected_index, randomize_seed, seed, wid
|
|
107 |
|
108 |
with calculateDuration("Unloading LoRA"):
|
109 |
pipe.unload_lora_weights()
|
|
|
110 |
|
111 |
# Load LoRA weights
|
112 |
with calculateDuration(f"Loading LoRA weights for {selected_lora['title']}"):
|
113 |
-
if
|
114 |
-
|
|
|
|
|
|
|
115 |
else:
|
116 |
-
|
117 |
-
|
|
|
|
|
|
|
118 |
# Set random seed for reproducibility
|
119 |
with calculateDuration("Randomizing seed"):
|
120 |
if randomize_seed:
|
121 |
seed = random.randint(0, MAX_SEED)
|
122 |
-
|
123 |
-
|
|
|
|
|
|
|
|
|
|
|
124 |
|
125 |
-
|
126 |
-
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
|
131 |
-
|
132 |
-
|
|
|
|
|
133 |
|
134 |
-
yield final_image, seed, gr.update(value=progress_bar, visible=False)
|
135 |
-
|
136 |
def get_huggingface_safetensors(link):
|
137 |
split_link = link.split("/")
|
138 |
if(len(split_link) == 2):
|
@@ -257,6 +297,9 @@ with gr.Blocks(theme=gr.themes.Soft(), css=css, delete_cache=(60, 3600)) as app:
|
|
257 |
|
258 |
with gr.Row():
|
259 |
with gr.Accordion("Advanced Settings", open=False):
|
|
|
|
|
|
|
260 |
with gr.Column():
|
261 |
with gr.Row():
|
262 |
cfg_scale = gr.Slider(label="CFG Scale", minimum=1, maximum=20, step=0.5, value=3.5)
|
@@ -288,7 +331,7 @@ with gr.Blocks(theme=gr.themes.Soft(), css=css, delete_cache=(60, 3600)) as app:
|
|
288 |
gr.on(
|
289 |
triggers=[generate_button.click, prompt.submit],
|
290 |
fn=run_lora,
|
291 |
-
inputs=[prompt, cfg_scale, steps, selected_index, randomize_seed, seed, width, height, lora_scale],
|
292 |
outputs=[result, seed, progress_bar]
|
293 |
)
|
294 |
|
|
|
5 |
import torch
|
6 |
from PIL import Image
|
7 |
import spaces
|
8 |
+
from diffusers import DiffusionPipeline, AutoencoderTiny, AutoencoderKL, AutoPipelineForImage2Image
|
9 |
from live_preview_helpers import calculate_shift, retrieve_timesteps, flux_pipe_call_that_returns_an_iterable_of_images
|
10 |
+
from diffusers.utils import load_image
|
11 |
from huggingface_hub import hf_hub_download, HfFileSystem, ModelCard, snapshot_download
|
12 |
import copy
|
13 |
import random
|
|
|
25 |
taef1 = AutoencoderTiny.from_pretrained("madebyollin/taef1", torch_dtype=dtype).to(device)
|
26 |
good_vae = AutoencoderKL.from_pretrained(base_model, subfolder="vae", torch_dtype=dtype).to(device)
|
27 |
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=dtype, vae=taef1).to(device)
|
28 |
+
pipe_i2i = AutoPipelineForImage2Image.from_pretrained(base_model,
|
29 |
+
vae=good_vae,
|
30 |
+
transformer=pipe.transformer,
|
31 |
+
text_encoder=pipe.text_encoder,
|
32 |
+
tokenizer=pipe.tokenizer,
|
33 |
+
text_encoder_2=pipe.text_encoder_2,
|
34 |
+
tokenizer_2=pipe.tokenizer_2,
|
35 |
+
torch_dtype=dtype
|
36 |
+
)
|
37 |
|
38 |
MAX_SEED = 2**32-1
|
39 |
|
|
|
97 |
):
|
98 |
yield img
|
99 |
|
100 |
+
@spaces.GPU(duration=70)
|
101 |
+
def generate_image_to_image(prompt_mash, image_input_path, image_strength, steps, cfg_scale, width, height, lora_scale, seed):
|
102 |
+
generator = torch.Generator(device="cuda").manual_seed(seed)
|
103 |
+
pipe_i2i.to("cuda")
|
104 |
+
image_input = load_image(image_input_path)
|
105 |
+
final_image = pipe_i2i(
|
106 |
+
prompt=prompt_mash,
|
107 |
+
image=image_input,
|
108 |
+
strength=image_strength,
|
109 |
+
num_inference_steps=steps,
|
110 |
+
guidance_scale=cfg_scale,
|
111 |
+
width=width,
|
112 |
+
height=height,
|
113 |
+
generator=generator,
|
114 |
+
joint_attention_kwargs={"scale": lora_scale},
|
115 |
+
output_type="pil",
|
116 |
+
).images[0]
|
117 |
+
return final_image
|
118 |
+
|
119 |
+
def run_lora(prompt, image_input, image_strength, cfg_scale, steps, selected_index, randomize_seed, seed, width, height, lora_scale, progress=gr.Progress(track_tqdm=True)):
|
120 |
if selected_index is None:
|
121 |
raise gr.Error("You must select a LoRA before proceeding.")
