-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
932 lines (858 loc) · 30.8 KB
/
index.html
File metadata and controls
932 lines (858 loc) · 30.8 KB
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
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description" content="IP-to-Portrait: SDXL Inpainting + IP-Adapter FaceID Plus v2 기반 얼굴 합성 파이프라인">
<meta name="keywords" content="AI, Face Synthesis, SDXL, IP-Adapter, Inpainting, Diffusion Models">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>IP-to-Portrait: Face Synthesis Pipeline</title>
<!-- Google Fonts -->
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet">
<!-- Font Awesome -->
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css">
<!-- Bulma CSS -->
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bulma@0.9.4/css/bulma.min.css">
<style>
body {
font-family: 'Noto Sans', sans-serif;
}
.hero-body {
padding: 3rem 1.5rem;
}
.publication-title {
font-family: 'Google Sans', sans-serif;
font-size: 2.5rem;
font-weight: 700;
margin-bottom: 1rem;
}
.publication-authors {
font-size: 1.1rem;
margin-bottom: 0.5rem;
}
.author-block {
display: inline-block;
margin: 0 0.3rem;
}
.publication-links {
margin-top: 1.5rem;
}
.publication-links .button {
margin: 0.3rem;
}
.section {
padding: 3rem 1.5rem;
}
.container {
max-width: 1200px;
}
h2.title {
font-size: 2rem;
font-weight: 700;
margin-bottom: 1rem;
}
.content {
font-size: 1.1rem;
line-height: 1.8;
}
.image-container {
margin: 2rem 0;
text-align: center;
}
.image-container img {
max-width: 90%;
height: auto;
border-radius: 8px;
box-shadow: 0 4px 8px rgba(0,0,0,0.1);
}
.team-member {
text-align: center;
margin: 1rem;
}
.team-member img {
border-radius: 50%;
width: 120px;
height: 120px;
margin-bottom: 0.5rem;
}
.tech-stack .tag {
margin: 0.3rem;
font-size: 0.9rem;
}
footer {
padding: 2rem;
background-color: #f5f5f5;
text-align: center;
}
table {
margin: 1.5rem auto;
border-collapse: collapse;
background: white;
}
table th, table td {
padding: 0.75rem;
border: 1px solid #dbdbdb;
}
table th {
background-color: #f5f5f5;
font-weight: 700;
}
.pipeline-diagram {
background: #f9f9f9;
padding: 2rem;
border-radius: 8px;
margin: 2rem 0;
}
/* Modal 스타일 */
.modal {
display: none;
position: fixed;
z-index: 1000;
left: 0;
top: 0;
width: 100%;
height: 100%;
overflow: auto;
background-color: rgba(0,0,0,0.95);
}
.modal.is-active {
display: flex;
align-items: center;
justify-content: center;
}
.modal-content-custom {
position: relative;
width: 95vw;
height: 95vh;
display: flex;
align-items: center;
justify-content: center;
}
.modal-content-custom img {
max-width: 100%;
max-height: 100%;
object-fit: contain;
}
.close-modal {
position: absolute;
top: 20px;
right: 40px;
color: #fff;
font-size: 40px;
font-weight: bold;
cursor: pointer;
z-index: 1001;
text-shadow: 0 0 10px rgba(0,0,0,0.5);
}
.close-modal:hover,
.close-modal:focus {
color: #ccc;
}
</style>
</head>
<body>
<!-- Title Section -->
