Image generation
Diffusers + SDXL + LoRA stacks running locally. Batch CSV/XML pipelines for marketing, OG image automation, and bulk content.
Real outputs from real users. Every video, image, and clip below was produced by Guaardvark running on consumer hardware — no cloud APIs, no rented GPUs.
All clips below were rendered on a single RTX 4070 Ti SUPER. Click through to the gallery for full-resolution sources and prompts.
Cinematic banner shot, golden-hour lighting, motion blur. Wan2.2 t2v.
View clip → Joker — rain, nightCharacter-consistent expressive close-up. Wan2.2 t2v.
View clip → Knight + holographic bookMixed-genre fantasy/sci-fi. Surface detail, parallax, glow. Wan2.2.
View clip → Batman perched, nightCinematic compositing, depth, atmospheric haze.
View clip → Knight silhouette — desertMinimal palette, dramatic backlighting, slow camera move.
View clip → View the full gallery130+ clips and stills generated through Guaardvark's image and video pipelines.
All work →Diffusers + SDXL + LoRA stacks running locally. Batch CSV/XML pipelines for marketing, OG image automation, and bulk content.
Wan2.2 + CogVideoX through the ComfyUI integration. 5–30 second clips, t2v and i2v supported.
Whisper.cpp transcription + Piper TTS synthesis. Audio Foundry stitches, edits, and overlays.
The same agent layer that powers the platform also generates outputs — content drafts, code, transcripts, web research summaries.