Can this do image+prompt to image+text?
#75
by darkbyte - opened
For example, can I send a picture (input image) together with an instruction (input prompt) to segment the input image into regions in some way by producing an image with solid color regions (the output image) and also return a JSON which tells what each color is (the output text)?
use this instead Qwen2.5-VL-72B-Instruct
with instructions
Qwen cannot output an image. That's the whole reason why I asked. It seems Janus in theory could accept and output text and image simultaneously on both end, it's just there's no example on how to do it.
这些图标是什么应用？
这些图标是什么应用？