工作流 /
视频 /
Pika H3 Flex：参考生视频
Pika H3 Flex：参考生视频
Pika H3 Flex: Reference to Video
使用 Pika H3 Flex 从参考图像生成带同步音频的视频，在提示词承载故事的同时保持参考图像的身份、服装和外观。两个图像输入分别接入为 Picture 1 和 Picture 2，Pika 适配器以强度 1.0 运行，配合 lcm 采样器和六步 sigma 列表，基于 MiniMax H3 Ref2VA 基础模型。非常适合制作角色一致且带原生声音的短片。
视频
22 个节点 2 个输入 3,324 次使用 ComfyUI ≥ 0.44.0 ComfyUI 39 KB
需要的权重
这个工作流引用了 5 个权重文件，其中 5 个能在本站模型库里找到。
minimax_h3_audio_vae_fp32.safetensors
在模型库查看
minimax_h3_ref2va_pruned_int8_convrot.safetensors
在模型库查看
minimax_h3_video_vae_int8_convrot.safetensors
在模型库查看
pika_h3_flex_ref2va_v1_bf16.safetensors
在模型库查看
qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors
在模型库查看
节点构成
VAELoader
× 2
LoadImage
× 2
SaveVideo
× 1
ResolutionSelector
× 1
MarkdownNote
× 1
VAEDecodeAudio
× 1
VAEDecode
× 1
KSamplerSelect
× 1
SamplerCustomAdvanced
× 1
BasicGuider
× 1
UNETLoader
× 1
CLIPLoader
× 1
RandomNoise
× 1
ComfyMathExpression
× 1
PrimitiveFloat
× 1
MiniMaxH3ReferenceToVideo
× 1
官方标注的模型
MiniMax H3
工作流 JSON
和上面「下载工作流 JSON」是同一份文件。
1{
2 "id": "e3f2b845-8f2c-4b5a-9caf-eac1029d3e7e",
3 "revision": 0,
4 "last_node_id": 152,
5 "last_link_id": 301,
6 "nodes": [
7 {
8 "id": 92,
9 "type": "SaveVideo",
10 "pos": [
11 750,
12 4870
13 ],
14 "size": [
15 1380,
16 970
17 ],
18 "flags": [],
19 "order": 21,
20 "mode": 0,
21 "inputs": [
22 {
23 "name": "video",
24 "type": "VIDEO",
25 "link": 299
26 }
27 ],
28 "outputs": [
29 {
30 "name": "video",
31 "type": "VIDEO",
32 "links": null
33 }
34 ],
35 "properties": {
36 "cnr_id": "comfy-core",
37 "ver": "0.33.0"
38 },
39 "widgets_values": [
40 "video/Pika_H3_Flex_REF2V",
41 "auto",
42 "auto",
43 "auto"
44 ],
45 "widgets_values_named": {
46 "filename_prefix": "video/Pika_H3_Flex_REF2V",
47 "format": "auto",
48 "format.codec": "auto",
49 "codec": "auto"
50 }
51 },
52 {
53 "id": 115,
54 "type": "ResolutionSelector",
55 "pos": [
56 -1490,
57 5980
58 ],
59 "size": [
60 270,
61 170
62 ],
63 "flags": [],
64 "order": 0,
65 "mode": 0,
66 "showAdvanced": true,
67 "inputs": [],
68 "outputs": [
69 {
70 "name": "width",
71 "type": "INT",
72 "links": [
73 276
74 ]
75 },
76 {
77 "name": "height",
78 "type": "INT",
79 "links": [
80 277
81 ]
82 }
83 ],
84 "title": "Resolution Selector (Size)",
85 "properties": {
86 "Node name for S&R": "ResolutionSelector",
87 "cnr_id": "comfy-core",
88 "ver": "0.33.0"
89 },
90 "widgets_values": [
91 "16:9 (Widescreen)",
92 0.98,
93 32
94 ],
95 "widgets_values_named": {
96 "aspect_ratio": "16:9 (Widescreen)",
97 "megapixels": 0.98,
98 "multiple": 32
99 },
100 "color": "#322",
101 "bgcolor": "#533"
102 },
103 {
104 "id": 117,
105 "type": "MarkdownNote",
106 "pos": [
107 -2020,
108 4890
109 ],
110 "size": [
111 440,
112 740
113 ],
114 "flags": [],
115 "order": 1,
116 "mode": 0,
117 "inputs": [],
118 "outputs": [],
119 "title": "Note: Model Links",
120 "properties": [],
121 "widgets_values": [
122 "## Input Assets\n\n- [ref_pilot_profile.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/ref_pilot_profile.png)\n- [ref_subject_station.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/ref_subject_station.png)\n\n## Model Links\n\n- [Hugging Face:Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3)\n\n**ModelScope mirrors** (alternative repo location)\n\n- [ModelScope:Comfy-Org/MiniMax-H3](https://modelscope.cn/models/Comfy-Org/MiniMax-H3)\n\n**vae**\n\n- [minimax_h3_video_vae_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_int8_convrot.safetensors) (2.62 GB)\n- [minimax_h3_audio_vae_fp32.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors) (577.2 MB)\n\n**diffusion_models**\n\n- [minimax_h3_ref2va_pruned_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors) (19.53 GB)\n\n**text_encoders**\n\n- [qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors) (14.61 GB)\n\n**loras**\n\n- [pika_h3_flex_ref2va_v1_bf16.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/loras/pika_h3_flex_ref2va_v1_bf16.safetensors)\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 vae/\n│ │ ├── minimax_h3_video_vae_int8_convrot.safetensors\n│ │ └── minimax_h3_audio_vae_fp32.safetensors\n│ ├── 📂 diffusion_models/\n│ │ └── minimax_h3_ref2va_pruned_int8_convrot.safetensors\n│ ├── 📂 text_encoders/\n│ │ └── qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors\n│ └── 📂 loras/\n│ └── pika_h3_flex_ref2va_v1_bf16.safetensors\n```\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
123 ],
124 "widgets_values_named": {
