Introducing batch generation
Generate up to 4 variations of the same sketch in one click — perfect for exploring styles before committing to a direction.
One of the most common requests we've heard since launch: "Can I generate multiple versions at once?" Today, the answer is yes. Batch generation is now available for all SketchSnap users.
What it does
Instead of generating a single render and deciding whether to keep it or try again, you can now generate 2 or 4 variations simultaneously. Each variation uses the same sketch and settings — but the model explores different interpretations within those constraints, giving you a spread of outputs to evaluate side by side.
The variations aren't random noise. They represent genuinely different aesthetic choices: different lighting directions, different color temperatures, different levels of detail in the background. Think of it as the model showing you its top answers rather than committing to just one.
How to use it
Open any canvas and look for the generation count selector next to the Generate button. Switch from 1 to 2 or 4, then generate as normal. Results appear in a grid. Click any result to expand it, download it, or continue editing from it.
You can also select a result as your base image — the starting point for a new round of generation or upscaling. This makes batch generation a natural first step in a workflow where you want to explore before you refine.
When to use it
Batch generation is most useful early in a project, when you're still deciding on direction. A client presentation is easier to put together when you have four distinct options rather than one. A commission is easier to discuss when the client can point to what they like and dislike across multiple examples.
It's also useful when you're experimenting with a new style preset for the first time. Seeing four variations reveals the range of the preset — which details it tends to emphasize, how it handles shadows, how loose or tight the rendering style is.
A note on generation credits
Each variation counts as one generation. Generating 4 variations uses 4 credits. This is the same cost as generating one at a time four times — but significantly faster and with more stylistic diversity in the outputs, since the model explores different directions rather than repeating a similar result.
What's next
We're working on a few features that build on batch generation: the ability to lock specific variations while regenerating others, and a comparison view that lets you place two results side by side at full size. Both are in progress and will ship in the coming weeks.
As always, let us know what you think. Tag us on X / Twitter or reach out directly — your feedback shapes what we build next.