工作流 /
视频 /
MiniMax H3：参考生视频
MiniMax H3：参考生视频
MiniMax H3: Reference to Video
使用MiniMax H3从参考图或一组多模态输入生成视频，该模型能够同时理解文本、图像、视频和音频，生成5-15秒的连贯片段并附带原生立体声音频。可输入最多2张图像来控制首帧和尾帧，或在全参考模式下输入最多9张图像、3个视频片段和3个音频片段，以您选择的比例输出单个视频。适用于商业内容创作、角色驱动叙事和快速视觉概念验证。
视频
3 个节点 1 个输入 131 次使用 ComfyUI ≥ 0.30.0 ComfyUI 4 KB
节点构成
LoadImage
× 1
SaveVideo
× 1
MinimaxHailuo03ReferenceNode
× 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": "d652d420-0e2b-42ec-a5a9-262f21f0bb6c",
3 "revision": 0,
4 "last_node_id": 4,
5 "last_link_id": 4,
6 "nodes": [
7 {
8 "id": 2,
9 "type": "LoadImage",
10 "pos": [
11 -2960,
12 -20
13 ],
14 "size": [
15 330,
16 360
17 ],
18 "flags": [],
19 "order": 0,
20 "mode": 0,
21 "inputs": [],
22 "outputs": [
23 {
24 "name": "IMAGE",
25 "type": "IMAGE",
26 "links": [
27 4
28 ]
29 },
30 {
31 "name": "MASK",
32 "type": "MASK",
33 "links": null
34 }
35 ],
36 "properties": {
37 "Node name for S&R": "LoadImage"
38 },
39 "widgets_values": [
40 "9panel_storyboard_golden_hour_clay_court.png",
41 "image"
42 ]
43 },
44 {
45 "id": 3,
46 "type": "SaveVideo",
47 "pos": [
48 -2140,
49 -20
50 ],
51 "size": [
52 480,
53 110
54 ],
55 "flags": [],
56 "order": 2,
57 "mode": 0,
58 "inputs": [
59 {
60 "name": "video",
61 "type": "VIDEO",
62 "link": 3
63 }
64 ],
65 "outputs": [
66 {
67 "name": "video",
68 "type": "VIDEO",
69 "links": null
70 }
71 ],
72 "properties": [],
73 "widgets_values": [
74 "video/MiniMax_H3",
75 "auto",
76 "auto"
77 ]
78 },
79 {
80 "id": 4,
81 "type": "MinimaxHailuo03ReferenceNode",
82 "pos": [
83 -2590,
84 -20
85 ],
86 "size": [
87 400,
88 320
89 ],
90 "flags": [],
91 "order": 1,
92 "mode": 0,
93 "inputs": [
94 {
95 "label": "image_1",
96 "name": "model.reference_images.image_1",
97 "shape": 7,
98 "type": "IMAGE",
99 "link": 4
100 },
101 {
102 "label": "image_2",
103 "name": "model.reference_images.image_2",
104 "shape": 7,
105 "type": "IMAGE",
106 "link": null
107 },
108 {
109 "label": "video_1",
110 "name": "model.reference_videos.video_1",
111 "shape": 7,
112 "type": "VIDEO",
113 "link": null
114 },
115 {
116 "label": "audio_1",
117 "name": "model.reference_audios.audio_1",
118 "shape": 7,
119 "type": "AUDIO",
120 "link": null
121 }
122 ],
123 "outputs": [
124 {
125 "name": "VIDEO",
126 "type": "VIDEO",
127 "links": [
128 3
129 ]
130 }
131 ],
132 "properties": {
133 "Node name for S&R": "MinimaxHailuo03ReferenceNode"
134 },
135 "widgets_values": [
136 "MiniMax H3",
137 "Use the uploaded 9-panel storyboard image as the exact visual reference.Create a single continuous cinematic shot of a golden-hour tennis match on a red clay court. Follow the storyboard from left to right, top to bottom, as one smooth unbroken sequence.A Black male player in a yellow polo, yellow shorts, and sporty sunglasses plays against a female player with braided hair, a beige tank top, and a light blue tennis skirt. Preserve their faces, outfits, court, lighting, and color palette exactly as shown.The sequence should move naturally through these moments:1. male player overhead smash2. female player sprinting across the court3. male player backhand follow-through4. extreme close-up of the male player's sunglasses and sweat5. female player jumping to hit the ball6. female player forehand contact with the yellow tennis ball7. male player low lunge on clay8. both players meeting at the net and shaking handsCinematic sports film style, golden hour sunlight, warm orange tones, photoreal, sharp focus, shallow depth of field, smooth steadicam movement, no cuts, no text, no logos, no broadcast graphics.",
138 "768P",
139 "adaptive",
140 5,
141 42,
142 "randomize",
143 false
144 ],
145 "color": "#432",
146 "bgcolor": "#653"
147 }
148 ],
149 "links": [
150 [
151 3,
152 4,
153 0,
154 3,
155 0,
156 "VIDEO"
157 ],
158 [
159 4,
160 2,
161 0,
162 4,
163 0,
164 "IMAGE"
165 ]
166 ],
167 "groups": [],
168 "config": [],
169 "extra": {
170 "ds": {
171 "scale": 0.9818181818181818,
172 "offset": [
173 3053.6458333333335,
174 386.2170862268519
175 ]
176 },
177 "frontendVersion": "1.47.11"
178 },
179 "version": 0.4
180}