125 "text": "## Input Assets\n\n- [ref_pilot_profile.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/ref_pilot_profile.png)\n- [ref_subject_station.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/ref_subject_station.png)\n\n## Model Links\n\n- [Hugging Face:Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3)\n\n**ModelScope mirrors** (alternative repo location)\n\n- [ModelScope:Comfy-Org/MiniMax-H3](https://modelscope.cn/models/Comfy-Org/MiniMax-H3)\n\n**vae**\n\n- [minimax_h3_video_vae_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_int8_convrot.safetensors) (2.62 GB)\n- [minimax_h3_audio_vae_fp32.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors) (577.2 MB)\n\n**diffusion_models**\n\n- [minimax_h3_ref2va_pruned_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors) (19.53 GB)\n\n**text_encoders**\n\n- [qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors) (14.61 GB)\n\n**loras**\n\n- [pika_h3_flex_ref2va_v1_bf16.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/loras/pika_h3_flex_ref2va_v1_bf16.safetensors)\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 vae/\n│ │ ├── minimax_h3_video_vae_int8_convrot.safetensors\n│ │ └── minimax_h3_audio_vae_fp32.safetensors\n│ ├── 📂 diffusion_models/\n│ │ └── minimax_h3_ref2va_pruned_int8_convrot.safetensors\n│ ├── 📂 text_encoders/\n│ │ └── qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors\n│ └── 📂 loras/\n│ └── pika_h3_flex_ref2va_v1_bf16.safetensors\n```\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
126 },
127 "color": "#222",
128 "bgcolor": "#000"
129 },
130 {
131 "id": 119,
132 "type": "VAELoader",
133 "pos": [
134 -1490,
135 5490
136 ],
137 "size": [
138 640,
139 70
140 ],
141 "flags": [],
142 "order": 2,
143 "mode": 0,
144 "inputs": [],
145 "outputs": [
146 {
147 "name": "VAE",
148 "type": "VAE",
149 "links": [
150 251,
151 273
152 ]
153 }
154 ],
155 "properties": {
156 "Node name for S&R": "VAELoader",
157 "cnr_id": "comfy-core",
158 "ver": "0.33.0",
159 "models": [
160 {
161 "name": "minimax_h3_video_vae_int8_convrot.safetensors",
162 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_int8_convrot.safetensors",
163 "directory": "vae"
164 }
165 ]
166 },
167 "widgets_values": [
168 "minimax_h3_video_vae_int8_convrot.safetensors"
169 ],
170 "widgets_values_named": {
171 "vae_name": "minimax_h3_video_vae_int8_convrot.safetensors"
172 }
173 },
174 {
175 "id": 120,
176 "type": "VAELoader",
177 "pos": [
178 -1490,
179 5630
180 ],
181 "size": [
182 650,
183 70
184 ],
185 "flags": [],
186 "order": 3,
187 "mode": 0,
188 "inputs": [],
189 "outputs": [
190 {
191 "name": "VAE",
192 "type": "VAE",
193 "links": [
194 250,
195 274
196 ]
197 }
198 ],
199 "properties": {
200 "Node name for S&R": "VAELoader",
201 "cnr_id": "comfy-core",
202 "ver": "0.33.0",
203 "models": [
204 {
205 "name": "minimax_h3_audio_vae_fp32.safetensors",
206 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors",
207 "directory": "vae"
208 }
209 ]
210 },
211 "widgets_values": [
212 "minimax_h3_audio_vae_fp32.safetensors"
213 ],
214 "widgets_values_named": {
215 "vae_name": "minimax_h3_audio_vae_fp32.safetensors"
216 }
217 },
218 {
219 "id": 121,
220 "type": "VAEDecodeAudio",
221 "pos": [
222 20,
223 5730
224 ],
225 "size": [
226 230,
227 60
228 ],
229 "flags": {
230 "collapsed": false
231 },
232 "order": 19,
233 "mode": 0,
234 "inputs": [
235 {
236 "name": "samples",
237 "type": "LATENT",
238 "link": 281
239 },
240 {
241 "name": "vae",
242 "type": "VAE",
243 "link": 250
244 }
245 ],
246 "outputs": [
247 {
248 "name": "AUDIO",
249 "type": "AUDIO",
250 "links": [
251 301
252 ]
253 }
254 ],
255 "properties": {
256 "Node name for S&R": "VAEDecodeAudio",
257 "cnr_id": "comfy-core",
258 "ver": "0.33.0"
259 }
260 },
261 {
262 "id": 122,
263 "type": "VAEDecode",
264 "pos": [
265 20,
266 5610
267 ],
268 "size": [
269 230,
270 60
271 ],
272 "flags": {
273 "collapsed": false
274 },
275 "order": 18,
276 "mode": 0,
277 "inputs": [
278 {
279 "name": "samples",
280 "type": "LATENT",
281 "link": 280
282 },
283 {
284 "name": "vae",
285 "type": "VAE",
286 "link": 251
287 }
288 ],
289 "outputs": [
290 {
291 "name": "IMAGE",
292 "type": "IMAGE",
293 "links": [
294 300
295 ]
296 }
297 ],
298 "properties": {
299 "Node name for S&R": "VAEDecode",
300 "cnr_id": "comfy-core",
301 "ver": "0.33.0"
302 }
303 },
304 {
305 "id": 123,
306 "type": "KSamplerSelect",
307 "pos": [
308 0,
309 5200
310 ],
311 "size": [
312 370,
313 70
314 ],
315 "flags": [],
316 "order": 4,
317 "mode": 0,
318 "inputs": [],
319 "outputs": [
320 {
321 "name": "SAMPLER",
322 "type": "SAMPLER",
323 "links": [
324 255
325 ]
326 }
327 ],
328 "properties": {
329 "Node name for S&R": "KSamplerSelect",
330 "cnr_id": "comfy-core",
331 "ver": "0.33.0"
332 },
333 "widgets_values": [