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="publication-title">IP-to-Portrait</h1>
<p style="font-size: 1.3rem; margin-bottom: 1.5rem;">
SDXL Inpainting + IP-Adapter FaceID Plus v2 기반 얼굴 합성 파이프라인
</p>
<div class="publication-authors">
<span class="author-block">
<a href="https://github.com/danlee-dev">이성민</a>,
</span>
<span class="author-block">
<a href="https://github.com/byungkun0823">임병건</a>,
</span>
<span class="author-block">
<a href="https://github.com/seoyeon-eo">최서연</a>,
</span>
<span class="author-block">
<a href="https://github.com/hongjiyeon56">홍지연</a>
</span>
</div>
<div class="publication-authors">
<span class="author-block">Project Prometheus 2025-2 Team_diffusers </span>
</div>
<!-- Links -->
<div class="publication-links">
<!-- Page 버튼 -->
<span class="link-block">
<a href="#" id="openPageBtn" class="button is-dark is-rounded">
<span class="icon">
<i class="fas fa-images"></i>
</span>
<span>Page</span>
</a>
</span>
<!-- Code 버튼 -->
<span class="link-block">
<a href="https://github.com/Diffusion-planet/ip-to-portrait" class="button is-dark is-rounded">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
<span class="link-block">
<a href="#demo" class="button is-dark is-rounded">
<span class="icon">
<i class="fas fa-play"></i>
</span>
<span>Demo</span>
</a>
</span>
</div>
<!-- Badges -->
<div style="margin-top: 1.5rem;">
<img src="https://img.shields.io/badge/SDXL-Inpainting-blue?style=for-the-badge" alt="SDXL">
<img src="https://img.shields.io/badge/IP--Adapter-FaceID%20Plus%20v2-green?style=for-the-badge" alt="IP-Adapter">
<img src="https://img.shields.io/badge/BiSeNet-Face%20Parsing-orange?style=for-the-badge" alt="BiSeNet">
<img src="https://img.shields.io/badge/InsightFace-Face%20Swap-red?style=for-the-badge" alt="InsightFace">
<img src="https://img.shields.io/badge/GFPGAN-Face%20Enhance-FF6B6B?style=for-the-badge" alt="GFPGAN">
<img src="https://img.shields.io/badge/Gemini-2.5%20Flash-8E75B2?style=for-the-badge&logo=googlegemini&logoColor=white" alt="Gemini">
</div>
</div>
</div>
</div>
</div>
</section>
<!-- Modal for Page Gallery -->
<div id="pageModal" class="modal">
<span class="close-modal" id="closeModalBtn">×</span>
<div class="modal-content-custom">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/face-inpainting-pipeline.png" alt="IP-to-Portrait Pipeline">
</div>
</div>
<!-- Abstract -->
<section class="section">
<div class="container is-max-desktop">
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title">Abstract</h2>
<div class="content has-text-justified">
<p>
IP-to-Portrait는 레퍼런스 이미지의 얼굴을 교체하면서 원본 배경, 포즈, 조명을 완벽하게 보존하는 AI 기반 얼굴 합성 파이프라인입니다.
최신 디퓨전 모델(RealVisXL V4.0 Inpainting)과 아이덴티티 보존 어댑터(IP-Adapter FaceID Plus v2)를 결합하여
사실적인 얼굴 합성을 수행합니다.
</p>
<p>
본 시스템은 정밀한 얼굴 파싱(BiSeNet), 얼굴 임베딩 추출(InsightFace), Face Swap(InsightFace inswapper_128),
Face Enhance(GFPGAN v1.4), 자동 프롬프트 생성(Gemini 2.5 Flash)을 통합하여
사용자가 쉽게 고품질의 얼굴 합성 결과를 얻을 수 있도록 지원합니다. 또한 병렬 처리와 실시간 미리보기 기능을 통해
빠르고 직관적인 사용자 경험을 제공합니다.
