Fabrice-TIERCELIN
commited on
Commit
•
93e1d77
1
Parent(s):
0879b61
Fix function
Browse files- gradio_demo.py +15 -17
gradio_demo.py
CHANGED
@@ -160,7 +160,6 @@ def stage2_process(
|
|
160 |
if 1 < downscale:
|
161 |
input_height, input_width, input_channel = np.array(input_image).shape
|
162 |
input_image = input_image.resize((input_width // downscale, input_height // downscale), Image.LANCZOS)
|
163 |
-
torch.cuda.set_device(SUPIR_device)
|
164 |
event_id = str(time.time_ns())
|
165 |
event_dict = {'event_id': event_id, 'localtime': time.ctime(), 'prompt': prompt, 'a_prompt': a_prompt,
|
166 |
'n_prompt': n_prompt, 'num_samples': num_samples, 'upscale': upscale, 'edm_steps': edm_steps,
|
@@ -181,23 +180,8 @@ def stage2_process(
|
|
181 |
input_image = upscale_image(input_image, upscale, unit_resolution=32,
|
182 |
min_size=min_size)
|
183 |
|
184 |
-
LQ = np.array(input_image) / 255.0
|
185 |
-
LQ = np.power(LQ, gamma_correction)
|
186 |
-
LQ *= 255.0
|
187 |
-
LQ = LQ.round().clip(0, 255).astype(np.uint8)
|
188 |
-
LQ = LQ / 255 * 2 - 1
|
189 |
-
LQ = torch.tensor(LQ, dtype=torch.float32).permute(2, 0, 1).unsqueeze(0).to(SUPIR_device)[:, :3, :, :]
|
190 |
-
if use_llava:
|
191 |
-
captions = [prompt]
|
192 |
-
else:
|
193 |
-
captions = ['']
|
194 |
-
|
195 |
-
model.ae_dtype = convert_dtype(ae_dtype)
|
196 |
-
model.model.dtype = convert_dtype(diff_dtype)
|
197 |
-
|
198 |
samples = restore(
|
199 |
model,
|
200 |
-
LQ,
|
201 |
captions,
|
202 |
edm_steps,
|
203 |
s_stage1,
|
@@ -255,7 +239,6 @@ def stage2_process(
|
|
255 |
@spaces.GPU(duration=600)
|
256 |
def restore(
|
257 |
model,
|
258 |
-
LQ,
|
259 |
captions,
|
260 |
edm_steps,
|
261 |
s_stage1,
|
@@ -273,6 +256,21 @@ def restore(
|
|
273 |
spt_linear_CFG,
|
274 |
spt_linear_s_stage2
|
275 |
):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
276 |
return model.batchify_sample(LQ, captions, num_steps=edm_steps, restoration_scale=s_stage1, s_churn=s_churn,
|
277 |
s_noise=s_noise, cfg_scale=s_cfg, control_scale=s_stage2, seed=seed,
|
278 |
num_samples=num_samples, p_p=a_prompt, n_p=n_prompt, color_fix_type=color_fix_type,
|
|
|
160 |
if 1 < downscale:
|
161 |
input_height, input_width, input_channel = np.array(input_image).shape
|
162 |
input_image = input_image.resize((input_width // downscale, input_height // downscale), Image.LANCZOS)
|
|
|
163 |
event_id = str(time.time_ns())
|
164 |
event_dict = {'event_id': event_id, 'localtime': time.ctime(), 'prompt': prompt, 'a_prompt': a_prompt,
|
165 |
'n_prompt': n_prompt, 'num_samples': num_samples, 'upscale': upscale, 'edm_steps': edm_steps,
|
|
|
180 |
input_image = upscale_image(input_image, upscale, unit_resolution=32,
|
181 |
min_size=min_size)
|
182 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
183 |
samples = restore(
|
184 |
model,
|
|
|
185 |
captions,
|
186 |
edm_steps,
|
187 |
s_stage1,
|
|
|
239 |
@spaces.GPU(duration=600)
|
240 |
def restore(
|
241 |
model,
|
|
|
242 |
captions,
|
243 |
edm_steps,
|
244 |
s_stage1,
|
|
|
256 |
spt_linear_CFG,
|
257 |
spt_linear_s_stage2
|
258 |
):
|
259 |
+
torch.cuda.set_device(SUPIR_device)
|
260 |
+
LQ = np.array(input_image) / 255.0
|
261 |
+
LQ = np.power(LQ, gamma_correction)
|
262 |
+
LQ *= 255.0
|
263 |
+
LQ = LQ.round().clip(0, 255).astype(np.uint8)
|
264 |
+
LQ = LQ / 255 * 2 - 1
|
265 |
+
LQ = torch.tensor(LQ, dtype=torch.float32).permute(2, 0, 1).unsqueeze(0).to(SUPIR_device)[:, :3, :, :]
|
266 |
+
if use_llava:
|
267 |
+
captions = [prompt]
|
268 |
+
else:
|
269 |
+
captions = ['']
|
270 |
+
|
271 |
+
model.ae_dtype = convert_dtype(ae_dtype)
|
272 |
+
model.model.dtype = convert_dtype(diff_dtype)
|
273 |
+
|
274 |
return model.batchify_sample(LQ, captions, num_steps=edm_steps, restoration_scale=s_stage1, s_churn=s_churn,
|
275 |
s_noise=s_noise, cfg_scale=s_cfg, control_scale=s_stage2, seed=seed,
|
276 |
num_samples=num_samples, p_p=a_prompt, n_p=n_prompt, color_fix_type=color_fix_type,
|