334 "lcm"
335 ],
336 "widgets_values_named": {
337 "sampler_name": "lcm"
338 }
339 },
340 {
341 "id": 125,
342 "type": "SamplerCustomAdvanced",
343 "pos": [
344 410,
345 5120
346 ],
347 "size": [
348 230,
349 140
350 ],
351 "flags": [],
352 "order": 17,
353 "mode": 0,
354 "inputs": [
355 {
356 "name": "noise",
357 "type": "NOISE",
358 "link": 253
359 },
360 {
361 "name": "guider",
362 "type": "GUIDER",
363 "link": 254
364 },
365 {
366 "name": "sampler",
367 "type": "SAMPLER",
368 "link": 255
369 },
370 {
371 "name": "sigmas",
372 "type": "SIGMAS",
373 "link": 296
374 },
375 {
376 "name": "latent_image",
377 "type": "LATENT",
378 "link": 271
379 }
380 ],
381 "outputs": [
382 {
383 "name": "output",
384 "type": "LATENT",
385 "links": [
386 280,
387 281
388 ]
389 },
390 {
391 "name": "denoised_output",
392 "type": "LATENT",
393 "links": []
394 }
395 ],
396 "properties": {
397 "Node name for S&R": "SamplerCustomAdvanced",
398 "cnr_id": "comfy-core",
399 "ver": "0.33.0"
400 }
401 },
402 {
403 "id": 126,
404 "type": "BasicGuider",
405 "pos": [
406 0,
407 5090
408 ],
409 "size": [
410 360,
411 60
412 ],
413 "flags": [],
414 "order": 16,
415 "mode": 0,
416 "inputs": [
417 {
418 "name": "model",
419 "type": "MODEL",
420 "link": 294
421 },
422 {
423 "name": "conditioning",
424 "type": "CONDITIONING",
425 "link": 270
426 }
427 ],
428 "outputs": [
429 {
430 "name": "GUIDER",
431 "type": "GUIDER",
432 "links": [
433 254
434 ]
435 }
436 ],
437 "properties": {
438 "Node name for S&R": "BasicGuider",
439 "cnr_id": "comfy-core",
440 "ver": "0.33.0"
441 }
442 },
443 {
444 "id": 127,
445 "type": "UNETLoader",
446 "pos": [
447 -1490,
448 4980
449 ],
450 "size": [
451 640,
452 90
453 ],
454 "flags": [],
455 "order": 5,
456 "mode": 0,
457 "inputs": [],
458 "outputs": [
459 {
460 "name": "MODEL",
461 "type": "MODEL",
462 "links": [
463 285
464 ]
465 }
466 ],
467 "properties": {
468 "Node name for S&R": "UNETLoader",
469 "cnr_id": "comfy-core",
470 "ver": "0.33.0",
471 "models": [
472 {
473 "name": "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
474 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors",
475 "directory": "diffusion_models"
476 }
477 ]
478 },
479 "widgets_values": [
480 "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
481 "default"
482 ],
483 "widgets_values_named": {
484 "unet_name": "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
485 "weight_dtype": "default"
486 }
487 },
488 {
489 "id": 128,
490 "type": "CLIPLoader",
491 "pos": [
492 -1490,
493 5300
494 ],
495 "size": [
496 640,
497 120
498 ],
499 "flags": [],
500 "order": 6,
501 "mode": 0,
502 "inputs": [],
503 "outputs": [
504 {
505 "name": "CLIP",
506 "type": "CLIP",
507 "links": [
508 272
509 ]
510 }
511 ],
512 "properties": {
513 "Node name for S&R": "CLIPLoader",
514 "cnr_id": "comfy-core",
515 "ver": "0.33.0",
516 "models": [
517 {
518 "name": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
519 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
520 "directory": "text_encoders"
521 }
522 ]
523 },
524 "widgets_values": [
525 "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
526 "minimax",
527 "default"
528 ],
529 "widgets_values_named": {
530 "clip_name": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
531 "type": "minimax",
532 "device": "default"
533 }
534 },
535 {
536 "id": 129,
537 "type": "RandomNoise",
538 "pos": [
539 0,
540 4950
541 ],
542 "size": [
543 360,
544 90
545 ],
546 "flags": [],
547 "order": 7,
548 "mode": 0,
549 "inputs": [],
550 "outputs": [
551 {
552 "name": "NOISE",
553 "type": "NOISE",
554 "links": [
555 253
556 ]
557 }
558 ],
559 "properties": {
560 "Node name for S&R": "RandomNoise",
561 "cnr_id": "comfy-core",
562 "ver": "0.33.0"
563 },
564 "widgets_values": [
565 261662374822964,
566 "randomize"
567 ],
568 "widgets_values_named": {
569 "noise_seed": 261662374822964,
570 "control_after_generate": "randomize"
571 }
572 },
573 {
574 "id": 131,
575 "type": "ComfyMathExpression",
576 "pos": [
577 -1470,
578 6370
579 ],
580 "size": [
581 230,
582 80
583 ],
584 "flags": {
585 "collapsed": true
586 },
587 "order": 14,
588 "mode": 0,
589 "inputs": [
590 {
591 "label": "a",
592 "name": "values.a",
593 "type": "FLOAT,INT,BOOLEAN",
594 "link": 261
595 },
596 {
597 "label": "b",
598 "name": "values.b",
599 "shape": 7,
600 "type": "FLOAT,INT,BOOLEAN",
601 "link": null
602 }
603 ],
604 "outputs": [
605 {
606 "name": "FLOAT",
607 "type": "FLOAT",
608 "links": []
609 },
610 {
611 "name": "INT",
612 "type": "INT",
613 "links": [
614 275
615 ]
616 },
617 {
618 "name": "BOOL",