</p>
</div>
</div>
</div>
</div>
</section>
<!-- Demo Images -->
<section class="section" id="demo">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">주요 기능</h2>
<div class="content">
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/generation-screen.png" alt="Generation Screen">
<p style="margin-top: 1rem;">노드 기반 워크플로우로 직관적인 얼굴 합성</p>
</div>
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/face-inpainting-pipeline.png" alt="AI Pipeline">
<p style="margin-top: 1rem;">AI 파이프라인 구조 (5단계: INPUT → PREPROCESSING → GENERATION → POST-PROCESSING → OUTPUT)</p>
</div>
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/inpainting-parameters-setting-node.png" alt="Parameters Node">
<p style="margin-top: 1rem;">인페인팅 파라미터 설정 노드 (IP-Adapter 모드, Diffusion, Mask, Pre-paste, POST-PROCESSING)</p>
</div>
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/auto-prompt-generation-node.png" alt="Auto Prompt">
<p style="margin-top: 1rem;">Gemini 2.5 Flash를 활용한 자동 프롬프트 생성</p>
</div>
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/generation-control-and-step-result-latent-image-return-node.png" alt="Generation Control">
<p style="margin-top: 1rem;">생성 제어 및 실시간 Latent 이미지 미리보기</p>
</div>
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/generation-history-modal.png" alt="History">
<p style="margin-top: 1rem;">생성 히스토리 관리 (파라미터 리콜 기능)</p>
</div>
<div class="image-container">
<img src="https://raw.githubusercontent.com/Diffusion-planet/ip-to-portrait/feature/pre-paste-faceswap/images/generation-report-modal.png" alt="Report">
<p style="margin-top: 1rem;">상세 생성 리포트 (모든 파라미터 및 다운로드 가능한 결과)</p>
</div>
</div>
</div>
</section>
<!-- AI Pipeline -->
<section class="section">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">AI 파이프라인 구조</h2>
<div class="content pipeline-diagram">
<h3>파이프라인 흐름 (5단계)</h3>
<ol>
<li><strong>INPUT (입력):</strong> 레퍼런스 이미지(배경/포즈), 얼굴 이미지(정체성 소스), 프롬프트(텍스트 가이드)</li>
<li><strong>PREPROCESSING (전처리):</strong>
<ul>
<li>Face Detection: InsightFace (buffalo_l)로 얼굴 위치 및 랜드마크 감지</li>
<li>Face Parsing: BiSeNet으로 얼굴 영역 세그멘테이션 (피부, 눈, 코, 입, 머리카락, 목)</li>
<li>Mask Generation: 인페인팅 영역 정의 (확장, 블러, 패딩 적용)</li>
<li>Embedding Extraction: InsightFace (512D 얼굴 임베딩) + CLIP ViT-H/14 (257x1280 이미지 임베딩)</li>
</ul>
</li>
<li><strong>GENERATION (생성):</strong>
<ul>
<li>IP-Adapter Injection: FaceID Plus v2로 얼굴 임베딩을 SDXL 어텐션 레이어에 주입</li>
<li>Inpainting: RealVisXL V4.0로 마스크 영역 재생성 (배경 보존, Stop-At 파라미터 적용)</li>
</ul>
</li>
<li><strong>POST-PROCESSING (후처리):</strong>
<ul>
<li>Face Swap: InsightFace (inswapper_128) 또는 Ghost로 생성된 얼굴을 소스 얼굴로 교체</li>
<li>Refinement (선택): Face Swap 경계 블렌딩 인페인팅</li>
<li>Face Enhance: GFPGAN v1.4로 얼굴 화질 향상/복원</li>
</ul>
</li>
<li><strong>OUTPUT (출력):</strong> 최종 합성 결과 생성 및 WebSocket 실시간 업데이트</li>
</ol>
<h3>모델 구성</h3>
<table>
<thead>
<tr>
<th>모델</th>
<th>용도</th>
<th>크기</th>
</tr>
</thead>
<tbody>
<tr>
<td>RealVisXL V4.0 Inpainting</td>
<td>기본 디퓨전 모델 (SDXL 기반, 포토리얼리스틱)</td>
<td>~6GB</td>
</tr>
<tr>
<td>IP-Adapter FaceID Plus v2</td>
<td>얼굴 정체성 주입</td>
<td>~100MB</td>
</tr>
<tr>
<td>CLIP ViT-H/14</td>
<td>이미지 임베딩 추출</td>
<td>~2GB</td>
</tr>
<tr>
<td>InsightFace (buffalo_l)</td>
<td>얼굴 임베딩 + Face Swap</td>
<td>~300MB</td>
</tr>
<tr>
<td>BiSeNet</td>
<td>얼굴 파싱/세그멘테이션</td>
<td>~50MB</td>
</tr>
<tr>
<td>GFPGAN v1.4</td>
<td>얼굴 화질 향상</td>
<td>~350MB</td>
</tr>
<tr>
<td>Gemini 2.5 Flash</td>
<td>자동 프롬프트 생성 (API)</td>
<td>Cloud</td>
</tr>
</tbody>
</table>
<h3>Stop-At 기능</h3>
<p>
Stop-At 파라미터는 IP-Adapter FaceID가 생성에 영향을 미치는 것을 언제 중단할지 제어합니다.
