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

<a rel="ugc" href="https://github.com/Creepybits/ComfyUI-Creepy_nodes">Added a node </a>to switch between Gemini models. Currently supporting:Gemini 2.0 FlashGemini 2.5 Flash PreviewGemini 2.5 Pro Experimental

Files with SHA256: 9c591d1d6636f3cc21d0518d7bc4a441f9be6bf6820601d91f53f3999833eedb