619 "type": "BOOLEAN",
620 "links": []
621 }
622 ],
623 "properties": {
624 "Node name for S&R": "ComfyMathExpression",
625 "cnr_id": "comfy-core",
626 "ver": "0.33.0"
627 },
628 "widgets_values": [
629 "max(5, round(a * 24)) + (5 - (max(5, round(a * 24)) % 17)) % 17"
630 ],
631 "widgets_values_named": {
632 "expression": "max(5, round(a * 24)) + (5 - (max(5, round(a * 24)) % 17)) % 17"
633 }
634 },
635 {
636 "id": 132,
637 "type": "PrimitiveFloat",
638 "pos": [
639 -1480,
640 6230
641 ],
642 "size": [
643 270,
644 70
645 ],
646 "flags": [],
647 "order": 8,
648 "mode": 0,
649 "inputs": [],
650 "outputs": [
651 {
652 "name": "FLOAT",
653 "type": "FLOAT",
654 "links": [
655 261
656 ]
657 }
658 ],
659 "title": "Float (Duration)",
660 "properties": {
661 "Node name for S&R": "PrimitiveFloat",
662 "cnr_id": "comfy-core",
663 "ver": "0.33.0"
664 },
665 "widgets_values": [
666 5
667 ],
668 "widgets_values_named": {
669 "value": 5
670 },
671 "color": "#322",
672 "bgcolor": "#533"
673 },
674 {
675 "id": 136,
676 "type": "MiniMaxH3ReferenceToVideo",
677 "pos": [
678 -610,
679 5070
680 ],
681 "size": [
682 400,
683 410
684 ],
685 "flags": [],
686 "order": 15,
687 "mode": 0,
688 "inputs": [
689 {
690 "name": "clip",
691 "type": "CLIP",
692 "link": 272
693 },
694 {
695 "name": "vae",
696 "shape": 7,
697 "type": "VAE",
698 "link": 273
699 },
700 {
701 "name": "audio_vae",
702 "shape": 7,
703 "type": "VAE",
704 "link": 274
705 },
706 {
707 "label": "ref_image_0",
708 "name": "ref_images.ref_image_0",
709 "shape": 7,
710 "type": "IMAGE",
711 "link": 297
712 },
713 {
714 "label": "ref_image_1",
715 "name": "ref_images.ref_image_1",
716 "shape": 7,
717 "type": "IMAGE",
718 "link": 298
719 },
720 {
721 "label": "ref_image_2",
722 "name": "ref_images.ref_image_2",
723 "shape": 7,
724 "type": "IMAGE",
725 "link": null
726 },
727 {
728 "label": "ref_video_0",
729 "name": "ref_videos.ref_video_0",
730 "shape": 7,
731 "type": "IMAGE",
732 "link": null
733 },
734 {
735 "label": "ref_video_audio_0",
736 "name": "ref_video_audios.ref_video_audio_0",
737 "shape": 7,
738 "type": "AUDIO",
739 "link": null
740 },
741 {
742 "label": "ref_audio_0",
743 "name": "ref_audios.ref_audio_0",
744 "shape": 7,
745 "type": "AUDIO",
746 "link": null
747 },
748 {
749 "name": "ref_audios.ref_audio_1",
750 "shape": 7,
751 "type": "AUDIO",
752 "link": null
753 },
754 {
755 "name": "ref_audios.ref_audio_2",
756 "shape": 7,
757 "type": "AUDIO",
758 "link": null
759 },
760 {
761 "name": "prompt",
762 "type": "STRING",
763 "widget": {
764 "name": "prompt"
765 },
766 "link": 279
767 },
768 {
769 "name": "width",
770 "type": "INT",
771 "widget": {
772 "name": "width"
773 },
774 "link": 276
775 },
776 {
777 "name": "height",
778 "type": "INT",
779 "widget": {
780 "name": "height"
781 },
782 "link": 277
783 },
784 {
785 "name": "length",
786 "type": "INT",
787 "widget": {
788 "name": "length"
789 },
790 "link": 275
791 }
792 ],
793 "outputs": [
794 {
795 "name": "positive",
796 "type": "CONDITIONING",
797 "links": [
798 270
799 ]
800 },
801 {
802 "name": "LATENT",
803 "type": "LATENT",
804 "links": [
805 271
806 ]
807 }
808 ],
809 "properties": {
810 "Node name for S&R": "MiniMaxH3ReferenceToVideo",
811 "cnr_id": "comfy-core",
812 "ver": "0.33.0"
813 },
814 "widgets_values": [
815 "",
816 1344,
817 768,
818 124,
819 "match"
820 ],
821 "widgets_values_named": {
822 "prompt": "",
823 "width": 1344,
824 "height": 768,
825 "length": 124,
826 "ref_image_size": "match"
827 },
828 "color": "#322",
829 "bgcolor": "#533"
830 },
831 {
832 "id": 138,
833 "type": "PrimitiveStringMultiline",
834 "pos": [
835 -1480,
836 6430
837 ],
838 "size": [
839 360,
840 480
841 ],
842 "flags": [],
843 "order": 9,
844 "mode": 0,
845 "inputs": [],
846 "outputs": [
847 {
848 "name": "STRING",
849 "type": "STRING",
850 "links": [
851 279
852 ]
853 }
854 ],
855 "title": "Input Text (Prompt)",
856 "properties": {
857 "Node name for S&R": "PrimitiveStringMultiline",
858 "cnr_id": "comfy-core",
859 "ver": "0.33.0"
860 },
861 "widgets_values": [
