ArchAi3D_Qwen_Scene_Photographer

ArchAi3D_Qwen_Scene_Photographer is the pack's answer to the most common camera instruction people actually give: "go stand in front of that thing and take a picture of it." The README's example is literally "Go in front of some object and take a photo with that subject in front of camera view" - and this node turns that impulse into a structured Qwen-Edit prompt with 14 professional position presets on top.

The design insight, straight from the pack's research docs, is that Qwen-Edit understands natural-language positioning far better than pixel coordinates - so everything here is phrased in meters and directions, not grid positions. You say "3m to the right facing the espresso machine," the node says it properly. A key discovery the README is proud of: pixel coordinates are not supported by Qwen, and distance-based positioning beats arbitrary degree rotations for reliability.

The inputs that matter

• target_subject - what to frame ("the espresso machine on the counter", "the leather sofa"). The node's whole job is positioning the camera relative to this.
• preset - 14 professional positions (product front/hero-low/overhead, interior corner/opposite-wall/ceiling, architectural ground-up/elevated, food 45°/overhead, fashion, landscape). Set it and the direction/distance/height fields populate for you.
• direction + distance - which way and how far (1–20m). The tooltip repeats the pack's standard warning: left/right is picture-left/right, not the subject's.
• height - ground_level (dramatic low), face_level (natural), higher (overview).
• tilt - none, way_up (ceiling/sky), way_down (bird's-eye).
• auto_facing - default true: bakes "facing the target" into the prompt so the subject stays centered. Turn it off for wide landscape shots where you want the general view.

Outputs: prompt (into the encoder), preset_description, and system_prompt (into the encoder's system input - carries the "you are a photographer" framing that stabilizes the scene).

How it works

A preset (or manual fields) maps to a "vantage point at [height] [distance]m to the [direction] facing [target]" instruction - the same vantage-point pattern the pack rates as its most reliable. The node then pairs it with a scene-preservation system prompt so the furniture, materials and lighting stay put while the camera repositions.

Installing it

Ships in ComfyUI-ArchAi3d-Qwen by Amir Ferdos (ArchAi3d):

# ComfyUI Manager: search "ArchAi3d Qwen"
cd ComfyUI/custom_nodes
git clone https://github.com/amir84ferdos/ComfyUI-ArchAi3d-Qwen.git
cd ComfyUI-ArchAi3d-Qwen
pip install -r requirements.txt

Restart. Bring your own Qwen-Image-Edit model (GGUF-quantized in practice) and the pack's encoder.

Common issues

• "Front of" vs "in front of" ambiguity. The pack's research says the model may read "in front of the camera" and "in front of the subject" differently; that's why auto_facing and the explicit "facing" clause exist. If the subject ends up off-frame, check whether auto_facing got switched off.
• Distance is approximate. 3m is directionally reliable, not photogrammetrically exact. Iterate rather than micromanage.
• Scene drift. Qwen-Edit re-emits the frame, so repositioning is also a subtle redesign. The system prompt is your stabilizer - keep it connected.

For product photography sessions, interior framings and the "just put me in front of that thing" requests, this is the pack's most immediately useful camera node.