Hi3D v1.0 Wins 83.5% of 562 Blind Comparisons Against Eight Image-to-3D Models
rohanpaul_ai · x · 2026-10-27
From rohanpaulai's Hi3D evaluation thread (3/3):
- Lineup: Hi3D v1.0 is the open base; the closed model is already at v3.0 refining at 2048 cubes per side. The new Hi3D Agent unifies generating, editing and processing 3D models in one flow, with an MCP server letting other AI agents call Hi3D directly
- Cost: the team claims over 50% reduction in users' credit costs
- Method: most image-to-3D tests only grade the input view, so every result's back and bottom were also inspected (hidden sides are necessarily plausible guesses)
- Results: 13 people with professional 3D-algorithm experience made 562 pairwise judgments vs eight image-to-3D models; Hi3D v1.0 won 83.5% overall — 73.2% vs TRELLIS.2, but only 51.4% vs Pixel3D
- Example: on a winged armored figure, TRELLIS.2 collapsed into a flat panel while Hi3D v1.0 kept wings and limbs as connected parts
Related event: Hi3D Beats Eight Rivals in 83.5% of 562 Image-to-3D Comparisons(2 posts)→
More from Multimodal
- Prompt Motion: Free Gallery of 200+ Claude Opus Motion Videos with Exact Prompts — victor_explore · 2026-10-27
- Depianoer: a joke app to strip the piano Seedance adds to every scene — umpherj · 2026-10-27
- kohya's kv-cache Control LoRA Implementation Cuts MOSS-Image Overhead to Half of Concat — bdsqlsz · 2026-10-27
- Opus 5.5 codes music, 3D stage and lighting for an anime MV with 38 synced dance moves — majidmanzarpour · 2026-10-27
- Redditor argues full HDR workflows, not SDR-to-HDR conversion, kill the "AI slop" look — robertwellesley · 2026-10-27
- $1M AI Film Award Returns: Entries Must Be 70% Seedance-Generated — HashemGhaili · 2026-10-27