Using poses and sprtiesheet for one Image

Unity Engine
,
1

Dear Games Development Community,

As a newcomer to the world of artistry, I have a question regarding character development. I have successfully generated an ideal image front view for my character using AI, specifically utilizing Midjourney and Stable Diffusion. However, I am facing the challenge of generating other views and poses that the AI did not assist with, despite my repeated attempts. Should I decide to proceed with a 2D or 3D approach, what techniques and tools can I employ to achieve the desired result?

Thank you in advance for your guidance and expertise.

An Example :

2

From my experience with stable diffusion, once you manage to generate a single pose (possibly with few variations), then you can easily generate different poses with plugins like:

for example, play around with this while using ControlNet:

8966580--1232349--4PGjFvX.png
8966580--1232349--4PGjFvX.png1024×512 31 KB

1 Like
3

You can’t do that reliably with image generators. Most of the time you’ll get different characters.

Now, to be more specific, Civitai has CharTurner embedding.
And reddit had those posts:
https://www.reddit.com/r/StableDiffusion/comments/zyfaeb/charturner_a_work_in_progress_resource_for/
https://www.reddit.com/r/StableDiffusion/comments/1180fls/sharing_my_openpose_template_for_character/

But the process is not reliable, and generatio nof other views is not guaranteed.

Meaning once you get a template you’re happy with, you’d want a human to work with it. Meaning the human would need to paint the other views or make a model based on a single image.

1 Like