Uses custom prompt generation to allow Image to Video generation with only 8gb vram. The workflow uses Florence2 to analyze your starting image then send that information to Groq. Groq has a custom prompt for generating LTX video prompts. You can also add custom details that are appended to the Florence2 prompt before Groq processes it.
v1.1
-Added an input prompt to offer user control of the final product. This additional prompt is added to the bottom of the Florence2 description before feeding it to Groq
-Replaced Height and Width fields with a CR Aspect Ratio which allows you to set your dimensions and quickly swap them without having to retype.
-Moved frame count to an external node for ease of use.
-Consolidated the seeds for Florence2 and Groq to allow easy duplication for rerenders.
-Color coded nodes and added titles to important nodes
-corrected the .json file name misspelling.
This is a continuation of the original model as Civit's update system wasn't allowing me to put the new workflow in because I'm probably an idiot.