Dynamic Poses LoRA V2 is here, now with even more versatility and precision! Available for SDXL and Flux, this updated version takes action-packed imagery to new heights. Trained on a larger, more diverse dataset over 10k steps, V2 features a broader range of poses—perfect for capturing the energy of athletic movements like acrobatics, gymnastics, running, and more. Whether you're crafting dynamic storytelling scenes or showcasing high-energy moments, this LoRA brings unparalleled realism and vibrancy to every frame.
Ready to see the difference? Explore the world of advanced reality generation with PolyPhaze on Patreon. Unlock exclusive content, tools, and techniques to power your creative journey—click here to join!
Description
This version of the Lora model focuses on generating highly dynamic and stylized images of men and women in various action-oriented scenes. The model has been fine-tuned to excel in creating cinematic compositions that blend strong lighting contrasts, fluid movement, and detailed expressions. It’s particularly well-suited for urban and modern settings, where dramatic lighting (like natural daylight or studio-controlled lights) can enhance the overall atmosphere.
Key features include:
Hair Movement and Light Play: The model captures the natural flow and movement of hair, especially when illuminated by dynamic light sources, creating realistic and dramatic effects.
Clothing Contrast: A strong emphasis on black-and-white clothing contrasts gives each image a sleek, modern feel, while still maintaining a high level of detail in the fabric and textures.
Background Blurring for Focus: Urban backdrops are rendered with a slight blur to evoke depth while keeping the subject in sharp focus, mimicking the effect of professional studio photography.
Playful and Cinematic Action: The model excels in rendering energetic poses and playful body language, making it ideal for action or movement shots, whether in intense, cinematic scenes or lighthearted, joyful moments.
Balanced Lighting: The Lora model effectively uses different light sources—whether it’s soft daylight or stark studio lighting—to create mood and depth, balancing shadows and highlights to enhance realism and visual appeal.
This version is versatile, designed for creators looking to explore a mix of action, urban themes, and modern aesthetics while maintaining a professional, studio-quality look.
FAQ
Comments (15)
Can you make a sexy version too? Flux seriously needs better sexy poses.
Well, actually I did one specific for that...
New lora Flux - ProReal Vision | Patreon
I think if you're asking for 5000 buzz for early access you need to provide better comparison images, some of these poses work without a lora.
@futureflix good examples thanks, with something like flux that is quite versatile you should show off the difference
Was looking forward to trying this in Flux / Forge. I can see it beside my other loras, but when I select it it doesn't show up in the prompt. Downloaded it twice with no luck, and all other loras work just fine. Can someone please help me!
Well the author couldn't find a solution to my problem - it doesn't work for me in Forge. Has anyone got it working in Forge, and if so, how?
I'm not using Forge but this is a Onetrainer LoRA with only attention defined (rank=64, alpha=1). Onetrainer LoRA's are somewhat rare. Maybe ask in the Forge github if this is supported or maybe at Onetrainer's github if he might kindly supply some conversion tools.
@Alentejo Thanks for the tip! I'll look into it after the stress of the festive season has died down a bit :-)
When i try to reproduce the picture i get the following Missing nodes output from ComfyUI:
AutoResSDXL(In group node 'workflow>FLUX Model Loader')
workflow>FLUX Model LoaderRemove from workflow
workflow>JoyCaption GenRemove from workflow
workflow>PromtpGen1.5 GenRemove from workflow
And the ComfyUI manager can't find them - any ideas waht i should install ?
Hello! I used JoyCaption and PromptGen to caption the dataset.
FLUX Model Loader was node I created that has CLIP and FLUX Model all in one
@futureflixHello! How can I intall this nodes?
@futureflix Thanks, I'm trying to run this
Works fine on my system, but takes a lot of time, about 1h per picture in this resolution ... anyway, the result is great!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.






