Images generated by the bot are not consistent with images drawn by ChatGPT for the same prompt. Some thoughts:
Might be an issue with the default drawing params (e.g natural vs `vivid)
Might be differences in the size of the image generated by chatgpt vs the bot (we default to 1024x1024)
Are we not invoking DALL-E3 correctly when the draw command is executed? (This one is unlikely I think, I'm fairly confident that d3 is being invoked correctly)
Images generated by the bot are not consistent with images drawn by ChatGPT for the same prompt. Some thoughts:
natural
vs `vivid)