예를 들어 stop_at=0.7로 설정하면 생성 과정의 70% 지점까지는 FaceID가 강하게 정체성과 구조를 가이드하고,
이후 30%는 FaceID 영향 없이 프롬프트만으로 자연스러운 디테일을 완성합니다.
</p>
<table>
<thead>
<tr>
<th>Stop-At 값</th>
<th>효과</th>
<th>사용 시점</th>
</tr>
</thead>
<tbody>
<tr>
<td>1.0</td>
<td>전체 과정에서 FaceID 적용</td>
<td>최대 정체성 보존</td>
</tr>
<tr>
<td>0.85-0.95</td>
<td>85-95%에서 FaceID 중단</td>
<td>권장 균형점</td>
</tr>
<tr>
<td>0.7-0.8</td>
<td>70-80%에서 FaceID 중단</td>
<td>프롬프트 영향력 증가</td>
</tr>
</tbody>
</table>
</div>
</div>
</section>
<!-- Hyperparameters -->
<section class="section">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">하이퍼파라미터</h2>
<div class="content">
<h3>IP-Adapter 모드</h3>
<table>
<thead>
<tr>
<th>모드</th>
<th>정체성</th>
<th>스타일</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>Simple Inpainting</td>
<td>없음</td>
<td>없음</td>
<td>IP-Adapter 없이 순수 인페인팅 (Pre-paste와 함께 사용)</td>
</tr>
<tr>
<td>Standard (CLIP)</td>
<td>낮음</td>
<td>중간</td>
<td>CLIP 기반 이미지 임베딩만 사용</td>
</tr>
<tr>
<td>FaceID</td>
<td>높음</td>
<td>낮음</td>
<td>InsightFace 얼굴 임베딩만 사용</td>
</tr>
<tr style="background-color: #fffacd;">
<td><strong>FaceID Plus v2 ⭐</strong></td>
<td><strong>높음</strong></td>
<td><strong>높음</strong></td>
<td><strong>얼굴 임베딩 + CLIP 이미지 결합 (권장)</strong></td>
</tr>
<tr>
<td>CLIP Blend</td>
<td>중간</td>
<td>높음</td>
<td>CLIP 임베딩에 얼굴 특징 블렌딩</td>
</tr>
</tbody>
</table>
<h3>Diffusion 파라미터</h3>
<table>
<thead>
<tr>
<th>파라미터</th>
<th>범위</th>
<th>기본값</th>
<th>권장값</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>steps</td>
<td>1-100</td>
<td>50</td>
<td>30-50</td>
<td>디퓨전 스텝 수 (높을수록 품질↑, 속도↓)</td>
</tr>
<tr>
<td>guidance_scale</td>
<td>1-20</td>
<td>3.0</td>
<td>2.5-5.0</td>
<td>CFG 스케일 - 낮을수록 자연스러움</td>
</tr>
<tr>
<td>denoise_strength</td>
<td>0.0-1.0</td>
<td>0.92</td>
<td>0.85-0.95</td>
<td>생성 강도 - 높을수록 더 많이 재생성</td>
</tr>
<tr>
<td>face_strength</td>
<td>0.0-1.5</td>
<td>0.85</td>
<td>0.7-0.9</td>
<td>IP-Adapter 얼굴 가이드 강도</td>
</tr>
<tr>
<td>stop_at</td>
<td>0.0-1.0</td>
<td>1.0</td>
<td>0.85-0.95</td>
<td>FaceID 영향 중단 시점 (조기 종료)</td>
</tr>
</tbody>
</table>
<h3>Mask 파라미터</h3>
<table>
<thead>
<tr>
<th>파라미터</th>
<th>범위</th>
<th>기본값</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>include_hair</td>
<td>bool</td>
<td>true</td>
<td>마스크에 머리카락 영역 포함</td>
</tr>
<tr>
<td>include_neck</td>
<td>bool</td>
<td>true</td>
<td>마스크에 목 영역 포함</td>
</tr>
<tr>
<td>mask_expand</td>
<td>0.0-1.0</td>
<td>0.3</td>
<td>마스크 확장 비율 (1.0 = 30% 확장)</td>
</tr>
<tr>
<td>mask_blur</td>
<td>0-100</td>
<td>15</td>
<td>마스크 경계 블러 (픽셀)</td>
</tr>
<tr>
<td>mask_padding</td>
<td>-100-100</td>
<td>10</td>
<td>추가 마스크 확장/축소 (픽셀)</td>
