comfyui_gr85 Introduction
comfyui_gr85 is a collection of 13 custom nodes for ComfyUI. It helps you choose image dimensions, add controlled variety to prompts, select text from several options, and generate random numbers or seeds.
You can use these nodes when you want to:
- Plan image dimensions around an aspect ratio or pixel budget.
- Try prompt variations without rewriting an entire prompt.
- Make random choices repeatable by reusing a seed.
- Pass a selected pose prompt into another part of a workflow.
The nodes are organized in ComfyUI under GR85/Resolution, GR85/Prompt, and GR85/Random, with smaller categories for specific tasks.
How comfyui_gr85 Works
Think of each node as a small tool that takes an input and hands its result to the next step in your workflow. A resolution node calculates width and height; a prompt node produces text; a random-number node produces a number. You can connect those outputs wherever your workflow needs them.
Several nodes use a seed, which is a number that guides a choice. Reusing the same seed and inputs lets a seeded node make the same choice again. Changing the seed gives you a way to explore another variation.
For image sizing, it helps to distinguish two ideas:
- Aspect ratio describes an image’s shape, such as wide, square, or tall.
- Pixel budget describes its approximate total size: width multiplied by height.
Some GR85 nodes calculate dimensions from those constraints. Others adjust dimensions to a specified tolerance, such as multiples of 16, so the returned sizes follow the increment you choose.
comfyui_gr85 Features
Resolution helpers
| Node | What it does | When to use it |
|---|---|---|
| GR85_ImageDimensionResizer | Calculates a new width and height from the original image dimensions and a selected target resolution, while taking the original pixel count and aspect ratio into account. | When you have existing dimensions and want the node to calculate a corresponding size for a target resolution. |
| GR85_ImageSizer | Calculates dimensions for a target aspect ratio. You can choose original, landscape, or portrait orientation and set a tolerance for the returned dimensions. | When you know the shape you want, such as a wide composition, and need usable width and height values. |
| GR85_ImageSizerAll | Calculates dimensions from a pixel budget and aspect-ratio components, taking orientation and a specified tolerance into account. | When you want to work within a pixel budget while controlling the image’s shape. |
| GR85_RandomRatio | Takes two width-and-height pairs, randomly chooses an aspect ratio between their ratios, and returns integer width-and-height components. | When you want to explore shapes within a defined range instead of using one fixed ratio. |
For example, you might set GR85_ImageSizer to landscape for a wide scene and choose a tolerance of 16 so its dimensions follow 16-pixel increments. With GR85_RandomRatio, two input pairs act as the ends of a range of possible shapes.
Prompt helpers
GR85_SeedBasedOutputSelector chooses one string from up to 10 inputs. It considers only inputs that are not None, then uses the seed to select among them. Connect several prompt alternatives when you want one of them to continue through the workflow. Reusing the same seed and the same available inputs repeats the selection.
GR85_SimpleWildcardPicker expands choices written in the form {a|b|c}. It also supports nested braces. For example, a prompt containing a {quiet|busy} street can become either “a quiet street” or “a busy street.” Its seeded random choice lets you revisit a variation.
GR85_FanServicePrompt provides a dropdown of fan-service pose prompts. After you select one, the node displays it in a read-only box, offers a Copy prompt button, and outputs the prompt as text for use elsewhere in your workflow.
The tag injector nodes fill named placeholders in a prompt template:
- TagInjectorSingle inserts one tag into a placeholder such as
__elements__. - TagInjectorDuo works with two tags, such as
elementsandstuff. - TagInjectorLarge supports a larger set of descriptive placeholders, including
__location__,__weather__, and__style__.
For example, a template containing __location__ in __weather__, __style__ can serve as a reusable prompt structure. You change the supplied tags to make a new variation without rebuilding the whole sentence.
Random and seed helpers
- GR85_NextSeed takes a seed and produces a new random seed in the full 64-bit range. Use it when a later step needs a fresh seed.
- GR85_RandomFloat produces a decimal number between
min_valueandmax_value. You can set its decimal precision and use a seed for repeatable results. - GR85_RandomInt produces a whole number between
min_valueandmax_value. It is also seeded for repeatable results.
Choose RandomInt when a setting needs a whole number, and RandomFloat when it needs a decimal value.
Troubleshooting comfyui_gr85
A prompt selector does not choose the option you expected
- Check which of the selector’s inputs contain text. Inputs set to
Noneare not included in the choice. - Check the seed value. The node uses the seed to choose an index among the available inputs.
- Keep both the seed and the available inputs unchanged when you want to repeat a selection.
A wildcard prompt does not vary
- Check that each choice uses braces and vertical bars, such as
{quiet|busy}. - Change the seed to request a different seeded choice.
- Remember that a new seed does not guarantee a different result every time: the picker can choose the same option again.
The calculated dimensions differ from the ratio you entered
- Check whether the node has a tolerance setting.
- Remember that adjusting dimensions to an increment, such as multiples of 16, can change the exact width-to-height ratio.
- Review the orientation setting, if the node provides one, to make sure it matches the shape you intended.
Frequently asked questions
Does comfyui_gr85 include image-generation models?
No models are described for this extension. The listed nodes are helpers for dimensions, prompts, random values, and seeds.
Can I repeat a random prompt choice?
Yes. For a seeded node, reuse the same seed and inputs to repeat its choice.
What is the difference between a wildcard picker and a tag injector?
A wildcard picker chooses from alternatives written directly in prompt text, such as {sunny|rainy}. A tag injector fills named placeholders in a reusable template, such as __weather__.
Can I use a fan-service pose prompt outside its node?
Yes. The node outputs the selected prompt as text and also provides a Copy prompt button.
Learn More about comfyui_gr85
The node categories in ComfyUI are a useful starting point for exploring the collection:
- GR85/Resolution for image dimensions and aspect ratios.
- GR85/Prompt/Selection, GR85/Prompt/Wildcards, GR85/Prompt/Collections, and GR85/Prompt/Tags for different ways to build or choose prompt text.
- GR85/Random/Seed and GR85/Random/Numbers for seeds and numeric values.
Start with one helper node connected to the part of your workflow that needs its output. Once you see what it returns, you can combine it with other GR85 nodes to build more varied, reusable workflows.
