📦 Object Focus Camera v5 (Professional Presets)

You know how product shots only look professional when the prompt describes why the material is shiny - the facets, the veining, the brushing - and how nobody actually types that stuff out for the hundredth time? V5 ("Professional Presets") is the node that got tired of it for you. It's V4's camera system plus two dropdowns stuffed with pre-written material and photography language.

The two new dropdowns

material_detail_preset is the star. 37 material-specific descriptions organized by category - Crystalline/Glass (Crystal Facets, Glass Transparency, Diamond Brilliance, Frosted Glass, Stained Glass, Ice Crystals), Metallic Surfaces (Polished Metal, Brushed Metal, Oxidized Patina), and more - plus "None (Manual entry)" if you'd rather type. Pick "Brushed Metal" and the node appends the exact technical description the model needs to render brushed metal convincingly instead of "shiny grey blob." This is the difference between a generic AI product render and something you could put in a spec sheet.

photography_quality_preset is the finishing move. 15 presets in two buckets: Technical Excellence (Razor Sharp Focus, Professional Lighting, High Dynamic Range, Color Accuracy, Bokeh Background) and Artistic Style (Editorial Quality, Commercial Product, Fine Art Photography, Documentary Realism). Each one is a pre-written clause appended to the prompt.

The rest of the panel is the V4 camera stack: camera_position (19), camera_movement (7), camera_distance (5), lens_type (10), prompt_language (English/Chinese/Hybrid - Chinese if you run dx8152's LoRAs), focus_transition_mode (Standard vs. Focus Transition), and add_detailed_explanation. target_object takes more specific subjects here - the sample choices are "chandelier crystal", "brass handle", "silk fabric" - and show_details still accepts a manual override that combines with the presets rather than replacing them.

How it works and what comes out

Same plain-language mechanism as the rest of the family, composed in layers: base camera instruction, then the material clause, then the quality clause, then the explanation depth. The three outputs are the usual prompt, system_prompt, and description. The system prompt stays the camera-operator-preserve-everything framing, which is what keeps "keep everything else identical" behavior when you're swapping in strong material language.

Where it fits

V5 is the one I'd actually reach for in the V3–V5 range for anything that will be seen by a client. The material presets are the killer feature - they're the kind of vocabulary that separates a convincing render from an obviously-AI one, and you get them without writing a paragraph per shot. V3 is fine for quick detail shots; V5 is for the shots that matter.

Install

Pack install, once:

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

Or search "ArchAi3d Qwen" in ComfyUI Manager. Restart, then look under ArchAi3d/Qwen/Camera. You still need the Qwen-Image-Edit weights themselves - quantized for consumer GPUs. Note the pack's dual license too: free for personal and non-commercial use, but client work counts as commercial and needs a license from the author. If you're generating for paying clients, that's not a footnote to skip.