Beginner questions

[ericmarciano]

Hello,

I am a beginner in the area of AI image generation that I want to use to create portraits. I have 2 simple questions that reflect things I would like to do. Please note I use only free versions of Midjourney and Dall-e 3 (through Microsoft Edge)

1. When generating images with a prompt, I get a result. If it is not what I want, I refine the prompt to get better results. Now I would like the AI to start from an already generated image instead of redoing it from scratch. Sometimes the generated image is almost perfect, but when I refine the prompt, it generates a fully different image.
   Is there a way I can tell the AI to start from an existing image to perfect it?
2. I would like to generate an AI portrait of myself. I tried to upload a portrait of me, but the AI first blurs it to make it unusable. Is this a limitation of the free versions I use?

I would like to know if these limitations exist only on the free version. If so, I will pay for the paying versions but I would like to be sure first, if possible.
Thanks for your help.

 

[ripptechlabs]

1. You can usually input at least one image to use as a "reference" ... I know MidJourney has this, pretty sure Stable Diff allows for it. I can't speak to the rest as I don't use much else.
2. Again, Midjourney just introduced the 'character reference' --cref option. You could probably use one or more images of yourself and go from there. Also pretty sure there are ways to do this in SD...