Vision Vanguard 24 [Krea 2]
Documentary-style realism with micro-textures, dramatic natural light, and both gritty black-and-white and saturated color.
Base model: Krea 2
Trigger words: documentary photograph, eye-level medium shot
Modifiers: high-contrast black and white, saturated color palette, film grain, shallow depth of field
Strength: 0.8–1.0
Prompting: start with the camera language, then the subject and environment, then texture and lighting details.
Good for: documentary scenes, aerial landscapes (salt flats, dunes, mineral pools), industrial and rural decay, and macro still life.
Description
Base model: Krea 2
Strength: 0.8–1.0
Resolution: trained at 512, generate at 1024–1536
CFG: 3.5 (Base) / 1.0 (Turbo)
Steps: 35–50 (Base) / 8 (Turbo)
FAQ
Comments (4)
Should be really useful for cinematic images, im a bit confused, why is it named vanguard 24, if you uploaded it after vanguard 25?
They use a dataset based on images related to different years. So each vision vanguard is all based on photography. But they all have completely different datasets.
Vision Vanguard 25 = images from 2025
24 = images from 2024 etc.
I actually have datasets heading all the way back to 2012. Maybe I'll release them one day.
They should all be very good at generalization and any photographic content. Their databases are all at least 1000 hi-res images covering large diverse subject matter so they're really more just like an overall realism style if you want to view it like that.
@thedivergentai Speaking of years, Krea desperately needs some 80s/90s/early 2000s loras to get retro styles, can I make a request for them?
@TheEventHorizon Hey so I heard you on the decades themed stuff. I'm working on it. Trying to build up a nice large dataset. I'll start with the 80's.
Makeup of dataset will be something like:
Documentary: ~40% | Street, home, transit, candid life
Retail ephemera: ~30% | Ads, catalogs, packaging, mall signage, magazine pages
Fashion/music/events: ~20% | Concerts, sports, youth culture, neon fashion
Technology/gaming: ~10% | Consoles, arcades, computers, VHS-era objects
Details
Files
(Krea 2) Vision Vanguard 24 V2026.1.safetensors
Mirrors
179-Vision-Vanguard.safetensors
Vision Vanguard 24 v2026.1 - Krea2 - 0.8-1str,cfg3.5raw,1turbo,8stepsturbo,35-50raw,documentary photograph,eye-level medium shot,high-contrast black and white,saturated color palette,film grain,shallow depth of field.safetensors
Vision Vanguard 24 v2026.1 - Krea2 - 0.8-1str,cfg3.5raw,1turbo,8stepsturbo,35-50raw,documentary photograph,eye-level medium shot,high-contrast black and white,saturated color palette,film grain,shallow depth of field.safetensors
Vision Vanguard 24 v2026.1 - Krea2 - 0.8-1str,cfg3.5raw,1turbo,8stepsturbo,35-50raw,documentary photograph,eye-level medium shot,high-contrast black and white,saturated color palette,film grain,shallow depth of field.safetensors
Vision Vanguard 24 v2026.1 - Krea2 - 0.8-1str,cfg3.5raw,1turbo,8stepsturbo,35-50raw,documentary photograph,eye-level medium shot,high-contrast black and white,saturated color palette,film grain,shallow depth of field.safetensors