CinematicRedmond SDXL LoRA for Film-Still Imaging
Learn how to use CinematicRedmond SDXL, a LoRA trained on film still images for cinematic image generation with consistent results.
On this page
CinematicRedmond SDXL: Film-Still LoRA for Consistent Cinematic Imaging
TL;DR with 3 actionable bullets:
- Use the trained words
Cinematic. Film Still.as your key trigger phrase - Best results with ComfyUI; apply to SDXL base models for consistent outputs
- Apache 2.0 licensed with 1.1K+ Civitai downloads and 69 likes
What It Is
CinematicRedmond SDXL is a LoRA (Low-Rank Adaptation) adapter trained on film still imagery. Unlike a full model, LoRAs modify base models through lightweight weight adapters, making them efficient for style transfer. This particular LoRA targets the cinematic film still aesthetic—a look characterized by dramatic lighting, shallow depth of field, and professional cinematography framing.
Training Background
The LoRA was trained on curated film still images, producing outputs with high detail and visual consistency across different subjects and scenarios. Training was supported by Redmond.AI, which provided GPU resources for the experiment.
How to Use
The recommended usage pattern:
- Load your base SDXL model (tested with Flux.2 Klein 9B, Qwen, and ZImageTurbo variants)
- Apply the CinematicRedmond SDXL LoRA at appropriate strength (typically 0.6-0.8)
- Use the trained trigger words:
Cinematic. Film Still. - Generate in 1024x1024 resolution for SDXL-based models

For ComfyUI workflows, the LoRA can be loaded as a separate node and connected to your UNet loader. In Automatic1111, place it in the models/Lora directory and enable via the LoRA tab.
Example Prompt
Cinematic. Film Still., masterpiece, high quality, detailed composition, cinematic lighting, clean background, shallow depth of field


Settings Recommendations
- CFG scale: 3-7 for optimal style balance
- Steps: 20-30 typically sufficient
- Sampler: Euler a or DPM++ 2M SDE
- LoRA strength: 0.6-0.8 (adjust based on base model)
Common Errors and Fixes
Error | Cause | Fix
|---|---|---
Output looks flat/unstyled | Trigger words not included | Add "Cinematic. Film Still." to prompt Lighting inconsistent | LoRA strength too high | Reduce weight to 0.5-0.7 Composition feels off | CFG too high | Lower CFG to 5-6
License
Licensed under Apache 2.0. Commercial and personal use permitted with attribution.