.Net: Add Support for "dall-e-3" Model in OpenAITextToImageService Class

[Nurgo]

Hi,

I propose we add support for the "dall-e-3" model within the OpenAITextToImageService class. The integration seems straightforward and the benefit to the users would be substantial.

Thank you for considering this request.

[dmytrostruk]

@Nurgo Please check these notebooks, they use "dall-e-3" model as example for image generation:

• https://github.com/microsoft/semantic-kernel/blob/main/dotnet/notebooks/07-DALL-E-3.ipynb
• https://github.com/microsoft/semantic-kernel/blob/main/dotnet/notebooks/08-chatGPT-with-DALL-E-3.ipynb

[Nurgo]

Hi @dmytrostruk , thank you for your reply. These notebooks only use "dall-e-3" if AddAzureOpenAIChatCompletion is used, and not AddOpenAITextToImage. From what I can see in the code, OpenAITextToImageService still uses "dall-e-2". Moreover, the image formats supported (256x256, 512x512, 1024x1024) are those of "dall-e-2".

[dmytrostruk]

@Nurgo Got it. Yes, looks like dall-e-3 is only implemented in AzureOpenAITextToImageService class, but not in OpenAITextToImageService.