This workflow (and nodes) will load between 1-100 images from a folder, sequentially or randomly.The images are then analyzed and a number of keywords that represents the images are output.The keywords are then made into a prompt, using Gemini.The idea is to get more randomness and more inspirations.V1.0 requires that <a target="_blank" rel="ugc" href="https://github.com/ShmuelRonen/ComfyUI-Gemini_Flash_2.0_Exp">Gemini Flash 2.0</a> node and <a target="_blank" rel="ugc" href="https://github.com/Creepybits/ComfyUI-Creepy_nodes">Creepy Nodes</a> are installed.For Gemini PRO version there's a Gemini node included in Creepy Nodes<h2 id="audio-to-random-promptimage-4w766ivb7">Audio to random prompt/image</h2>Similar to previous versions, this workflow will extract random prompts and create images from them.The difference is that this workflow will use audio to create the prompts.<ol><li>Upload an audio clip (max 10 minutes long)</li><li>Set how long the clip you want to use should be in the Random/Fixed Audio Picker node</li></ol>

Create random prompts (and images) from batches of images

Similar to previous versions, this workflow will extract random prompts and create images from them.The difference is that this workflow will use audio to create the prompts.<ol><li>Upload an audio clip (max 10 minutes long) </li><li>Set how long the clip you want to use should be in the Random/Fixed Audio Picker node</li><li>Either set the starting time in seconds for the audio or set it to random (-1)</li></ol>Works with Flux models, Shuttle 3.1 (and probably some others as well).

Files with SHA256: 4d5a936a294470fecc430f226400a8e259d84c84f2d2865fd0f5072af39e0245