</tr>
</tbody>
</table>
<h3>Pre-paste 파라미터</h3>
<table>
<thead>
<tr>
<th>파라미터</th>
<th>범위</th>
<th>기본값</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>use_pre_paste</td>
<td>bool</td>
<td>false</td>
<td>Pre-paste 모드 활성화</td>
</tr>
<tr>
<td>pre_paste_denoising</td>
<td>0.0-1.0</td>
<td>0.65</td>
<td>붙여넣기 후 디노이징 강도 (낮을수록 원본 유지)</td>
</tr>
</tbody>
</table>
<h3>POST-PROCESSING 파라미터</h3>
<h4>Face Swap</h4>
<table>
<thead>
<tr>
<th>파라미터</th>
<th>범위</th>
<th>기본값</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>use_face_swap</td>
<td>bool</td>
<td>false</td>
<td>Face Swap 활성화</td>
</tr>
<tr>
<td>face_swap_model</td>
<td>string</td>
<td>"insightface"</td>
<td>모델 선택 ("insightface", "ghost")</td>
</tr>
<tr>
<td>use_swap_refinement</td>
<td>bool</td>
<td>false</td>
<td>Face Swap 후 블렌딩 인페인팅</td>
</tr>
<tr>
<td>swap_refinement_strength</td>
<td>0.0-1.0</td>
<td>0.3</td>
<td>블렌딩 인페인팅 강도</td>
</tr>
</tbody>
</table>
<h4>Face Enhance</h4>
<table>
<thead>
<tr>
<th>파라미터</th>
<th>범위</th>
<th>기본값</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>use_face_enhance</td>
<td>bool</td>
<td>false</td>
<td>GFPGAN 얼굴 향상 활성화</td>
</tr>
<tr>
<td>face_enhance_strength</td>
<td>0.0-1.0</td>
<td>0.8</td>
<td>향상 강도 (1.0 = 최대 복원)</td>
</tr>
</tbody>
</table>
<h3>권장 세팅 가이드</h3>
<table>
<thead>
<tr>
<th>섹션</th>
<th>파라미터</th>
<th>권장값</th>
<th>설명</th>
</tr>
</thead>
<tbody>
<tr>
<td>IP-Adapter</td>
<td>Mode</td>
<td>FaceID Plus v2</td>
<td>얼굴 유사도 + 자연스러움 최적 균형</td>
</tr>
<tr>
<td>Diffusion</td>
<td>Stop At</td>
<td>85-95%</td>
<td>낮출수록 프롬프트 영향↑ (얼굴 윤곽 조정 가능)</td>
</tr>
<tr>
<td>Mask</td>
<td>Include Hair</td>
<td>✅ On</td>
<td>헤어스타일 변경 허용</td>
</tr>
<tr>
<td>Mask</td>
<td>Include Neck</td>
<td>✅ On</td>
<td>목까지 자연스럽게 블렌딩</td>
</tr>
<tr>
<td>Pre-paste</td>
<td>활성화</td>
<td>❌ Off</td>
<td>비정상적 얼굴 생성 가능성 있음</td>
</tr>
<tr>
<td>Face Swap</td>
<td>활성화</td>
<td>✅ On</td>
<td>얼굴 유사도 대폭 향상</td>
</tr>
<tr>
<td>Face Swap</td>
<td>Model</td>
<td>InsightFace 128</td>
<td>안정적이고 빠름</td>
</tr>
<tr>
<td>Face Swap</td>
<td>Refinement</td>
<td>❌ Off (기본)</td>
<td>얼굴이 비정상일 때만 활성화</td>
</tr>
<tr>
<td>Face Enhance</td>
<td>활성화</td>
<td>✅ On</td>
<td>화질 개선 필수</td>
</tr>
<tr>
<td>Face Enhance</td>
<td>Strength</td>
<td>80-95%</td>
<td>높여도 무방, 얼굴 변화 시 낮춤</td>
</tr>
</tbody>
</table>
</div>
</div>
</section>
<!-- Team -->
<section class="section">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">팀원</h2>
<div class="columns is-multiline is-centered">
<div class="column is-3">
<div class="team-member">
<img src="https://github.com/hongjiyeon56.png" alt="홍지연">
