File size: 4,976 Bytes
0997c23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Gemini Util - Configurable Options Guide

## Overview

The Gemini Describe functionality has been enhanced with configurable options through two nodes:

1. **Gemini Util - Options**: Configuration node that defines how descriptions should be generated
2. **Gemini Util - Media Describe**: Processing node that accepts media and optional configuration

## New Gemini Util - Options Node

### Purpose

This node provides granular control over description generation by separating configuration from media processing. Connect this node's output to the `gemini_options` input of the Media Describe node.

### Configuration Options

#### API & Model Settings

-   **Gemini API Key**: Your Google Gemini API key
-   **Gemini Model**: Choose from available models (2.5-flash, 2.5-flash-lite, 2.5-pro)
-   **Model Type**: Text2Image or ImageEdit workflow type

#### Description Control (New Boolean Options)

-   **Describe Clothing?** [Yes/No]: Include detailed clothing and accessory descriptions
-   **Describe Hair Style?** [Yes/No]: Include hair texture and motion (but not color/length)
-   **Describe Bokeh?** [Yes/No]: Allow depth of field effects and blur descriptions
-   **Replace Action with Twerking?** [Yes/No]: Replace video movement description with twerking content

#### Text Options

-   **Prefix Text**: Text to prepend to generated descriptions

## Updated Media Describe Node

### Changes

-   **Removed**: API key, model selection, description mode combo box, prefix text
-   **Added**: Optional `gemini_options` input that accepts configuration from Options node
-   **Maintained**: All media handling capabilities (upload, random selection, image/video support)

### Backward Compatibility

When no Options node is connected, the Media Describe node uses these defaults:

-   Describe Clothing: No
-   Describe Hair Style: Yes
-   Describe Bokeh: Yes
-   Replace Action with Twerking: No
-   API Key: Default development key
-   Model: gemini-2.5-flash

## Usage Examples

### Basic Usage

1. Add "Gemini Util - Options" node to workflow
2. Configure desired settings
3. Add "Gemini Util - Media Describe" node
4. Connect Options node output to Media Describe `gemini_options` input
5. Provide media input (image tensor or upload)

### Option Combinations

#### For Minimal Descriptions (No Clothing, No Hair, No Bokeh)

-   Clean, simple descriptions focusing on core elements
-   Results in 3-paragraph structure: Subject, Cinematic, Style

#### For Detailed Fashion Analysis (Clothing + Hair)

-   Comprehensive descriptions including garments and hairstyles
-   Results in 4-paragraph structure when clothing enabled

#### For Sharp Focus Photography (No Bokeh)

-   Explicitly prevents depth-of-field language
-   Useful for product photography or architectural scenes

### System Prompt Improvements

#### Enhanced Decisiveness

All prompts now include instructions to avoid uncertain language:

-   ❌ "She appears to be wearing either lace tights or leggings"
-   ✅ "She wears black lace tights"

#### Hair Style Option

New granular control over hair descriptions:

-   When enabled: Includes texture and movement
-   When disabled: Completely omits hair references
-   Always excludes: Color and length (as before)

#### Dynamic Paragraph Structure

-   3 paragraphs: Subject, Cinematic, Style (no clothing)
-   4 paragraphs: Subject, Cinematic, Style, Clothing (with clothing)
-   5-6 paragraphs for video: Adds Scene and Movement sections

## Migration Guide

### For Existing Workflows

Existing Media Describe nodes will continue working with default settings. To use new features:

1. Add Options node to workflow
2. Configure desired settings
3. Connect to Media Describe node
4. Remove any hardcoded API keys from Media Describe node

### Option Mapping

Old combo box "Description Mode" maps to new options as follows:

-   "Describe without clothing" → Clothing: No, Hair: Yes, Bokeh: Yes
-   "Describe with clothing" → Clothing: Yes, Hair: Yes, Bokeh: Yes
-   "Describe without clothing (No bokeh)" → Clothing: No, Hair: Yes, Bokeh: No
-   "Describe with clothing (No bokeh)" → Clothing: Yes, Hair: Yes, Bokeh: No

## Benefits

1. **Modularity**: Configure once, use with multiple media nodes
2. **Flexibility**: Mix and match options as needed
3. **Clarity**: Each option has a clear purpose
4. **Extensibility**: Easy to add new options in the future
5. **Precision**: More deterministic descriptions without uncertainty

## Troubleshooting

### No Description Generated

-   Verify Options node is connected to Media Describe
-   Check API key is valid
-   Ensure media input is provided

### Unexpected Content

-   Review individual option settings
-   Check if hair/clothing/bokeh options match expectations
-   Verify model type (Text2Image vs ImageEdit) is appropriate

### Backward Compatibility Issues

-   Media Describe works standalone with defaults
-   Only connect Options node if you need custom configuration