862 "subject_definitions:\n<Subject 1> is the station exterior in <Picture 2>. Reference role: form and material anchor. Features to follow exactly: a wide white saucer-shaped upper hull with visible panel seams and small antennas on top, a lit yellow-orange window band running around the rim, a cluster of white cylindrical modules hanging beneath the saucer with thin antenna spars at the bottom, weathered white paint with dark seams between plates, and the blue limb of a planet with white cloud bands behind it.\n<Subject 2> is the pilot in <Picture 1>. Reference role: identity and gear anchor. Features to follow exactly: a dark matte flight helmet with a wide glossy black visor, one thin orange glowing ring on the left side of the visor and two horizontal orange glowing bars on the right side, a segmented metal collar ring under the jaw, and a bulky dark flight suit with a soft padded cowl. Do not carry over the light grey studio background of <Picture 1>.\n<Subject 3> is the station interior, invented to match <Subject 1>: a dark rectangular hangar bay with ribbed white-metal walls, deep structural ribs, a grated catwalk running across the middle distance, small cargo clamps along the walls, and orange strip lights recessed between the ribs that match the colour of the pilot's visor glow.\n\nsummary:\n[reference generation] A five-second single continuous shot: the station of <Picture 2> hangs in orbit, the camera pushes in toward its saucer rim, the bay mouth opens and the camera travels inside, where the hangar lights come up to reveal the pilot of <Picture 1> standing on the catwalk. Exterior and interior are one move, no cut.\n\nretention_analysis:\n<Subject 1> (appears in [Shot 1]): fully_preserved - saucer hull, panel seams, yellow-orange rim band, hanging module cluster, antenna spars.\n<Subject 2> (appears at the end of [Shot 1]): fully_preserved - helmet silhouette, glossy visor, the orange ring and two orange bars, collar ring and padded cowl.\n<Subject 3> (appears at the end of [Shot 1]): attribute_transfer - the ribbed white-metal language and window-band orange of <Subject 1> are carried inside as hull ribs and strip lights; the bay layout itself is new.\n<Picture 1> background: weak_reference only - its light grey studio backdrop is discarded entirely.\nNo audio reference is used in this generation.\n\ndetailed_description:\nRefined retro-futurist cinematography in a single unbroken move, cold blue ambient outside against warm orange interior light, deep blacks in both halves, fine grain and no lens flares; the camera is the only thing that travels and it never cuts.\n\n[Shot 1] Wide establishing frame: the station of <Subject 1> hangs in the black, centred and slightly below eye level, the planet's blue limb and white cloud bands filling the lower third behind it. The saucer's rim band glows a soft orange, the hanging module cluster underneath reads as a dense white cylinder mass with thin spars, and the hull plates are weathered white with dark seams. The camera pushes in with small amplitude at slow speed, straight toward the saucer rim, and the station grows until the window band fills the upper half of the frame.\n\nAt about one and a half seconds a rectangular bay mouth opens in the rim below the band: two armour plates slide apart, spilling a warm orange rectangle of light into the vacuum, and a faint haze of vented gas drifts out of it. The camera does not stop and does not slow down; it continues into the opening, the frame going completely black for a beat as it crosses the threshold, then the interior lights of <Subject 3> come up in rows from the far end toward the lens, stripping in along the ribs one section at a time.\n\nAt about three seconds the camera is inside, gliding down the length of the ribbed hangar at chest height, the grated catwalk beneath it, cargo clamps and rib shadows sliding past on both sides, and the orange strip lights drawing long converging lines toward the far bulkhead. The camera decelerates and comes to rest at about four and a half seconds, framing <Subject 2> standing on the catwalk in three-quarter profile: dark helmet, glossy black visor with the orange ring and the two orange bars lit, padded cowl, suit reading as a soft dark mass against the white ribs. The last beat holds completely still while the pilot turns their head a few degrees toward the lens and the visor glow spills a thin orange edge onto the nearest rib.\n\noverall_soundscape:\nDeep vacuum silence outside with only a low sub-bass hum from the station's structure, then a hard change as the camera crosses the threshold: pressurised air rushing past the lens, the hydraulic whine of the bay plates closing behind, distant bulkhead clanks, the hum of the hangar settling into a steady tone, and one soft suit life-support hiss from the pilot at the end.\n\nnon_diegetic_music:\nA slow retro synthwave pad that holds one low chord through the exterior, drops to near silence for the threshold crossing, and returns as a warm rising two-note figure as the hangar lights come up, ending on the held final frame."
