Photographic images Dall-e 3

[zealous walrus]

I work mostly with Midjourney but I am trying to work with Dall-e 3 on GPT to make photographic images but they mostly seem to have a game render look. I am not using the following words -
photorealistic
realistic
realism
photorealism

I am only asking in the positive and not for things I don't want.

I have tried

"movie scene
cinematic photo
candid photo
stock photo"
and a combination of above.

I have included film types, "award winning photography" "by a professional photographer" "f/2.5" and other photographic references.

What I haven't done

Describe the subject in detail but rather used archetypes.

and of course I can't use any contemporary photographers names.

Do you have a suggestion on how to get more photographic images or are we sort of stuck with a more game render look when working with Dall-e 3 and GPT?

[lean cliff]

Is this closer to what you are looking for?

I wrote the following prompt based on what I saw in your sample image: "Professional photography, an orange tabby cat sleeping on the floor, contemporary living room in white and light wood in the background, diffuse sunlight."

I used the DALL-E GPT by OAI. Its prompt was: "A professional photography scene featuring an orange tabby cat sleeping peacefully on the floor. The background depicts a contemporary living room, styled with white walls and light wood furniture. Diffuse sunlight illuminates the scene, creating a warm and serene atmosphere. The focus is on the natural beauty and tranquil posture of the cat in this domestic setting."

[zealous walrus]

Thank you for this. Is it important that you describe the scene and subject in detail? Will that impact the photographic quality? I have been using Dall-e via GPT as well. Yours is definitely much better. I find that since GPT is sending the prompt through to Dall-e the results can vary wildly. Ideally, I would like to use a basic archetype instead of getting specific about the subject but that may not be as feasible as it is in Midjourney.

[lean cliff]

I don't use MJ but the idea of "archetype" may not translate in the same way in DE. In general, short prompts leave a lot of room for ChatGPT/DALL-E to interpret what you mean. If you are looking for a particular style, composition, etc., it's better to supply that information.

Also, just to make sure we are talking about the same thing, when you say "GPT," do you mean "ChatGPT 4.0"? I'm asking because there are many "GPTs" inside CHATGPT. DALL-E GPT I used is one of them, and I get better results from it than ChatGPT itself.

[zealous walrus]

That is good to know. No I am using GPT 4 and using Dall-e 3 in the conversation. In terms of capability, I see fairly good photographic output on Bing and Dall-e 3 itself but it seems when you involve GPT it tends to go toward that game render look. I could absolutely be wrong it's just a theory. Do you think that could be the case?

[lean cliff]

It depends on which one of the GPTs you are using, I guess. Some of them are built for specialized purposes (for example I have different custom GPTs for different art media). OAI's DALL-E GPT is a good versatile one to start with. There might also be ones specifically built for photography. Click on "Explore GPTs" to see an extensive list of publicly available GPTs.