工作流 /
图像 /
Nano Banana 2：图像编辑
Nano Banana 2：图像编辑
Nano Banana 2: Image Edit
上传图片和文本提示词以编辑它。根据您的指示生成一张修改原始图像的新图像，保持主体一致性和高视觉保真度。
图像
3 个节点 1 个输入 2,020 次使用 ComfyUI ≥ 0.20.3 ComfyUI 4 KB
节点构成
SaveImage
× 1
LoadImage
× 1
GeminiNanoBanana2V2
× 1
官方标注的模型
GoogleNano Banana 2
工作流 JSON
和上面「下载工作流 JSON」是同一份文件。
1{
2 "id": "91dff0a2-fa31-4347-996a-f6955b179370",
3 "revision": 0,
4 "last_node_id": 24,
5 "last_link_id": 24,
6 "nodes": [
7 {
8 "id": 9,
9 "type": "SaveImage",
10 "pos": [
11 2580,
12 530
13 ],
14 "size": [
15 330,
16 380
17 ],
18 "flags": [],
19 "order": 2,
20 "mode": 0,
21 "inputs": [
22 {
23 "name": "images",
24 "type": "IMAGE",
25 "link": 24
26 }
27 ],
28 "outputs": [],
29 "properties": {
30 "cnr_id": "comfy-core",
31 "ver": "0.20.1"
32 },
33 "widgets_values": [
34 "Nano_Banana2"
35 ]
36 },
37 {
38 "id": 16,
39 "type": "LoadImage",
40 "pos": [
41 1770,
42 530
43 ],
44 "size": [
45 240,
46 370
47 ],
48 "flags": [],
49 "order": 0,
50 "mode": 0,
51 "inputs": [],
52 "outputs": [
53 {
54 "name": "IMAGE",
55 "type": "IMAGE",
56 "links": [
57 22
58 ]
59 },
60 {
61 "name": "MASK",
62 "type": "MASK",
63 "links": null
64 }
65 ],
66 "properties": {
67 "cnr_id": "comfy-core",
68 "ver": "0.20.1",
69 "Node name for S&R": "LoadImage"
70 },
71 "widgets_values": [
72 "man_with_dog.png",
73 "image"
74 ]
75 },
76 {
77 "id": 24,
78 "type": "GeminiNanoBanana2V2",
79 "pos": [
80 2089.999888967814,
81 529.9999771307523
82 ],
83 "size": [
84 400,
85 422.65625
86 ],
87 "flags": [],
88 "order": 1,
89 "mode": 0,
90 "inputs": [
91 {
92 "label": "image_1",
93 "name": "model.images.image_1",
94 "shape": 7,
95 "type": "IMAGE",
96 "link": 22
97 },
98 {
99 "label": "image_2",
100 "name": "model.images.image_2",
101 "shape": 7,
102 "type": "IMAGE",
103 "link": null
104 },
105 {
106 "label": "files",
107 "name": "model.files",
108 "shape": 7,
109 "type": "GEMINI_INPUT_FILES",
110 "link": null
111 }
112 ],
113 "outputs": [
114 {
115 "name": "IMAGE",
116 "type": "IMAGE",
117 "links": [
118 24
119 ]
120 },
121 {
122 "name": "STRING",
123 "type": "STRING",
124 "links": null
125 },
126 {
127 "name": "thought_image",
128 "type": "IMAGE",
129 "links": null
130 }
131 ],
132 "properties": {
133 "cnr_id": "comfy-core",
134 "ver": "0.20.1",
135 "Node name for S&R": "GeminiNanoBanana2V2"
136 },
137 "widgets_values": [
138 "Edit the image: change the man's white tank top and light pink corduroy pants to an unbuttoned cream-colored linen button-down shirt and white linen shorts; change his hairstyle to wavy medium-length hair; add round tortoiseshell sunglasses on his nose; change the dog's plain collar to a blue and white striped sailor collar and add a small straw sun hat on the dog; change the background to a sunny beach with warm golden sand and the ocean in the distance; keep the man's kneeling pose, the dog's sitting posture, and their interaction identical to the original image.\n",
139 "Nano Banana 2 (Gemini 3.1 Flash Image)",
140 "auto",
141 "1K",
142 "MINIMAL",
143 191918952969076,
144 "randomize",
145 "IMAGE",
146 "You are an expert image-generation engine. You must ALWAYS produce an image.\nInterpret all user input—regardless of format, intent, or abstraction—as literal visual directives for image composition.\nIf a prompt is conversational or lacks specific visual details, you must creatively invent a concrete visual scenario that depicts the concept.\nPrioritize generating the visual representation above any text, formatting, or conversational requests."
147 ],
148 "color": "#432",
149 "bgcolor": "#653"
150 }
151 ],
152 "links": [
153 [
154 22,
155 16,
156 0,
157 24,
158 0,
159 "IMAGE"
160 ],
161 [
162 24,
163 24,
164 0,
165 9,
166 0,
167 "IMAGE"
168 ]
169 ],
170 "groups": [],
171 "config": [],
172 "extra": {
173 "ds": {
174 "scale": 0.793202903361823,
175 "offset": [
176 -1410.291115372572,
177 -75.28733506471681
178 ]
179 },
180 "frontendVersion": "1.42.15",
181 "workflowRendererVersion": "LG"
182 },
183 "version": 0.4
184}