863 ],
864 "widgets_values_named": {
865 "value": "subject_definitions:\n<Subject 1> is the station exterior in <Picture 2>. Reference role: form and material anchor. Features to follow exactly: a wide white saucer-shaped upper hull with visible panel seams and small antennas on top, a lit yellow-orange window band running around the rim, a cluster of white cylindrical modules hanging beneath the saucer with thin antenna spars at the bottom, weathered white paint with dark seams between plates, and the blue limb of a planet with white cloud bands behind it.\n<Subject 2> is the pilot in <Picture 1>. Reference role: identity and gear anchor. Features to follow exactly: a dark matte flight helmet with a wide glossy black visor, one thin orange glowing ring on the left side of the visor and two horizontal orange glowing bars on the right side, a segmented metal collar ring under the jaw, and a bulky dark flight suit with a soft padded cowl. Do not carry over the light grey studio background of <Picture 1>.\n<Subject 3> is the station interior, invented to match <Subject 1>: a dark rectangular hangar bay with ribbed white-metal walls, deep structural ribs, a grated catwalk running across the middle distance, small cargo clamps along the walls, and orange strip lights recessed between the ribs that match the colour of the pilot's visor glow.\n\nsummary:\n[reference generation] A five-second single continuous shot: the station of <Picture 2> hangs in orbit, the camera pushes in toward its saucer rim, the bay mouth opens and the camera travels inside, where the hangar lights come up to reveal the pilot of <Picture 1> standing on the catwalk. Exterior and interior are one move, no cut.\n\nretention_analysis:\n<Subject 1> (appears in [Shot 1]): fully_preserved - saucer hull, panel seams, yellow-orange rim band, hanging module cluster, antenna spars.\n<Subject 2> (appears at the end of [Shot 1]): fully_preserved - helmet silhouette, glossy visor, the orange ring and two orange bars, collar ring and padded cowl.\n<Subject 3> (appears at the end of [Shot 1]): attribute_transfer - the ribbed white-metal language and window-band orange of <Subject 1> are carried inside as hull ribs and strip lights; the bay layout itself is new.\n<Picture 1> background: weak_reference only - its light grey studio backdrop is discarded entirely.\nNo audio reference is used in this generation.\n\ndetailed_description:\nRefined retro-futurist cinematography in a single unbroken move, cold blue ambient outside against warm orange interior light, deep blacks in both halves, fine grain and no lens flares; the camera is the only thing that travels and it never cuts.\n\n[Shot 1] Wide establishing frame: the station of <Subject 1> hangs in the black, centred and slightly below eye level, the planet's blue limb and white cloud bands filling the lower third behind it. The saucer's rim band glows a soft orange, the hanging module cluster underneath reads as a dense white cylinder mass with thin spars, and the hull plates are weathered white with dark seams. The camera pushes in with small amplitude at slow speed, straight toward the saucer rim, and the station grows until the window band fills the upper half of the frame.\n\nAt about one and a half seconds a rectangular bay mouth opens in the rim below the band: two armour plates slide apart, spilling a warm orange rectangle of light into the vacuum, and a faint haze of vented gas drifts out of it. The camera does not stop and does not slow down; it continues into the opening, the frame going completely black for a beat as it crosses the threshold, then the interior lights of <Subject 3> come up in rows from the far end toward the lens, stripping in along the ribs one section at a time.\n\nAt about three seconds the camera is inside, gliding down the length of the ribbed hangar at chest height, the grated catwalk beneath it, cargo clamps and rib shadows sliding past on both sides, and the orange strip lights drawing long converging lines toward the far bulkhead. The camera decelerates and comes to rest at about four and a half seconds, framing <Subject 2> standing on the catwalk in three-quarter profile: dark helmet, glossy black visor with the orange ring and the two orange bars lit, padded cowl, suit reading as a soft dark mass against the white ribs. The last beat holds completely still while the pilot turns their head a few degrees toward the lens and the visor glow spills a thin orange edge onto the nearest rib.\n\noverall_soundscape:\nDeep vacuum silence outside with only a low sub-bass hum from the station's structure, then a hard change as the camera crosses the threshold: pressurised air rushing past the lens, the hydraulic whine of the bay plates closing behind, distant bulkhead clanks, the hum of the hangar settling into a steady tone, and one soft suit life-support hiss from the pilot at the end.\n\nnon_diegetic_music:\nA slow retro synthwave pad that holds one low chord through the exterior, drops to near silence for the threshold crossing, and returns as a warm rising two-note figure as the hangar lights come up, ending on the held final frame."
