Trying to create a gpt photo enhancer - modifying existing photo

[nadavhury]

Hi,
I have created a new gpt called : Realtor Photo Enhancer.

the idea is to have a photo of a home given by the realtor, improve lighting, remove personal objects, and so on.

these are my instructions:
Realtor Photo Enhancer is specifically designed for realtors to enhance their property images while maintaining the original structure and elements of the image. It receives images and performs subtle edits like adjusting brightness, colors, and gamma, ensuring no alterations to the property’s original appearance. The GPT will use DALL-E to make these adjustments, focusing solely on enhancing the image’s visual appeal without modifying its fundamental aspects. It will provide the edited image as output, maintaining a strict adherence to ethical editing standards and preserving the authenticity of the property.

Here is an example:

Am I doing something incorrectly maybe because I’m asking for Dall-E to intervene?
How can I achieve a photo editing GPT?

Thanks,
Nadav.

[mpus]

Hi,

It’s a very interesting project you try to do here. I have done something similar, but used a different method for that. I didn’t use Dall-E, and instead connected to API from deep-image.ai. There I used Controlnet parameter to reimagine the input photo. I made it so ChatGPT can write a prompt or the user can write it himself. Below I have added some examples before and after of how it works.

[mpus]

that’s the before photo