ComfyUI-Variationator Introduction
ComfyUI-Variationator helps you explore different versions of an image idea without rewriting your prompt each time. You start with a subject, such as “sports car on highway,” then use ComfyUI nodes to add directions for lighting, weather, mood, camera framing, and other visual choices.
This is useful when you know what you want to depict but want to compare how it could look. You might test a product at several times of day, give a landscape different seasonal moods, or try multiple approaches to a portrait while keeping its subject consistent. You can make one carefully chosen variation, build a layered prompt, or generate a batch of prompts for comparison.
Variationator changes prompt text. Your image-generation workflow still interprets that text, so the exact visual result depends on the rest of your ComfyUI setup.
A note on the documented counts: Parts of the original README describe 12 dimensions, 240 modifiers, and 19 nodes. Its detailed node list and version history instead describe 11 modifier dimensions, 220 modifiers, and 18 nodes; the overview also names Distance as a twelfth dimension without listing a corresponding single-modifier node. The counts should be treated as inconsistent until confirmed against the extension itself.
How ComfyUI-Variationator Works
Think of your base prompt as the subject of a photo shoot and each modifier as one creative direction you give the crew. “Luxury watch on pedestal” identifies the subject. A lighting modifier might add warm golden-hour light; a camera modifier might suggest a close-up with a shallow depth of field. The combined text gives the image generator more specific instructions while keeping the original idea recognizable.
A typical workflow has three steps:
- Write a base prompt. Keep the subject clear, such as “mountain landscape.”
- Choose how to vary it. Add a single modifier, connect several modifier nodes, select a preset, or generate a batch.
- Connect the resulting text to your prompt workflow. Variationator’s nodes output strings (text), which can feed compatible ComfyUI prompt inputs and other text-processing nodes.
For example, starting with “street scene,” you could add golden-hour lighting and rain for a warm scene with wet-road reflections. Adding a cyberpunk style would push the same starting idea toward a more stylized result. Each extra instruction gives you another creative lever, but too many competing directions can make a prompt harder to control. Starting with one or two modifiers makes it easier to see what each one contributes.
Some nodes offer an append or replace choice. Appending keeps the existing prompt and adds a modifier; replacing uses the modifier text in place of the existing text. Where a preserve_base setting is available, keeping it enabled is usually the clearest way to retain your original subject.
ComfyUI-Variationator Features
Modifier Dimensions
The original README describes 20 prepared modifiers per documented dimension. A modifier is a ready-made piece of descriptive prompt text that you select by its ID rather than writing it from scratch.
| Dimension | What it helps you change | Example uses |
|---|---|---|
| Lighting | Time of day and quality of light | Compare sunrise, golden hour, blue hour, and night. |
| Seasonal | Seasonal appearance and atmosphere | Give one landscape spring bloom or winter frost. |
| Location | The setting around your subject | Place a subject in an urban, coastal, desert, or mountain environment. |
| Weather | Sky and atmospheric conditions | Explore clear weather, rain, fog, snow, or storms. |
| Mood | The scene’s emotional tone | Make an image feel calm, mysterious, energetic, or luxurious. |
| Camera | Photographic viewpoint and technique | Try different angles, shot types, lenses, or depth of field. |
| Color Grading | Overall palette and tonal character | Compare warm, cool, muted, vibrant, or cinematic treatments. |
| Action | Movement and activity | Give a subject a more dynamic situation or interaction. |
| Era | Period or imagined time setting | Explore a historical decade, a modern look, or a futuristic one. |
| Detail | The degree and character of visual detail | Compare a minimalist treatment with a more detailed one. |
| Composition | Arrangement within the image | Try symmetry, the rule of thirds, or negative space. |
The overview also mentions Distance, with examples ranging from macro views to aerial views, but the detailed node list does not establish how to select it. Check the nodes available in your ComfyUI installation before planning a workflow around that dimension.
Lighting, Seasonal, and Weather modifiers are also described with DRY and WET categories. These provide a way to explore contrasting conditions. For instance, a clear, dry scene and a wet, reflective one may give you very different starting points for the same subject.
Single-Modifier Nodes
The README lists 11 single-modifier nodes, one for each dimension in the table above. Choose an ID from 1–20 to apply a specific prepared modifier. These nodes are best when you want to repeat a particular look or understand exactly which instruction changed your prompt.
For a first experiment, use a base prompt such as “sports car on highway” with Lighting Modifier (Single) set to ID 7, documented as Golden Hour. The resulting prompt adds warm, low-angle lighting directions. You can then change only that ID to compare lighting ideas while leaving the subject alone.
You can connect nodes from different dimensions in sequence. For example, Lighting → Seasonal → Camera lets you build a more specific scene one choice at a time. Two to four complementary modifiers are a practical starting range; review the combined text if you add more.
Style Nodes
- Style Transform applies a built-in or custom artistic style. The README describes more than 10 built-in styles. Use it to explore how the same subject reads in approaches such as noir, impressionist, or cyberpunk.
