Finish sketches for artist img2img

Hi, I am an artist and I wanted to try to integrate into my telegram bot the finalization of my sketches using image-to-image generation. But I managed to use and correctly get the result only with model stabilityai/stable-diffusion-xl-refiner-1.0 via inference_client . However, as I understand it is not suitable for my purposes because it does not change the original image only slightly correct it at any settings

Is there any way to solve my problem via API so that
1)it was free (within at least 200-300 free requests per month)
2)the calculations took place NOT on my computer
Thank you.

Translated with DeepL.com (free version)

1 Like

The Inference API doesn’t allow you to do complex image-to-image tasks, so if you want to use Hugging Face to do that, I think you’ll need to use the Endpoint API or Spaces.
The Gradio Spaces that are made public can often be used from the Gradio Client API.

Spaces usually don’t cost anything to the user, and at most they give you a slight advantage with subscriptions, so I think it’s a good idea to try them out first.