Open zjcs opened 1 week ago
If you mean image to text, there are many options, for example Florence2 with PromptGen 2.0 -model is pretty good and light.
Not a image to text node, I want to a pure text node, which will be linked to the 'caption' in 'SUPIR UpScale' Node, then I can change the text prompt as any worlds.
As I known, the 'caption' in 'SUPIR UpScale' Node is optional; And there are a positive prompt and a negative prompt, If I change the positive prompt to what I want, is it same to act as the input 'caption'?
Very nice work. But Now, I want to add a text prompt to test the SUPIR performance.
How can I add a text input node in ComfyUI?
I want to input a text prompt, which is similar to the LLaVa in SUPIR?
Is there any existing ComfyUI Node in the Node Manager?