🎨📝 Text Generator (Style)

The style feeder in the pack's Text Generator family, and the one node in that group aimed squarely at the look of the image rather than what's in it. Where character, outfit, and scene nodes describe subject matter, this one layers on medium, art movement, lighting, and mood - the kind of qualifiers people usually bolt onto the end of a prompt by hand, here broken out into structured categories you pick from instead of remembering the right phrase.

How it works

category sets the medium first - Painting, Photography, Digital Art, 3D Rendering, or Illustration - and style narrows it to a specific movement or technique within that medium, with over fifty options spanning Renaissance and Baroque through Impressionism, Cubism, and Futurism. From there, color_palette, lighting_type, mood, and ambiance each add their own layer - vibrant versus muted, dramatic versus soft lighting, peaceful versus tense mood, and so on - following the same NONE / RANDOM / pick-a-value pattern the rest of the family uses, with seed controlling which random picks land. CUSTOM_PROMPT sits on top for anything you want to add by hand. Like the scene node, this one connects directly into the main Text Generator node with no chaining input of its own - it's a self-contained style fragment, not something other generator nodes feed into.

The inputs and outputs that matter

• category - the medium: painting, photography, digital art, 3D rendering, or illustration. Worth setting explicitly since it changes what "style" even means underneath it.
• style - the specific movement or technique, the biggest list on this node at over 50 entries.
• color_palette, lighting_type, mood, ambiance - secondary layers, good candidates to leave on RANDOM while you're exploring, then pin down once you find a combination you like.
• CUSTOM_PROMPT (STRING, multiline) - freeform addition on top of everything the dropdowns generate.
• Output: GEN_STYLE - connects directly into the main Text Generator node.

How to install it

Via ComfyUI Manager: search Bjornulf_custom_nodes, install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/justUmen/Bjornulf_custom_nodes
cd Bjornulf_custom_nodes && pip install -r requirements.txt

then restart ComfyUI. No special dependency for this node - it's a dropdown/text builder - but the pack's shared requirements.txt still gets installed since it's all one repository.

Common issues & troubleshooting

Style words don't do much on their own. This is a general prompting reality, not specific to this node: naming an art movement or a mood shifts a model's output, but how strongly depends entirely on how well that concept was represented in the checkpoint's training data. A style name that works beautifully on one checkpoint can barely register on another - if a style pick doesn't seem to change anything, try a different checkpoint before assuming the node isn't working.

Output never shows up in your generated image. Same pattern as the rest of the Text Generator family - GEN_STYLE has to actually reach the pack's main Image Text Generator node to become part of your real prompt; on its own this node just builds a text fragment.

Combining a strong style with a strong custom prompt fights itself. If you've picked a specific style and also written a detailed CUSTOM_PROMPT describing a different look, the two can pull the generation in conflicting directions. Treat CUSTOM_PROMPT as an addition to the chosen style rather than a way to override it - for a genuinely different style, change the dropdown instead.