Dev tests SAM3 + VITMatte Photoshop plugin, finds masks still not fine enough
Just_Second9861 · reddit · 2026-10-09
A developer built a Photoshop plugin combining SAM3 and VITMatte for smart mask/selection generation, but results disappointed: SAM3's segmentation masks lack the resolution needed for photo editing and can only output binary masks, defeating the purpose of a smart masking tool. VITMatte enables feathered edges but struggles with complex scenes. After also trying SDMatte and other matting solutions—none combining contextual guidance, fine detail, and alpha transparency—the dev asked Reddit for a better smart/fast/high-quality masking approach.
More from Multimodal
- Autoregressive Retriever (ARR) Refines Queries with Retrieved Item Feedback via SFT and RL — _reachsumit · 2026-10-09
- Sony's Syn-Omni: Shared + Expert LoRA Paths Beat Omnimodal Embedding Baselines Across 81 Tasks — _reachsumit · 2026-10-09
- RISEBench++: 65 reasoning-based visual editing tasks; best model GPT-Image-2.5 hits only 56.6% — VisionXLab · 2026-10-09
- LEGO: lifting-free exocentric-to-egocentric video generation beats depth-lifting SOTA pipelines — 25frms · 2026-10-09
- VibeEdit Replaces Text Prompts with Canvas Marks, Scoring 79.9 on Edit Benchmark — Sydney-Uni · 2026-10-09
- Microsoft's Compo Shifts Poster Generation from Prompting to Spatial Composing — microsoft · 2026-10-09