MAIN GALLERY/AI-ENHANCED GALLERY/SIGNSHOP

1 Daz Model 4 Ethnicities 3D Render, Digital Artwork by signshop

! BLOCK USER

! REPORT

PROFILE

1 Daz Model 4 Ethnicities

AI-Enhanced 3D Render by signshop

This experiment is using a set up / posed and rendered Daz character that I wanted to see if AI could change the ethnicity of the original Daz Character used by typing an slight adjustments in the main Text prompt.

The basic posed figure had strategically placed items in specific areas, mainly focusing on the Gloves of 'Shadow Thief Outfit - Gloves', hoping AI would generate a cyborg looking arm, but I also allowed the AI freedom to influence certain aspects of the composition. While AI doesn’t always produce exactly what you envision, it did generate some great looking images. (personal opinion smile

)

I included the ethnicities African, Asian, Brazilian, and Swedish in the text prompt, along with a different background color, to create more variations in the final AI generated renders.

The SECOND image shows the original Daz posed character and other ethnicities generated.

The THIRD image shows all four ethnicities in the ComfyUI ControlNet utilizing the 'ControlNet Preprocessor - Lineart' to maintain composition stability. The adjustments were made in the 'APPLY CONTROLNET' box, where I experimented with the 'strength' and 'end-percent' settings at various levels.

While most of the generated images turned out distorted and unusable, a few had potential. To enhance the facial details, I processed them through 'Topaz Gigapixel AI - Redefine Beta,' improving their overall quality.

More than 300 images were generated to arrive at these four final selections, and although there where a number of other considerations, they would have had a lot more Photoshop work on them to get them to a finished state. The final four had light Photoshop adjustments to refine them, adding subtle 'Noise' to achieve a film-like quality in the final main image.

The below YOU TUBE Video that I used to learn ComfyUI, producing what I have learned to create the images here...