Building a character dataset from one image: img2img workflows keep failing to reproduce the same person
ThePerfectStormy · reddit · 2026-09-08
A Redditor wants to create a consistent AI character but can only rely on a single found image of a woman they like, since the exact person can't be regenerated from scratch.
Attempts to build img2img workflows with help from Grok and ChatGPT have all failed, and they're asking the community for proven character-consistency workflows to expand the dataset from that one reference image.
More from Multimodal
- "Sora Is Back": Viral Demo Shows Video Made with Astra — mallow610 · 2026-09-08
- Minimax H3 runs AI video locally at 720p — impressive but trails Seedance 2.5 — SimplyAnnisa · 2026-09-08
- fal extends 75% off H3 Max endpoints to Sept 15, adds real-time 1080p video — isidentical · 2026-09-08
- Dev turns every storefront he checked into at into photo-generated 3D miniatures — floguo · 2026-09-08
- Four Prompt-Free Grok Videos Show Off a Striking Visual Style — azed_ai · 2026-09-08
- Fully AI-generated video combines Qwen 3.8, Flux 2 Klein, Minimax H3, LTX 2.5 and Breeze TTS — CQDSN · 2026-09-08