File size: 6,475 Bytes
4a4fecc
 
 
 
 
6d97cab
 
 
 
4a4fecc
6d97cab
 
 
 
4a4fecc
6d97cab
4a4fecc
 
 
 
 
 
 
 
 
 
6d97cab
 
 
 
4a4fecc
6d97cab
 
 
4a4fecc
6d97cab
4a4fecc
 
 
 
 
 
 
 
 
 
6d97cab
4a4fecc
 
 
 
 
 
 
6d97cab
 
 
 
4a4fecc
6d97cab
4a4fecc
 
 
 
 
6d97cab
 
4a4fecc
6d97cab
4a4fecc
 
6d97cab
 
4a4fecc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6d97cab
4a4fecc
6d97cab
4a4fecc
 
6d97cab
 
 
4a4fecc
6d97cab
 
4a4fecc
6d97cab
4a4fecc
6d97cab
4a4fecc
6d97cab
 
 
4a4fecc
 
 
 
 
 
 
 
 
 
 
 
6d97cab
4a4fecc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6d97cab
 
 
4a4fecc
6d97cab
 
 
 
4a4fecc
6d97cab
4a4fecc
6d97cab
4a4fecc
 
6d97cab
 
 
 
4a4fecc
6d97cab
4a4fecc
 
 
 
 
 
 
 
 
6d97cab
4a4fecc
 
 
 
 
 
 
 
6d97cab
 
4a4fecc
6d97cab
4a4fecc
 
 
 
 
6d97cab
4a4fecc
 
 
 
 
 
6d97cab
 
4a4fecc
6d97cab
4a4fecc
6d97cab
4a4fecc
 
6d97cab
 
 
4a4fecc
 
 
 
 
 
 
6d97cab
4a4fecc
 
6d97cab
4a4fecc
 
 
 
6d97cab
 
4a4fecc
 
 
 
6d97cab
 
4a4fecc
6d97cab
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
# ==========================================================
#  Stable Diffusion v1-4 — CPU Optimized Diffusion Visualizer
#  REAL images (256×256) on free HuggingFace CPU
#  With: step-by-step latents, PCA path, norm plot, latents decode
# ==========================================================

import gradio as gr
import torch
import numpy as np
from diffusers import StableDiffusionPipeline, DDIMScheduler
from sklearn.decomposition import PCA
import plotly.graph_objects as go
from PIL import Image
import time
import warnings

warnings.filterwarnings("ignore")

# ------------------- CPU SETTINGS -------------------

DEVICE = "cpu"

# Disable MKLDNN for safety (prevents matmul errors on SD)
torch.backends.mkldnn.enabled = False

MODEL_ID = "CompVis/stable-diffusion-v1-4"

PIPE_CACHE = None


# ------------------- LOAD SD MODEL -------------------

def get_pipe():
    global PIPE_CACHE
    if PIPE_CACHE:
        return PIPE_CACHE

    pipe = StableDiffusionPipeline.from_pretrained(
        MODEL_ID,
        torch_dtype=torch.float32,
        low_cpu_mem_usage=True,
    )

    # Replace scheduler with DDIM (better for stepping)
    pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)

    pipe.to(DEVICE)

    # VERY IMPORTANT: disable safety checker to avoid weird errors on CPU
    pipe.safety_checker = lambda images, clip_input: (images, False)

    # Disable features not needed
    pipe.enable_attention_slicing(None)

    PIPE_CACHE = pipe
    return PIPE_CACHE


# ------------------- PCA + NORM -------------------

def compute_pca(latents):
    flat = [x.flatten() for x in latents]
    X = np.stack(flat)
    if X.shape[0] < 2:
        return np.zeros((X.shape[0], 2))
    try:
        pca = PCA(n_components=2)
        pts = pca.fit_transform(X)
        return pts
    except:
        return np.zeros((X.shape[0], 2))


def compute_norm(latents):
    return [float(np.linalg.norm(x.flatten())) for x in latents]


# ------------------- LATENT DECODER -------------------

def decode_latent(pipe, latent_np):
    latent = torch.from_numpy(latent_np).unsqueeze(0).to(DEVICE)
    scale = pipe.vae.config.scaling_factor
    with torch.no_grad():
        image = pipe.vae.decode(latent / scale).sample
        image = (image / 2 + 0.5).clamp(0, 1)
    np_img = (image[0].permute(1, 2, 0).cpu().numpy() * 255).astype("uint8")
    return Image.fromarray(np_img)


