Show HN: DreamOmni2 – Unified Multimodal Generation and Editing AI

Hi HN, We’ve been building DreamOmni2, a multimodal AI model that understands both text and image instructions, letting creators generate or edit visuals through natural, intuitive guidance.

Key Features:

Multimodal Understanding: Combine text + image prompts for richer, more accurate control.

Unified Generation & Editing: One model for both creation and modification — DreamOmni2 adapts intelligently to your goal.

Consistency Mastery: Maintain identity, pose, or layout across edits — ideal for storytelling, branding, and design workflows.

Advanced Editing Tools: Replace objects, backgrounds, or lighting — even imitate fonts and hairstyles through natural language.

Open Source & Research Ready: Built for creators and researchers with transparent benchmarks and reproducible pipelines.

Use Cases: Creative Direction & Art Design Portrait & Fashion Editing Product Visualization Typography & Branding Complex Compositional Creation

Why we built this: Creative workflows are often fragmented — one model for generation, another for editing, and none for maintaining consistent visual identity. DreamOmni2 unifies all of that, enabling seamless multimodal creation and iteration.

We’d love your feedback on how multimodal AI could improve your creative workflow — or what integrations you’d like to see next.

Try it here: https://www.dreamomni.net/