need a prompt to produce a described character that results in a head-toe image

[SoCalSteveT]

I have tried several prompt attempts to generate a described character that results in a full-body, head-t0-toe image.

[Kevin Stohlmeyer]

@SoCalSteveT can you give examples?

[SoCalSteveT]

Hi - great meeting you.  I am trying to create concept characters for a potentinal story.  Here is the prompt I used -Generate full length photorealistic image of an android, with the complete figure visible from head to toe in a vertical orientation showing the entire body. The android should be tall and slender, with metallic skin in bronze or rose-gold hues. Facial features blend Norwegian human traits with elfin characteristics, including high cheekbones, pointed ears, and arched eyebrows, complemented by a human-like nose and mouth. The hair is made of luminescent fiber-optic strands, shining in various colors, styled to frame the face. The figure should be standing in a full-length pose, clearly displaying head, torso, arms, legs, and feet.

 What are your thoughts?  Thanks, Steve

[Kevin Stohlmeyer]

Dont use a GPT to create overly complicated prompts is my first thought.

[SoCalSteveT]

Worked mostly well on Photoshop Beta generative fill.  I think the GPTs need to be built/trained on detail to be useful for comerical purposes.  What are your thoughts?

[Abambo]

Your prompt is very complex. Try to simplify. Some elements, like portrait mode, should be set with flags.

[Brandy Nova]

I've struggled with this as well.  I found it helps to specify the  surroundings in more detail, such as what the person/thing is standing on like a pedestal, or the room has a high ceiling.   I also take the GPT generated prompt and simplify it, and add the room descriptions early on so it is weighted more.   Once I get a full figure I use it as a style reference.  I would love to hear other people's experiences too.

Here's a prompt for example that gave me some good results:

"Full-length photorealistic image of an android standing confidently on a white pedestal in a room with a high ceiling. The android is tall and slender, with metallic skin in bronze or rose-gold hues. The android's hair is glowing in various colors, made of luminescent fiber-optic strands. The android has Norwegian facial features blended with elfin characteristics, including pointed ears, high cheekbones, and arched eyebrows. 3/4 full-length profile. Hair is styled to frame the face. Simple gray background. 4k, Long shot, dramatic"

[SoCalSteveT]

I tried a couple of times to attach my results pre and post your prompt suggestion on both Firefly and Phototshop Beta generative function.  Thanks for your help.  What types of images are you creating?

[Brandy Nova]

Right now, I am testing various image generators, in addition to Firefly, as well as different prompt styles-  art styles, themes, mediums, photo options, etc.  As an artist, I've been trying to recreate or enhance some of my own artwork using these ai tools.  Currently I have been working on two compositions.  One is a futuristic setting with various types of people (which is why your android example was relevant to me.)  The second is a composition of three cats (from real life) in a specific setting.  Getting the images to resemble their real-life counterparts has been challenging, as well as getting their full bodies in the scene without being cut off (as you know!)  It has been very rewarding, but sometimes very frustrating.  

[SoCalSteveT]

Sounds very interesting.  The text-to-image generation is great for artwork concept design.  Have you used the generative aspects of Photoshop Beta?

[Brandy Nova]

I haven't!  I need to give it a try too, thanks for the tip 🙂 

[SoCalSteveT]

Try having the people standing on a platform or other raised object in order to be above groound level.