866 },
867 "color": "#322",
868 "bgcolor": "#533"
869 },
870 {
871 "id": 145,
872 "type": "LoraLoaderModelOnly",
873 "pos": [
874 -1490,
875 5140
876 ],
877 "size": [
878 640,
879 90
880 ],
881 "flags": [],
882 "order": 13,
883 "mode": 0,
884 "inputs": [
885 {
886 "name": "model",
887 "type": "MODEL",
888 "link": 285
889 }
890 ],
891 "outputs": [
892 {
893 "name": "MODEL",
894 "type": "MODEL",
895 "links": [
896 294
897 ]
898 }
899 ],
900 "properties": {
901 "Node name for S&R": "LoraLoaderModelOnly",
902 "cnr_id": "comfy-core",
903 "ver": "0.33.0",
904 "models": [
905 {
906 "name": "pika_h3_flex_ref2va_v1_bf16.safetensors",
907 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/loras/pika_h3_flex_ref2va_v1_bf16.safetensors",
908 "directory": "loras"
909 }
910 ]
911 },
912 "widgets_values": [
913 "pika_h3_flex_ref2va_v1_bf16.safetensors",
914 1
915 ],
916 "widgets_values_named": {
917 "lora_name": "pika_h3_flex_ref2va_v1_bf16.safetensors",
918 "strength_model": 1
919 }
920 },
921 {
922 "id": 147,
923 "type": "ManualSigmas",
924 "pos": [
925 0,
926 5340
927 ],
928 "size": [
929 370,
930 100
931 ],
932 "flags": [],
933 "order": 10,
934 "mode": 0,
935 "inputs": [],
936 "outputs": [
937 {
938 "name": "SIGMAS",
939 "type": "SIGMAS",
940 "links": [
941 296
942 ]
943 }
944 ],
945 "properties": {
946 "Node name for S&R": "ManualSigmas"
947 },
948 "widgets_values": [
949 "1.0, 0.972973, 0.923077, 0.8, 0.571429, 0.333333, 0.0"
950 ],
951 "widgets_values_named": {
952 "sigmas": "1.0, 0.972973, 0.923077, 0.8, 0.571429, 0.333333, 0.0"
953 }
954 },
955 {
956 "id": 149,
957 "type": "LoadImage",
958 "pos": [
959 -1060,
960 5990
961 ],
962 "size": [
963 400,
964 480
965 ],
966 "flags": [],
967 "order": 11,
968 "mode": 0,
969 "inputs": [],
970 "outputs": [
971 {
972 "name": "IMAGE",
973 "type": "IMAGE",
974 "links": [
975 297
976 ]
977 },
978 {
979 "name": "MASK",
980 "type": "MASK",
981 "links": []
982 }
983 ],
984 "title": "LoadImage (Picture 1 — pilot)",
985 "properties": {
986 "Node name for S&R": "LoadImage"
987 },
988 "widgets_values": [
989 "ref_pilot_profile.png",
990 "image"
991 ],
992 "widgets_values_named": {
993 "image": "ref_pilot_profile.png",
994 "upload": "image"
995 }
996 },
997 {
998 "id": 150,
999 "type": "LoadImage",
1000 "pos": [
1001 -630,
1002 5990
1003 ],
1004 "size": [
1005 400,
1006 480
1007 ],
1008 "flags": [],
1009 "order": 12,
1010 "mode": 0,
1011 "inputs": [],
1012 "outputs": [
1013 {
1014 "name": "IMAGE",
1015 "type": "IMAGE",
1016 "links": [
1017 298
1018 ]
1019 },
1020 {
1021 "name": "MASK",
1022 "type": "MASK",
1023 "links": []
1024 }
1025 ],
1026 "title": "LoadImage (Picture 2 — station)",
1027 "properties": {
1028 "Node name for S&R": "LoadImage"
1029 },
1030 "widgets_values": [
1031 "ref_subject_station.png",
1032 "image"
1033 ],
1034 "widgets_values_named": {
1035 "image": "ref_subject_station.png",
1036 "upload": "image"
1037 }
1038 },
1039 {
1040 "id": 152,
1041 "type": "CreateVideo",
1042 "pos": [
1043 340,
1044 5600
1045 ],
1046 "size": [
1047 284.158203125,
1048 174
1049 ],
1050 "flags": [],
1051 "order": 20,
1052 "mode": 0,
1053 "inputs": [
1054 {
1055 "name": "images",
1056 "type": "IMAGE",
1057 "link": 300
1058 },
1059 {
1060 "name": "audio",
1061 "shape": 7,
1062 "type": "AUDIO",
1063 "link": 301
1064 }
1065 ],
1066 "outputs": [
1067 {
1068 "name": "VIDEO",
1069 "type": "VIDEO",
1070 "links": [
1071 299
1072 ]
1073 }
1074 ],
1075 "properties": {
