⚜️ Style Selector

Three dropdowns, one output, no prompt-mangling of its own. Style Selector doesn't touch your text at all - it just packages a Main Style, a Lighting choice, and a Camera choice into a bundle for other FOCUS nodes to actually apply. If you've landed here expecting a node that rewrites your prompt directly, you want Style Injector instead; this is the picker-only half.

Where it fits

FOCUS gives you three ways to pick a style, and it's worth knowing which is which before you build around one:

• SDXL Control Module has its own Main Style/Lighting/Camera dropdowns baked in alongside your stage/CFG settings, for the pack's two-stage workflow.
• Style Selector (this node) is the standalone version - just the picker, output as a DICTIONARY, for wiring into SDXL All-In-One's style input when you're not using the full Control Module/Control Pipe setup.
• Style Injector skips the dictionary entirely and rewrites your positive/negative prompt text directly, as plain strings.

Inputs and outputs that matter

• Main Style - 6 choices: Inactive, FOCUS Photography, FOCUS Portraits, FOCUS Macros, FOCUS Landscapes, FOCUS Fine Art. This is the master switch; leave it on Inactive and the rest doesn't matter downstream.
• Lighting - 9 choices: Inactive, Warm, Cold, Nightclub, Natural, Low Key, High Key, Flashlight, Studio.
• Camera - 22 choices, and this is the one worth taking seriously. It's not generic "professional camera" fluff - it's specific gear: a Canon EOS 5D Mark IV with a 24-70mm f/2.8L II, a Hasselblad X1D II with its 65mm f/2.8, a GoPro HERO9, a DJI Mavic Air 2 drone camera, even an "Old Polaroid." Naming a real physical camera and lens is, per the current community consensus on photorealistic prompting, one of the highest-leverage phrases you can use to knock a model out of default "beauty mode" - this pack baked that habit into a dropdown before most people had articulated why it worked.

Every category defaults to Inactive on its own axis, so you can pick a Camera and skip Lighting, or vice versa. Output is a single styles DICTIONARY - wire it into SDXL All-In-One's style socket.

Installing it

Via ComfyUI Manager: search Comfyui FOCUS nodes, install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/DJ-Tribefull/Comfyui_FOCUS_nodes

Restart. Nothing extra to download - the style categories live in the pack's bundled focus_style.csv.

Common issues & troubleshooting

Nothing seems to change. Two likely causes, in order of frequency: Main Style is still on Inactive (it defaults that way), or the styles output isn't actually wired into anything. This node has no effect sitting on its own - it has to feed SDXL All-In-One (or downstream through Control Pipe) to do anything visible.

Style feels weak even with everything set. Per the pack author's own note, effectiveness depends on your checkpoint and LoRA stack, and on how long your base prompt already is. SDXL runs on CLIP's 77-token attention window, so a long, cluttered prompt genuinely competes with the style keywords for the model's attention rather than just diluting them for taste reasons. Trim the base prompt before assuming the style categories don't work.

You want more than six Main Style presets. This node only exposes what's in focus_style.csv. Add your own rows to that file (matching its existing format) rather than expecting new categories to appear here automatically.