I successfully made a few model files with ollama and installed qwen2.5 coder 32b, llama 3.2-vision, and llava:34b. The first one to mimic Claude level coding as best as possible. It can not see images but the other 2 can.
Unfortunately the 2 vision models inside bolt.diy tell me they can not see the images as well even with built in image OCR abilities.
So how to i fix this or do i need to somehow get the image inside the file structure to have it be able to see it.
Is it possible to import code files from Vercel, Loveable or Bolt.new into the VS studio files so when i run bolt.diy i can use use the others image functionality to circumvent my issue?
Im not sure how to do either or if this isnt fixable or just not implemented yet.
Thank you for any information.