jan-prompt-presets

A collection of small utility nodes for ComfyUI that make common prompt-building and sampler configuration tasks faster. All nodes live under the prompt/presets category.

---

Nodes

Camera Presets

Outputs a prompt string for common camera angles and shot types.

| Option | Description | |---|---| | None / Custom | Empty output — use the custom override field | | Cinematic Hero | Low angle, dramatic perspective, foreshortening | | Dynamic Action | Dutch angle, tilted camera, wide angle lens | | Intimate Portrait | Close-up, eye level, shallow depth of field | | Character Showcase | Full body, eye level, centered composition | | Over-the-Shoulder Scene | OTS framing, medium shot, depth of field | | Epic Wide Scene | Wide shot, low angle, cinematic framing |

Output: camera_prompt (STRING) Optional input: custom_override — if filled, replaces the dropdown entirely.

---

Lighting Presets

Outputs a prompt string for common lighting setups.

| Option | Description | |---|---| | None / Custom | Empty output — use the custom override field | | Dramatic | Strong directional light, deep shadows, chiaroscuro | | Cinematic | Three-point lighting, colour grading, anamorphic flare | | Golden Hour | Warm sunlight, long soft shadows, amber tones | | Neon Night | Vibrant neon, cyan/magenta spill, cyberpunk atmosphere | | Moody Thriller | Low key, cool tones, noir atmosphere | | Soft Natural Daylight | Diffused overcast light, even illumination, airy feel | | Moonlit Night | Cool pale moonlight, silver and deep blue tones | | Studio | Professional softbox, clean white background |

Output: lighting_prompt (STRING) Optional input: custom_override — if filled, replaces the dropdown entirely.

---

SDXL Resolution Picker

Outputs width and height for all standard SDXL bucket resolutions (the aspect ratios the model was trained on).

| Option | Resolution | |---|---| | 1024 x 1024 (1:1 Square) | 1024 × 1024 | | 1152 x 896 (9:7 Landscape) | 1152 × 896 | | 1216 x 832 (3:2 Landscape) | 1216 × 832 | | 1344 x 768 (16:9 Landscape) | 1344 × 768 | | 1536 x 640 (12:5 Wide) | 1536 × 640 | | 896 x 1152 (7:9 Portrait) | 896 × 1152 | | 832 x 1216 (2:3 Portrait) | 832 × 1216 | | 768 x 1344 (9:16 Portrait) | 768 × 1344 | | 640 x 1536 (5:12 Tall) | 640 × 1536 |

Outputs: width (INT), height (INT)

---

Scheduler Selector

A clean dropdown for picking a scheduler, sourced directly from ComfyUI's core (comfy.samplers.KSampler.SCHEDULERS). Automatically includes any schedulers added by other installed packages.

Output: scheduler — type-compatible with the scheduler input on KSampler and similar nodes.

---

Sampler Selector

Same as Scheduler Selector but for samplers, sourced from comfy.samplers.KSampler.SAMPLERS.

Output: sampler_name — type-compatible with the sampler input on KSampler and similar nodes.

---

Text Split

Splits a multiline string into a list, one element per line.

Input: text (STRING, multiline), skip_empty_lines (BOOLEAN, default: true) Output: list (STRING_LIST)

---

List Get Index

Returns a single element from a STRING_LIST by index. Supports negative indices (-1 = last element).

Input: list (STRING_LIST), index (INT) Output: element (STRING)

---

LM Studio Vision Prompt Enhancer

Takes a start image and an optional short scene description, sends both to a vision-capable LLM running in LM Studio, and returns a Wan-optimised motion prompt ready to feed into a text encoder.

Designed to work alongside the ComfyUI-EBU-LMStudio node pack — use the EBU Load and Unload nodes to manage the LLM lifecycle around this node.

| Input | Type | Description | |---|---|---| | image | IMAGE | The start image to describe and expand | | model_name | STRING | Partial name of the model loaded in LM Studio (e.g. "qwen") | | url | STRING | LM Studio API endpoint (default: http://127.0.0.1:1234/v1/chat/completions) | | max_tokens | INT | Maximum response length (default: 300) | | temperature | FLOAT | Sampling temperature (default: 0.5) | | scene_description | STRING (optional) | Short hint to guide the output, e.g. "A girl dances on a tropical beach." — if empty, the LLM works from the image alone | | system_prompt | STRING (optional) | System prompt sent to the LLM — editable in the node for tuning without touching code |

Output: enhanced_prompt (STRING)

Requirements:

• LM Studio running with Developer Mode enabled
• A vision-capable model loaded (e.g. Qwen3 9B with mmproj file)

---

Installation

1. Clone or copy this folder into ComfyUI/custom_nodes/:

   git clone https://github.com/kursopiko/jan-prompt-presets.git
   

2. Restart ComfyUI.
3. All nodes will appear under the prompt/presets category.

---

Credits

Created by JAN (@kursopiko) with assistance from Claude (Anthropic).