Crafting stylized t-shirt image design prompts

Good day everyone,

I’m using DALL-E 3 image generator in Bing Copilot. Also using Promptmania prompt builder to enhance prompts at times. I have a print-on-demand t-shirt business that I generate social media prompts for in Gemini and ChatGPT 3.5. I’m trying to generate t-shirt designs for my products as well and need some help with this.

I’m new to AI, maybe 3 or 4 weeks now and I´m thrilled with what I have been able to product with these tools. The learning curve is steep but I’m doing ok with it. I’m stumped at the moment and would like to ask the advanced folks here for some assistance.

I am using Gemini and ChatGPT 3.5 to produce social media image prompts and getting good results. They are producing images that have a certain type of t-shirt design as part of the overall image. However the t-shirt designs I want are part of the larger image content produced by DALL-E 3. I am trying to produce prompts for the t-shirt designs similar in style to what you see on the t-shirts in the overall image. I have attached an example to demonstrate what I am describing:

Can anyone tell me how to craft prompts that will produce images in the style that appears on the t-shirt in this image but without being printed on a t-shirt and without the background? I can specify white background but if I eliminate the context of the prompt that produced this image I get results that are nothing like the designs I want. Here is the prompt I am using to generate this image. This image prompt was produced in ChatGPT 3.5:

Beautiful Lighting style, Artistic portrayal of a motivational quote on a tee, set against the raw urban landscape, evoking urban resilience. Motivation, urban grit DSLR Bold typography, urban backdrop.

I’m pretty much stumped here. Can anyone suggest wording that will produce images with designs like this without the other content?

Best regards to all. Much obliged for any help you can give.

2 Likes

Welcome to the community!

Hey, you’re 90% of the way there. How do you feel about these?

here’s your prompt, edited.

Beautiful Lighting style, Artistic three tone portrayal of a motivational quote set against the raw urban landscape, evoking urban resilience. Motivation, urban grit, bold typography, vector design, dithering.

this is what chatgpt ultimately rewrote it to:

Summary

A vector design featuring a motivational quote “Rise and Grind” in bold, artistic typography. The background showcases a raw urban landscape at sunrise, with buildings casting long shadows. The color palette is limited to three tones: a deep blue for the predawn sky, a vibrant orange for the sunlight, and a dark gray for the urban structures. The image incorporates elements of dithering to add texture and depth, creating a visually striking portrayal of urban resilience. The overall atmosphere is inspirational, with the beautiful lighting evoking a sense of a new beginning and grit. The style is reminiscent of high-quality ArtStation vector artwork.

Anyways, the trick to getting rid of the t-shirt is to get rid of the t-shirt. And the camera.

Don’t mention anything that you don’t want to see in the pic, and don’t ever tell it to not do something, because it will then do exactly that. Sometimes you gotta be a little creative in the rephrasing, But it’s doable in most cases.

What do you think about that?

3 Likes

Is this along the lines of what you’re looking for?

Your reply is incredibly helpful. I understand the login of your explanation and I’m already working on applying the methods for prompts.

As usual one thing inevitably leads to another so when new questions arise I now know there are some very knowledgeable people here on this forum that are willing to share their knowledge.

Many thanks Diet for your expertise.

1 Like

Hi elmstedt,

Yes this is in the neighborhood of the deigns I am looking for. Can you share the prompt you used for this? Do you have some suggestions on how to write prompts that produce images like the one you posted?

It’s really a new world learning to communicate with AI image and text models to get the results from them that I am looking for. Thankyou for your help.