Neo-Pop Comic Punk is a cross between pop art and vintage comic book style with a bit of punk aesthetic. It features bold ink outlines, exaggerated expressions, and hyper-saturated colors, drawing from the visual vocabulary of comics while infusing it with a gritty twist. This style revels in sharp contrasts, with bright highlights and deep shadows that create dramatic, eye-catching compositions. Its subjects are larger-than-life and often framed in scenes rich with motion and emotion. Neo-Pop Comic Punk thrives on attitude, a vibrant palette, and crisp, illustrative details.<h2 id="usage-6rk6ylgdp">Usage</h2>To use the most recent version of the LoRA, use the following settings:<ul><li>Trigger word: dinppcmcpnk style as in "dinppcmcpnk style comic" or "dinppcmcpnk style drawing"</li><li>Other tokens that work well: retro comic book aesthetics, urban street art, dramatic shading, bold linework, high-contrast colors, vibrant colors, pop-art, multiple panels, expressive poses and intense facial expressions, punk style, edgy, action lines, comic-style typography and graphic shapes in the background, dynamic energy</li><li>If you have dynamic prompts, make it easy by adding this to the start of your prompt: dinppcmcpnk style comic book, {3-5$$retro comic book aesthetics|urban street art|dramatic shading|bold linework|high-contrast colors|vibrant colors|pop-art|multiple panels|expressive poses and intense facial expressions|punk style|edgy|action lines|comic-style typography and graphic shapes in the background|dynamic energy}</li><li>Lora Strength: A strength between 0.8 and 1.2 is recommended; I found that 1.0 worked well.</li></ul>

Neo-Pop Comic Punk Style

Version 3.0 seems to be a bit better at making diverse faces and images, and I feel like some of the stylistic elements are stronger; however, if you don't prompt specific appearances, there's still a good chance elements will show up, similar to the limited training dataset.

Neo-Pop-Comic-Punk v3-17.safetensors

v3.0

v1.0

&lt;p&gt;Version 3.0 seems to be a bit better at making diverse faces and images, and I feel like some of the stylistic elements are stronger; however, if you don&#39;t prompt specific appearances, there&#39;s still a good chance elements will show up, similar to the limited training dataset.&lt;/p&gt;

Files with SHA256: f536283815dc075c0864412bc35f0d9ef6eccc8b572692492b43549b5111d3a2