Update app.py

app.py

@@ -24,8 +24,7 @@ from lxml.html import fromstring
 from diffusers.utils import export_to_video, load_image
 from huggingface_hub import hf_hub_download
 from safetensors.torch import load_file, save_file
-from diffusers import StableDiffusionPipeline, CogVideoXImageToVideoPipeline, DDIMScheduler #, AnimateDiffPipeline
-from diffusers.models import AutoencoderKL #, MotionAdapter
+from diffusers import StableDiffusionPipeline, CogVideoXImageToVideoPipeline
 from PIL import Image, ImageDraw, ImageFont
 
 # logging
@@ -48,23 +47,20 @@ else:
     device = "cpu"
     dtype = torch.float16
 
-#base = "emilianJR/epiCRealism"
-base = "SG161222/Realistic_Vision_V5.1_noVAE"
-vae_id = "stabilityai/sd-vae-ft-mse"
-#motion_adapter = "guoyww/animatediff-motion-adapter-v1-5-3"
+base = "emilianJR/epiCRealism"
 
 # variable data
 
-last_motion=""
+
 
 # precision data
 
 seq=512
-fast=False
 fps=20
 width=768
 height=768
-step=40
+image_steps=40
+video_steps=20
 accu=7
 
 # ui data
@@ -112,30 +108,14 @@ function custom(){
 # torch pipes
 
 image_pipe = StableDiffusionPipeline.from_pretrained(base, torch_dtype=dtype, safety_checker=None).to(device)
-#adapter = MotionAdapter.from_pretrained(motion_adapter, torch_dtype=dtype, safety_checker=None).to(device)
-vae = AutoencoderKL.from_pretrained(vae_id, torch_dtype=torch.float16).to(device)
-image_pipe.vae = vae
-
-scheduler = DDIMScheduler.from_pretrained(
-    base,
-    subfolder="scheduler",
-    clip_sample=False,
-    timestep_spacing="linspace",
-    beta_schedule="linear",
-    steps_offset=1,
-)
-
 video_pipe = CogVideoXImageToVideoPipeline.from_pretrained(
     "THUDM/CogVideoX-5b-I2V",
     torch_dtype=torch.bfloat16
 ).to(device)
-video_pipe.scheduler = scheduler
 
 video_pipe.vae.enable_tiling()
 video_pipe.vae.enable_slicing()
-#pipe.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
 video_pipe.enable_model_cpu_offload()
-#pipe.enable_free_init(method="butterworth", use_fast_sampling=fast)
 
 # functionality
 
@@ -199,7 +179,7 @@ def pipe_generate(img,p1,p2,motion,time,title):
             width=width,
             guidance_scale=accu,
             num_images_per_prompt=1,
-            num_inference_steps=step,
+            num_inference_steps=image_steps,
             max_sequence_length=seq,
             need_safetycheck=False,
             generator=torch.Generator(device).manual_seed(int(str(random.random()).split(".")[1]))
@@ -217,20 +197,11 @@ def pipe_generate(img,p1,p2,motion,time,title):
     if time == 0.0:
         return img
 
-    if last_motion != motion:
-        if last_motion != "":
-            pipe.unload_lora_weights()
-        if motion != "":
-            pipe.load_lora_weights(motion, adapter_name="motion")
-            pipe.fuse_lora()
-            pipe.set_adapters("motion", [0.7])
-        last_motion = motion
-   
     return video_pipe(
         prompt=p1,
         negative_prompt=p2,
         image=img,
-        num_inference_steps=step,
+        num_inference_steps=video_steps,
         guidance_scale=accu,
         num_videos_per_prompt=1,
         num_frames=(fps*time),
@@ -245,14 +216,14 @@ def handle_generate(*_inp):
     inp[2] = translate(inp[2],"english")
 
     if inp[2] != "":
-        inp[2] = ", related to: " + inp[2] + "."
+        inp[2] = ", related to: " + inp[2]
 
-    inp[2] = f"The content which is faked, errored, unreal, off topic, pixelated, deformed, and semi-realistic, cgi, 3d, sketch, cartoon, drawing, anime, cropped, out of frame, low quality, textual, jpeg artifacts, ugly, duplicated, weird, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutations, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck content{inp[2]}"
+    inp[2] = f"faked, errored, unreal, off topic, pixelated, deformed, and semi-realistic, cgi, 3d, sketch, cartoon, drawing, anime, cropped, out of frame, low quality, textual, jpeg artifacts, ugly, duplicated, weird, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutations, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck{inp[2]}"
 
     if inp[1] != "":
-        inp[1] = ", related to: " + inp[1] + "."
+        inp[1] = ", related to: " + inp[1]
     
-    inp[1] = f'The content which is photographed, realistic, true, genuine, dynamic poze, authentic, deep field, reasonable, natural, best quality, focused, highly detailed content{inp[1]}'
+    inp[1] = f'photographed, realistic, true, genuine, dynamic poze, authentic, deep field, reasonable, natural, best quality, focused, highly detailed{inp[1]}'
 
     print(f"""
 
@@ -305,27 +276,9 @@ def ui():
                 maximum=600.0,
                 value=0.0,
                 step=5.0,
-                label="MP4/PNG Duration (0s = PNG)"
+                label="Duration (0s = PNG)"
             )
-            motion = gr.Dropdown(
-                label='GIF camera movement',
-                show_label=True,
-                container=False,
-                choices=[
-                    ("(No Effect)", ""),
-                    ("Zoom in", "guoyww/animatediff-motion-lora-zoom-in"),
-                    ("Zoom out", "guoyww/animatediff-motion-lora-zoom-out"),
-                    ("Tilt up", "guoyww/animatediff-motion-lora-tilt-up"),
-                    ("Tilt down", "guoyww/animatediff-motion-lora-tilt-down"),
-                    ("Pan left", "guoyww/animatediff-motion-lora-pan-left"),
-                    ("Pan right", "guoyww/animatediff-motion-lora-pan-right"),
-                    ("Roll left", "guoyww/animatediff-motion-lora-rolling-anticlockwise"),
-                    ("Roll right", "guoyww/animatediff-motion-lora-rolling-clockwise"),
-                ],
-                value="",
-                interactive=True
-            )            
-        with gr.Row(elem_id="col-container"):
+         with gr.Row(elem_id="col-container"):
             with gr.Column():
                 img = gr.Image(label="Upload photo",show_label=True,container=False,type="pil")
             with gr.Column():