Docs for making better picks, not just more images
Varimuse Documentation
This guide is organized around the pages you use in the app. The goal is simple: help a first-time user go from a vague idea to a curated set of images that actually solves the problem in front of them.
How the Process Works
Varimuse is not built around getting one perfect image from one prompt. It is built for exploring nearby options, learning what matters, saving the strong candidates, and refining them into a set you can actually use.
Start with one clear goal
Decide what you are trying to solve before you generate: a product visual, a mood board direction, a children's book scene, or a social post concept. Varimuse works best when each run answers one question.
Generate a spread, not a single perfect image
Your first run is for exploration. Aim to compare a small set of useful directions so you can learn what should change next.
Keep your winners and branch from them
When something is close, save it as a Pick and branch from that result instead of restarting from scratch. This is how you gradually curate a set that solves the real need.
Download anything you need to keep
Treat Varimuse as your working studio, not your only archive. Save copies of outputs you care about as you narrow down your set.
Docs by Page
If you are lost inside the app, match the page you are on to the section below.
Create a Generation
Describe what you want, choose how much variation you want, and start your first batch.
Configure
Choose budget, aspects, and advanced controls to decide what changes between outputs.
Preview & Go
Review the exact variations before spending credits, then disable weak directions and launch.
Generations and Branching
Review results, save the strong ones, and branch from the closest hit to refine it.
Picks and Collections
Curate the results that matter so you can compare finalists, organize concepts, and present a narrowed set.
Explore and Profile
Publish selected work, earn credits from likes, and track your credit usage and earnings.
A Novice-Friendly Curation Loop
If your real goal is to walk away with a useful image set, follow this loop instead of trying to solve everything in one run.
Write down the job the images must do before you start.
Run a small first batch to discover what direction feels promising.
Save every serious contender as a Pick immediately.
Branch from the closest winner and change only one or two things at a time.
Move finalists into a Collection so you can compare a clean shortlist.
Download the images you need to keep before moving on.
Create a Generation
Describe what you want, choose how much variation you want, and start your first batch.
Write a prompt that names the subject, the use case, and the feeling or constraint that matters.
Keep the first batch manageable so you can actually compare it.
Use simple mode first unless you already know you need custom controls.
Configure
Choose budget, aspects, and advanced controls to decide what changes between outputs.
Use budget as your default model shortcut if you are new.
Pick one or two aspects that are most likely to change the outcome.
Open advanced controls only when you know what you want to lock, spread, or randomize.
Preview & Go
Review the exact variations before spending credits, then disable weak directions and launch.
Look for duplicates, nonsense, or obvious misses before you spend.
Disable any preview that is off-target so credits go toward better options.
Launch only the set you would genuinely compare afterward.
Generations and Branching
Review results, save the strong ones, and branch from the closest hit to refine it.
Pick the result that is closest to useful, not the one that is merely unusual.
Branch when you want to preserve most of a result and only change a few things.
Use image-iteration models when continuity matters across branches.
Picks and Collections
Curate the results that matter so you can compare finalists, organize concepts, and present a narrowed set.
Save Picks whenever an image is a serious candidate.
Use Collections to group options by concept, audience, or campaign direction.
Keep only images that help the decision in front of you.
Explore and Profile
Publish selected work, earn credits from likes, and track your credit usage and earnings.
Publish only work you are comfortable showing publicly.
Use your profile credits page to understand usage, earnings, and purchases.
Keep local copies of any published work you need to preserve.
How to Pick Aspects That Matter
This is where many users get stuck. The easiest rule is: only vary the things that are likely to change your decision. If changing an aspect would not help you choose a better image, leave it fixed.
Pick the variables, not the details
An aspect is a knob you want Varimuse to turn between outputs. Choose mood, lighting, style, composition, or color only if changing that thing could produce a meaningfully better option.
Start with one or two aspects
If everything changes at once, you will not learn why one image works better than another. Two aspects is enough to create a useful spread for most first passes.
Use aspects to answer a question
Examples: “Should this feel playful or premium?” “Do I need bright daylight or cinematic shadows?” “Is watercolor or editorial photography better for this audience?”
When to leave an aspect alone
If a detail is essential to the assignment, keep it fixed. Use variation only where you want options, not where you need consistency.
Quick examples
If you are choosing between marketing directions, vary mood and lighting. If you are choosing illustration treatment, vary style and color palette. If you are already happy with the palette, do not vary it just because it is available.
How to Choose Models in Advanced Mode
Advanced model controls are useful, but only after you know what you are testing. If you are new, the safest choice is to let the budget tier choose for you first.
Use budget first, advanced second
If you are unsure which model to pick, stay in the simple budget mode. It is the fastest route to a sensible starting point.
Pick one model when judging composition
If you are still deciding on framing, subject clarity, or broad art direction, a single model removes one source of noise from the comparison.
Use multiple models only for a reason
Advanced multi-model runs are useful when you already like the prompt and want to compare rendering behavior, realism, texture, or adherence across engines.
Understand the four advanced modes
Single keeps one model for everything. Round Robin cycles evenly through active models. Random mixes models unpredictably. Each Model multiplies the run by running every planned variation through every selected model.
For branching, continuity matters
If you are refining an existing image, favor models marked for image iteration so the next batch can stay visually connected to the current result.
A simple rule of thumb
If you are exploring ideas, use one sensible model. If you are comparing rendering behavior, use two or more. If you choose Each Model, remember that the total image count and cost multiply quickly.
Preserving Important Work
As you curate, download outputs you need to keep. Collections and Picks make comparison easier inside Varimuse, but your long-term archive should live somewhere you control.
Need developer details instead?