Update app.py

app.py

@@ -1,5 +1,6 @@
 import gradio as gr
 import spaces
+import random
 
 import torch
 from diffusers import StableDiffusionXLPipeline
@@ -12,9 +13,14 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
 good_vae = AutoencoderKL.from_pretrained("bobber/bigasp2", subfolder="vae", torch_dtype=dtype).to(device)
 pipeline = StableDiffusionXLPipeline.from_pretrained("bobber/bigasp2", torch_dtype=dtype, vae=good_vae).to(device)
 
+MAX_SEED = np.iinfo(np.int32).max
+
 @spaces.GPU
-def generate(prompt, negative_prompt, width, height, sample_steps, guidance_scale):
-    return pipeline(prompt=prompt, negative_prompt=negative_prompt, width=width, height=height, guidance_scale=guidance_scale, num_inference_steps=sample_steps).images[0]
+def generate(prompt, negative_prompt, width, height, sample_steps, guidance_scale, seed):
+    if seed ==0:
+        seed = random.randint(0, MAX_SEED)
+    generator = torch.Generator().manual_seed(seed)
+    return pipeline(prompt=prompt, generator=generator, negative_prompt=negative_prompt, width=width, height=height, guidance_scale=guidance_scale, num_inference_steps=sample_steps).images[0]
 
 with gr.Blocks() as interface:
         with gr.Column():
@@ -33,10 +39,11 @@ with gr.Blocks() as interface:
                             height = gr.Slider(label="Height", info="The height in pixels of the generated image.", value=1024, minimum=128, maximum=4096, step=64, interactive=True)
                         with gr.Column():
                             sampling_steps = gr.Slider(label="Sampling Steps", info="The number of denoising steps.", value=40, minimum=4, maximum=50, step=1, interactive=True)
+                            seed = gr.Slider(label="Seed", minimum=0, maximum=MAX_SEED, step=1, value=0,)
                         with gr.Column():
                             guidance_scale = gr.Slider(label="Guidance Scale", info="Guidance scale.", value=2.5, minimum=1, maximum=10, step=0.1, interactive=True)
         
-        generate_button.click(fn=generate, inputs=[prompt, negative_prompt, width, height, sampling_steps, guidance_scale], outputs=[output])
+        generate_button.click(fn=generate, inputs=[prompt, negative_prompt, width, height, sampling_steps, guidance_scale, seed], outputs=[output])
 
 if __name__ == "__main__":
     interface.launch()