Nomad Studio 1.1.0

Qwen Image 2.1 Turbo for editing, and output shape control in Edit Qwen 2.1.

New

• Turbo variation for Qwen 2.1 editing. Edit Qwen 2.1, Multi-reference Compose and every flow on the Qwen 2.1 edit engine have Advanced › Variation: Normal / Turbo. Turbo uses the distilled Qwen Image 2.1 Turbo checkpoint at 8 steps: on an RTX 4060 Ti, an edit goes from ~41 s to ~20 s and a multi-reference compose from ~77 s to ~40 s, with the subject's identity kept. Quantization now lists the precisions of the selected variation, and Steps follows it (8 / 25).
• Output aspect ratio in Edit Qwen 2.1. The Resolution panel offers Original (same as the source, the default), a set of ratios (9:16 … 21:9) that keep the source's pixel count in the new shape, and a typeable Custom resolution (width × height).
• MCP: nomad_studio_refine_image and nomad_studio_mix_images accept variation (normal / turbo); nomad_studio_refine_image accepts aspect_ratio or width/height on Edit Qwen 2.1. The action lists say which flows accept them.

Packs

Bundled: Core Refine 0.1.6, Extract 0.1.12, Presets Core 0.1.10. The Ko-fi packs that use the Qwen 2.1 edit engine have new versions with the Turbo variation; they need this version of the app.

Turbo needs qwen_image_2.1_turbo_int8_convrot.safetensors (or the BF16 one) in ComfyUI — the same checkpoint as the Qwen Image 2.1 Turbo engine in Create, so installing that engine is enough. Without it the Turbo pill shows as not installed.

Download

• Windows (x64): NomadStudio-1.1.0-windows-x64.zip — unzip and run NomadStudio.bat
• Linux (x64): NomadStudio-1.1.0-linux-x64.tar.gz
• macOS (Apple Silicon, experimental, unsigned): NomadStudio-1.1.0-macos-arm64.tar.gz

Already on 1.0.x? The app offers the update by itself ("Restart and update").

Checksums in SHA256SUMS.txt.