# ------------------- RUN DIFFUSION -------------------

def run_diffusion(prompt, steps, guidance, seed, simple):

    if not prompt.strip():
        return None, "Enter prompt", gr.update(), None, None, None, {}

    pipe = get_pipe()

    generator = torch.Generator("cpu").manual_seed(seed if seed >= 0 else int(time.time()))

    latents_list = []
    timesteps = []

    def cb(step, t, latents):
        latents_list.append(latents.detach().cpu().numpy()[0])
        timesteps.append(int(t))

    t0 = time.time()

    result = pipe(
        prompt,
        height=256,
        width=256,
        num_inference_steps=steps,
        guidance_scale=guidance,
        generator=generator,
        callback=cb,
        callback_steps=1,
    )

    total = time.time() - t0

    final = result.images[0]

    pca = compute_pca(latents_list)
    norms = compute_norm(latents_list)

    cur = len(latents_list) - 1
    step_image = decode_latent(pipe, latents_list[cur])

    explanation = (
        "🧒 **Simple Explanation**\n"
        "The model starts with noise, slowly removes it, and reveals an image.\n"
        if simple else
        "🔬 **Technical Explanation**\n"
        "We collect latents at each DDIM step, decode them via VAE, and visualize their PCA path."
    )
    explanation += f"\n⏱ Runtime: {total:.2f}s"

    state = {
        "latents": latents_list,
        "pca": pca,
        "norms": norms
    }

    return (
        final,
        explanation,
        gr.update(maximum=len(latents_list)-1, value=cur),
        step_image,
        plot_pca(pca, cur),
        plot_norm(norms, cur),
        state
    )


# ------------------- PLOT FUNCTIONS -------------------

def plot_pca(points, idx):
    fig = go.Figure()
    fig.add_trace(go.Scatter(x=points[:,0], y=points[:,1], mode="lines+markers"))
    fig.add_trace(go.Scatter(
        x=[points[idx,0]], y=[points[idx,1]],
        mode="markers", marker=dict(size=12, color="red")
    ))
    fig.update_layout(height=350, title="PCA Trajectory")
    return fig

def plot_norm(norms, idx):
    fig = go.Figure()
    fig.add_trace(go.Scatter(y=norms, mode="lines+markers"))
    fig.add_trace(go.Scatter(
        x=[idx], y=[norms[idx]], mode="markers", marker=dict(size=12, color="red")
    ))
    fig.update_layout(height=350, title="Latent Norm Over Steps")
    return fig


# ------------------- SLIDER UPDATE -------------------

def update_step(state, idx):
    latents = state["latents"]
    pca = state["pca"]
    norms = state["norms"]
    pipe = get_pipe()

    img = decode_latent(pipe, latents[idx])
    return (
        img,
        plot_pca(pca, idx),
        plot_norm(norms, idx)
    )


# ------------------- UI -------------------

with gr.Blocks(title="SD v1-4 CPU Diffusion Visualizer") as demo:

    gr.Markdown("# 🧠 Stable Diffusion v1-4 — CPU Visualizer (256×256)")
    gr.Markdown("This version produces **real images**, optimized for free HF CPU.")

    with gr.Row():
        with gr.Column():
            prompt = gr.Textbox(label="Prompt", value="a cute cat in watercolor")
            steps = gr.Slider(10, 30, value=20, step=1, label="Steps")
            guidance = gr.Slider(3, 12, value=7.5, step=0.5, label="Guidance")
            seed = gr.Number(label="Seed (-1 for random)", value=-1)
            simple = gr.Checkbox(label="Simple Explanation", value=True)
            run = gr.Button("Run Diffusion", variant="primary")

        with gr.Column():
            final = gr.Image(label="Final Image")
            expl = gr.Markdown()

    step_slider = gr.Slider(0, 0, value=0, step=1, label="View Step")
    step_img = gr.Image(label="Latent Image at Step")
    pca_plot = gr.Plot(label="PCA")
    norm_plot = gr.Plot(label="Norm Plot")
    state = gr.State()

    run.click(
        run_diffusion,
        inputs=[prompt, steps, guidance, seed, simple],
        outputs=[final, expl, step_slider, step_img, pca_plot, norm_plot, state]
    )

    step_slider.change(update_step, [state, step_slider], [step_img, pca_plot, norm_plot])

demo.launch()