- Multi-Style Layer stacks up to three styles. This offers more experimentation, though the styles you combine should support rather than contradict one another.
For example, “vintage motorcycle” can remain your subject while Style Transform changes the artistic treatment. Style choices are prompt directions, not separate image-generation models.
Batch and Preset Nodes
- Batch Variation Generator creates multiple prompt variations from one base prompt, with up to 10 variations described in the feature list. It is suited to side-by-side exploration, such as testing several lighting choices before committing to an image series.
- Preset Collection provides prepared sets for common creative goals. The README names
commercialfor product-oriented scenes,artisticfor creative exploration,key_momentsfor dramatic shots,full_day_cyclefor time-of-day progression, andwet_nightfor low-light, reflective scenes.
For a product study, you could begin with “luxury watch on pedestal” and try the commercial preset to explore different lighting directions. Presets offer a starting point; you can still refine the prompt for your specific subject.
Utility Nodes
- Modifier Info helps you look up a modifier before using it. This is useful when an ID alone does not tell you enough about the effect.
- Prompt Combiner joins text segments, helping you assemble a prompt from separate ideas.
- Random Modifier selects a modifier using a seed. Keep the same seed when you want to reproduce a selection; change it to explore another.
- Smart Variation (NLP) uses natural-language instructions to choose relevant modifiers. For example, an instruction like “make it golden hour with a wet road and cyberpunk style” expresses the intended look without requiring you to select every ID manually. Clear, specific terms are more useful than vague requests.
The README says Variationator nodes appear under Prompt Orchestration in ComfyUI.
What’s New with ComfyUI-Variationator
The original README documents these releases:
| Version | Documented changes | Why they matter to you |
|---|---|---|
| 1.0.0 | Initial release with 20 lighting modifiers, nine basic nodes, and style transformation. | Established a way to explore lighting and style changes from a base prompt. |
| 2.0.0 | A redesigned system described in the version history as having 11 dimensions, 220 modifiers, and 18 nodes. It added Seasonal, Location, Weather, Mood, Camera, Color Grading, Action, Era, Detail, and Composition dimensions; multi-style layering; Smart Variation; a seed-controlled Random Modifier; and expanded documentation. | Broadened exploration beyond lighting, with more ways to make precise choices, discover unexpected combinations, and compare prompt directions. |
These are the versions described in the supplied README; it does not document later releases. As noted above, its headline totals conflict with its detailed version history.
Troubleshooting ComfyUI-Variationator
My prompts are getting too long or unfocused
- Return to your base prompt and identify the visual decision you want to test.
- Start with one modifier, then add another only if it contributes something distinct.
- Remove directions that compete or repeat one another.
- Where the node provides it, consider
preserve_base=Falseif retaining all the incoming text is contributing to the length. Check the resulting prompt so you do not accidentally lose your subject.
My variations look too similar
- Change a visually significant dimension, such as Lighting, Weather, Era, or Style.
- For relevant categories, compare DRY and WET choices.
- Try modifiers from different dimensions rather than several near-identical choices from one.
- Read the resulting prompts to confirm the directions actually differ.
I cannot find the modifier I want
- Use Modifier Info to inspect available choices.
- Consult the Modifier Reference for documented IDs and descriptions.
- If you know the look you want but not its ID, try a direct instruction with Smart Variation.
Random Modifier keeps choosing the same result
- Change the seed to request a different repeatable selection.
- Check whether its category is limited to
dry,wet, or another subset rather thanall. - Keep a seed you like so you can return to that selection later.
Smart Variation does not respond as expected
- Use concrete visual terms, such as “golden hour,” “rain,” or “cyberpunk.”
- Simplify the instruction and test one idea at a time.
- If you need an exact modifier, choose its ID with a single-modifier node instead.
Frequently Asked Questions
Does Variationator generate images by itself?
No. It prepares varied prompt text for use in a compatible image-generation workflow.
Do I need to use every modifier dimension?
No. Use only the dimensions that serve your image. A lighting change alone may be enough.
Can I repeat a variation I discovered at random?
The Random Modifier is described as seed-controlled. Record the seed and relevant category settings for a repeatable selection.
What is the difference between a batch and a layered prompt?
A batch gives you multiple prompt variations to compare. Layering combines several directions into one prompt.
Are the styles separate models?
The README describes them as style transformations within the extension, not as separate downloadable image-generation models. It does not list any distinct Variationator models.
Learn More about ComfyUI-Variationator
- Quick Start Guide: A short introduction to building your first variation.
- Node Reference: Detailed guidance when you need to understand a particular node or setting.
- Modifier Reference: Descriptions to help you choose modifiers and their IDs.
- Variationator on GitHub: The project repository and its documentation.
- GitHub Issues: Existing reports and a place to describe a reproducible problem or request a feature.
- ComfyUI on GitHub: Background on the application in which Variationator runs.