<h4 class="title is-5">홍지연 (팀장)</h4>
<p><a href="https://github.com/hongjiyeon56">@hongjiyeon56</a></p>
</div>
</div>
<div class="column is-3">
<div class="team-member">
<img src="https://github.com/byungkun0823.png" alt="임병건">
<h4 class="title is-5">임병건</h4>
<p><a href="https://github.com/byungkun0823">@byungkun0823</a></p>
</div>
</div>
<div class="column is-3">
<div class="team-member">
<img src="https://github.com/danlee-dev.png" alt="이성민">
<h4 class="title is-5">이성민</h4>
<p><a href="https://github.com/danlee-dev">@danlee-dev</a></p>
</div>
</div>
<div class="column is-3">
<div class="team-member">
<img src="https://github.com/seoyeon-eo.png" alt="최서연">
<h4 class="title is-5">최서연</h4>
<p><a href="https://github.com/seoyeon-eo">@seoyeon-eo</a></p>
</div>
</div>
</div>
</div>
</section>
<!-- Tech Stack -->
<section class="section">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">기술 스택</h2>
<div class="content tech-stack">
<h3>AI / ML</h3>
<div class="tags">
<span class="tag is-info is-medium">Python</span>
<span class="tag is-info is-medium">PyTorch</span>
<span class="tag is-info is-medium">Hugging Face Diffusers</span>
<span class="tag is-info is-medium">OpenCV</span>
<span class="tag is-info is-medium">RealVisXL V4.0</span>
<span class="tag is-info is-medium">IP-Adapter FaceID Plus v2</span>
<span class="tag is-info is-medium">BiSeNet</span>
<span class="tag is-info is-medium">InsightFace</span>
<span class="tag is-info is-medium">GFPGAN v1.4</span>
<span class="tag is-info is-medium">CLIP ViT-H/14</span>
<span class="tag is-info is-medium">Gemini 2.5 Flash</span>
</div>
<h3>Backend</h3>
<div class="tags">
<span class="tag is-success is-medium">FastAPI</span>
<span class="tag is-success is-medium">Uvicorn</span>
<span class="tag is-success is-medium">Celery</span>
<span class="tag is-success is-medium">SQLAlchemy</span>
<span class="tag is-success is-medium">Alembic</span>
<span class="tag is-success is-medium">JWT</span>
<span class="tag is-success is-medium">WebSocket</span>
</div>
<h3>Frontend</h3>
<div class="tags">
<span class="tag is-warning is-medium">Next.js</span>
<span class="tag is-warning is-medium">React</span>
<span class="tag is-warning is-medium">TypeScript</span>
<span class="tag is-warning is-medium">Tailwind CSS</span>
</div>
<h3>Database & Infrastructure</h3>
<div class="tags">
<span class="tag is-link is-medium">PostgreSQL</span>
<span class="tag is-link is-medium">Redis</span>
<span class="tag is-link is-medium">Docker</span>
<span class="tag is-link is-medium">Docker Compose</span>
</div>
