Update app.py

app.py

@@ -5,7 +5,7 @@ import numpy as np
 from transformers import pipeline
 from diffusers.utils import load_image
 from accelerate import Accelerator
-import torch, os, random
+import torch, os, random, gc
 from diffusers import StableDiffusionControlNetPipeline, StableDiffusionPipeline, ControlNetModel, UniPCMultistepScheduler
 from controlnet_aux import OpenposeDetector
 
@@ -126,6 +126,7 @@ def plex(mput, prompt, neg_prompt, stips, modal_id, dula, blip, blop):
     pipe = accelerator.prepare(pipe.to("cpu"))
     
     tilage = pope(prompt,num_inference_steps=5,height=512,width=512,generator=generator).images[0]
+    tilage.save('til.png', 'PNG')
     cannyimage = np.array(tilage)
     low_threshold = 100
     high_threshold = 200
@@ -135,12 +136,15 @@ def plex(mput, prompt, neg_prompt, stips, modal_id, dula, blip, blop):
     cannyimage[:, zero_start:zero_end] = 0
     cannyimage = cannyimage[:, :, None]
     cannyimage = np.concatenate([cannyimage, cannyimage, cannyimage], axis=2)
-    canny_image = Image.fromarray(cannyimage)  
+    canny_image = Image.fromarray(cannyimage)
+    canny_image.save('can.png', 'PNG')
     pose_image = load_image(mput).resize((512, 512))
+    pose_image.save('./pos.png', 'PNG')
     openpose_image = openpose(pose_image)
-    images = [openpose_image, canny_image]
+    openpose_image.save('./fin.png','PNG')
+    ##images = [openpose_image, canny_image]
 
-    imoge = pipe(prompt,images,num_inference_steps=stips,negative_prompt=neg_prompt,controlnet_conditioning_scale=[blip, blop],height=512,width=512,generator=generator).images[0]
+    imoge = pipe(prompt,[openpose_image, canny_image],num_inference_steps=stips,negative_prompt=neg_prompt,controlnet_conditioning_scale=[blip, blop],height=512,width=512,generator=generator).images[0]
     return imoge
 
 iface = gr.Interface(fn=plex,inputs=[gr.Image(type="filepath"), gr.Textbox(label="prompt"), gr.Textbox(label="neg_prompt", value="monochrome, lowres, bad anatomy, worst quality, low quality"), gr.Slider(label="infer_steps", value=20, minimum=1, step=1, maximum=100), gr.Dropdown(choices=models, value=models[0], type="value", label="select a model"), gr.Dropdown(choices=sdulers, value=sdulers[0], type="value", label="schedulrs"), gr.Slider(label="condition_scale_canny", value=0.5, minimum=0.05, step=0.05, maximum=0.95), gr.Slider(label="condition_scale_pose", value=0.5, minimum=0.05, step=0.05, maximum=0.95)], outputs=gr.Image(), title="Img2Img Guided Multi-Conditioned Canny/Pose Controlnet Selectable StableDiffusion Model Demo", description="by JoPmt.")