1076 "Node name for S&R": "CreateVideo"
1077 },
1078 "widgets_values": [
1079 24,
1080 "auto",
1081 "sRGB",
1082 "auto",
1083 "none"
1084 ],
1085 "widgets_values_named": {
1086 "fps": 24,
1087 "bit_depth": "auto",
1088 "color_space": "sRGB",
1089 "chroma_subsampling": "auto",
1090 "codec": "none"
1091 }
1092 }
1093 ],
1094 "links": [
1095 [
1096 250,
1097 120,
1098 0,
1099 121,
1100 1,
1101 "VAE"
1102 ],
1103 [
1104 251,
1105 119,
1106 0,
1107 122,
1108 1,
1109 "VAE"
1110 ],
1111 [
1112 253,
1113 129,
1114 0,
1115 125,
1116 0,
1117 "NOISE"
1118 ],
1119 [
1120 254,
1121 126,
1122 0,
1123 125,
1124 1,
1125 "GUIDER"
1126 ],
1127 [
1128 255,
1129 123,
1130 0,
1131 125,
1132 2,
1133 "SAMPLER"
1134 ],
1135 [
1136 270,
1137 136,
1138 0,
1139 126,
1140 1,
1141 "CONDITIONING"
1142 ],
1143 [
1144 271,
1145 136,
1146 1,
1147 125,
1148 4,
1149 "LATENT"
1150 ],
1151 [
1152 280,
1153 125,
1154 0,
1155 122,
1156 0,
1157 "LATENT"
1158 ],
1159 [
1160 281,
1161 125,
1162 0,
1163 121,
1164 0,
1165 "LATENT"
1166 ],
1167 [
1168 285,
1169 127,
1170 0,
1171 145,
1172 0,
1173 "MODEL"
1174 ],
1175 [
1176 294,
1177 145,
1178 0,
1179 126,
1180 0,
1181 "MODEL"
1182 ],
1183 [
1184 296,
1185 147,
1186 0,
1187 125,
1188 3,
1189 "SIGMAS"
1190 ],
1191 [
1192 261,
1193 132,
1194 0,
1195 131,
1196 0,
1197 "FLOAT"
1198 ],
1199 [
1200 272,
1201 128,
1202 0,
1203 136,
1204 0,
1205 "CLIP"
1206 ],
1207 [
1208 273,
1209 119,
1210 0,
1211 136,
1212 1,
1213 "VAE"
1214 ],
1215 [
1216 274,
1217 120,
1218 0,
1219 136,
1220 2,
1221 "VAE"
1222 ],
1223 [
1224 297,
1225 149,
1226 0,
1227 136,
1228 3,
1229 "IMAGE"
1230 ],
1231 [
1232 298,
1233 150,
1234 0,
1235 136,
1236 4,
1237 "IMAGE"
1238 ],
1239 [
1240 279,
1241 138,
1242 0,
1243 136,
1244 11,
1245 "STRING"
1246 ],
1247 [
1248 276,
1249 115,
1250 0,
1251 136,
1252 12,
1253 "INT"
1254 ],
1255 [
1256 277,
1257 115,
1258 1,
1259 136,
1260 13,
1261 "INT"
1262 ],
1263 [
1264 275,
1265 131,
1266 1,
1267 136,
1268 14,
1269 "INT"
1270 ],
1271 [
1272 299,
1273 152,
1274 0,
1275 92,
1276 0,
1277 "VIDEO"
1278 ],
1279 [
1280 300,
1281 122,
1282 0,
1283 152,
1284 0,
1285 "IMAGE"
1286 ],
1287 [
1288 301,
1289 121,
1290 0,
1291 152,
1292 1,
1293 "AUDIO"
1294 ]
1295 ],
1296 "groups": [
1297 {
1298 "id": 1,
1299 "title": "Models",
1300 "bounding": [
1301 -1520,
1302 4840,
1303 700,
1304 990
1305 ],
1306 "color": "#3f789e",
1307 "flags": []
1308 },
1309 {
1310 "id": 2,
1311 "title": "Sampling",
1312 "bounding": [
1313 -30,
1314 4840,
1315 690,
1316 660
1317 ],
1318 "color": "#3f789e",
1319 "flags": []
1320 },
1321 {
1322 "id": 4,
1323 "title": "Decoding and create video",
1324 "bounding": [
1325 -30,
1326 5530,
1327 700,
1328 300
1329 ],
1330 "color": "#3f789e",
1331 "flags": []
1332 },
1333 {
1334 "id": 5,
1335 "title": "User Inputs",
1336 "bounding": [
1337 -1520,
1338 5860,
1339 2190,
1340 1090
1341 ],
1342 "color": "#3f789e",
1343 "flags": []
1344 },
1345 {
1346 "id": 6,
1347 "title": "Conditioning",
1348 "bounding": [
1349 -790,
1350 4840,
1351 730,
1352 990
1353 ],
1354 "color": "#3f789e",
1355 "flags": []
1356 }
1357 ],
1358 "config": [],
1359 "extra": {
1360 "ds": {
1361 "scale": 0.27625899280575594,
1362 "offset": [
1363 2510.6424563615274,
1364 -3570.7729058279415
1365 ]
1366 },
1367 "frontendVersion": "1.58.5",
1368 "VHS_latentpreview": false,
1369 "VHS_latentpreviewrate": 0,
1370 "VHS_MetadataImage": true,
1371 "VHS_KeepIntermediate": true
1372 },
1373 "version": 0.4
1374}