|
122 |
selected_lora = loras[selected_index]
|
|
|
135 |
|
136 |
with calculateDuration("Unloading LoRA"):
|
137 |
pipe.unload_lora_weights()
|
138 |
+
pipe_i2i.unload_lora_weights()
|
139 |
|
140 |
# Load LoRA weights
|
141 |
with calculateDuration(f"Loading LoRA weights for {selected_lora['title']}"):
|
142 |
+
if(image_input is not None):
|
143 |
+
if "weights" in selected_lora:
|
144 |
+
pipe_i2i.load_lora_weights(lora_path, weight_name=selected_lora["weights"])
|
145 |
+
else:
|
146 |
+
pipe_i2i.load_lora_weights(lora_path)
|
147 |
else:
|
148 |
+
if "weights" in selected_lora:
|
149 |
+
pipe.load_lora_weights(lora_path, weight_name=selected_lora["weights"])
|
150 |
+
else:
|
151 |
+
pipe.load_lora_weights(lora_path)
|
152 |
+
|
153 |
# Set random seed for reproducibility
|
154 |
with calculateDuration("Randomizing seed"):
|
155 |
if randomize_seed:
|
156 |
seed = random.randint(0, MAX_SEED)
|
157 |
+
|
158 |
+
if(image_input is not None):
|
159 |
+
|
160 |
+
final_image = generate_image_to_image(prompt_mash, image_input, image_strength, steps, cfg_scale, width, height, lora_scale, seed)
|
161 |
+
yield final_image, seed, gr.update(visible=False)
|
162 |
+
else:
|
163 |
+
image_generator = generate_image(prompt_mash, steps, seed, cfg_scale, width, height, lora_scale, progress)
|
164 |
|
165 |
+
# Consume the generator to get the final image
|
166 |
+
final_image = None
|
167 |
+
step_counter = 0
|
168 |
+
for image in image_generator:
|
169 |
+
step_counter+=1
|
170 |
+
final_image = image
|
171 |
+
progress_bar = f'<div class="progress-container"><div class="progress-bar" style="--current: {step_counter}; --total: {steps};"></div></div>'
|
172 |
+
yield image, seed, gr.update(value=progress_bar, visible=True)
|
173 |
+
|
174 |
+
yield final_image, seed, gr.update(value=progress_bar, visible=False)
|
175 |
|
|
|
|
|
176 |
def get_huggingface_safetensors(link):
|
177 |
split_link = link.split("/")
|
178 |
if(len(split_link) == 2):
|
|
|
297 |
|
298 |
with gr.Row():
|
299 |
with gr.Accordion("Advanced Settings", open=False):
|
300 |
+
with gr.Row():
|
301 |
+
input_image = gr.Image(label="Input image", type="filepath")
|
302 |
+
image_strength = gr.Slider(label="Image Strength", minimum=0.1, maximum=1.0, step=0.01, value=0.75)
|
303 |
with gr.Column():
|
304 |
with gr.Row():
|
305 |
cfg_scale = gr.Slider(label="CFG Scale", minimum=1, maximum=20, step=0.5, value=3.5)
|
|
|
331 |
gr.on(
|
332 |
triggers=[generate_button.click, prompt.submit],
|
333 |
fn=run_lora,
|
334 |
+
inputs=[prompt, input_image, image_strength, cfg_scale, steps, selected_index, randomize_seed, seed, width, height, lora_scale],
|
335 |
outputs=[result, seed, progress_bar]
|
336 |
)
|
337 |
|