Feature request: Image support for Gemini Live

[aristid]

Gemini Flash now can create nice graphs, other images may follow. Could you please add support for that to gemini.py?

I added the following to _handle_evt_model_turn()

        elif mime_type.startswith("image/"):
            logger.debug(f"Received image from Gemini: {mime_type}")
            image = base64.b64decode(inline_data.data)
            if not image:
                return

            image_message = {
                "role": "assistant",
                "content": [{
                    "type": "image_url",
                    "image_url": {"url": f"data:{mime_type};base64,{inline_data.data}"}
                }],
                "source": "tool"
            }

            await self.push_frame(LLMFullResponseStartFrame())
            await self.push_frame(LLMMessagesAppendFrame(messages=[image_message]))
            await self.push_frame(LLMFullResponseEndFrame())

Maybe it should be added to context as well, not sure. I added an RTVI processor to the pipeline to forward image frames to the client. It more or less works, but I'm sure there is a better way to add this functionality, I don't yet have a good understanding of Pipecat internals.

Could you say whether you plan to add this in the next weeks or is it low priority? Thank you.