Fashion photographer on AI product fidelity: Nano Banana best so far, FLUX 2/3 fall short
Active_Operation_546 · reddit · 2026-10-23
A fashion photographer using AI to place real garments, bags and accessories into generated scenes says product fidelity remains the core blocker: images look convincing overall, but shape, proportions, materials, stitching, logos and hardware often shift — unacceptable for commercial work.
He has tested Nano Banana (current pick) and found FLUX 2/FLUX 3 disappointing; several other models output too little detail to even verify text and logos. He asks:
- Which models or tools best preserve the actual product while changing scene or model?
- Do dedicated virtual try-on models beat general image editors?
- Do multiple reference views, product-specific training, or structured specs (measurements, materials) meaningfully help?
He's open to commercial tools, open-source pipelines and hybrid Photoshop workflows, and wants before/after examples.
Related event: Fashion photographer finds AI image generators fail product fidelity(2 posts)→
More from Multimodal
- Bria's Fibo-2.0-edit-turbo Image-Editing Model Trends on Hugging Face — briaai · 2026-10-24
- Kimi K3.1 Pro lands on VoxelBench with a detailed voxel cat sculpture — legit_api · 2026-10-24
- Google Gemini's Omni pops popcorn shaped like a middle finger in viral video prompt — michaelrabone · 2026-10-24
- SeedAudio 1.5 brings one-prompt cinematic audio generation to CapCut PC — thetripathi58 · 2026-10-24
- Open-source Phonon-2.5: 156MB speech model beats 20x-larger Whisper large-v3 — julianweisser · 2026-10-24
- MiniMax H3 reference-to-video preserves actors' performances but caps at 9s on a 3090 — ItsLukeHill · 2026-10-24