Hunyuan Image 3.5 realism test: lighting, materials and camera language
HeyAmit_ · x · 2026-09-23
Part 3 of HeyAmit's Hunyuan Image 3.5 preview test pushes realism: the model focuses on details like lighting, materials, camera language, environmental details and textures, which make a huge difference in how believable the final image feels.
More from Multimodal
- Skillry's free GrokBot Avatar skill generates mascot and soft-vinyl style avatars — yihui_indie · 2026-09-23
- NetEase Youdao's R2T2 + T3PO form a fully open-source real-time interpretation stack — thetripathi58 · 2026-09-23
- NetEase Youdao open-sources Confucius4-R2T2 and T3PO, a full real-time voice AI stack — thetripathi58 · 2026-09-23
- Midjourney LOTR caricature style with sref 2458935905, then Nano Banana for photoreal conversion — michaelrabone · 2026-09-23
- PixVerse world model lets you walk through AI-generated video with WASD — Aiden_Tech_Ai · 2026-09-23
- What do agents think they look like? This project has a painter draw their self-descriptions — Mastbubbles · 2026-09-23