</div>
</div>
</section>
<!-- References -->
<section class="section">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">참고 문헌</h2>
<div class="content">
<ul>
<li><a href="https://github.com/tencent-ailab/IP-Adapter">IP-Adapter</a> - Tencent AI Lab</li>
<li><a href="https://huggingface.co/OzzyGT/RealVisXL_V4.0_inpainting">RealVisXL V4.0</a> - Photorealistic SDXL Inpainting</li>
<li><a href="https://huggingface.co/diffusers/stable-diffusion-xl-1.0-inpainting-0.1">SDXL Inpainting</a> - Stability AI</li>
<li><a href="https://github.com/deepinsight/insightface">InsightFace</a> - Face Analysis & Swap</li>
<li><a href="https://github.com/TencentARC/GFPGAN">GFPGAN</a> - Face Restoration</li>
<li><a href="https://github.com/zllrunning/face-parsing.PyTorch">BiSeNet</a> - Face Parsing</li>
<li><a href="https://github.com/huggingface/diffusers">Diffusers</a> - Hugging Face</li>
<li><a href="https://ai.google.dev/">Gemini API</a> - Google AI</li>
</ul>
</div>
</div>
</section>
<!-- BibTeX -->
<section class="section">
<div class="container is-max-desktop">
<h2 class="title has-text-centered">BibTeX</h2>
<pre><code>@misc{diffusionplanet2025ipportrait,
title={IP-to-Portrait: SDXL Inpainting + IP-Adapter FaceID Plus v2 Face Synthesis Pipeline},
author={Hong, Jiyeon and Lim, Byungkun and Lee, Seongmin and Choi, Seoyeon},
year={2025},
howpublished={\url{https://github.com/Diffusion-planet/ip-to-portrait}},
note={Project Prometheus 2025-2 Demo Day - Team 4}
}</code></pre>
</div>
</section>
<!-- Footer -->
<footer>
<div class="container">
<div class="content has-text-centered">
<p>
Website template based on <a href="https://nerfies.github.io/">Nerfies</a>.
</p>
<p>
<a href="https://github.com/Diffusion-planet/ip-to-portrait">View on GitHub</a>
</p>
</div>
</div>
</footer>
<!-- JavaScript for Modal -->
<script>
// Modal 열기
document.getElementById('openPageBtn').addEventListener('click', function(e) {
e.preventDefault();
document.getElementById('pageModal').classList.add('is-active');
});
// Modal 닫기
document.getElementById('closeModalBtn').addEventListener('click', function() {
document.getElementById('pageModal').classList.remove('is-active');
});
// 배경 클릭시 Modal 닫기
window.addEventListener('click', function(event) {
const modal = document.getElementById('pageModal');
if (event.target === modal) {
modal.classList.remove('is-active');
}
});
// ESC 키로 Modal 닫기
document.addEventListener('keydown', function(event) {
if (event.key === 'Escape') {
document.getElementById('pageModal').classList.remove('is-active');
}
});
</script>
</body>
</html>