lllyasviel / lllyasviel/stable-diffusion-webui-forge

Feature Request: "Batch Process Prompt List" for Parallel Generation

Open
#3,020 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
13k
Forks
1.7k
PR merge metrics
No merged PRs in 30d

Description

**Description**
The Batch size parameter is misleading for my use case. It generates multiple images in parallel, but only from a single, identical prompt. I need a feature that can process a list of different prompts in a single, parallel batch job.

**The Core Issue Explained**
What Batch size: 4 does now: It takes ONE prompt (e.g., "a cat") and generates FOUR images of "a cat" all at once.

What I need it to do: I want to provide FOUR different prompts (e.g., "a cat", "a dog", "a car", "a house") and have the system generate ONE image for each of those prompts, all at once, in a single parallel operation.

The current system lacks a method to do this. The Prompts from file or textbox option processes the list sequentially, which is slow. There is no option for parallel execution of a prompt list.

**The Requested Feature**
I propose a new feature, which could be an option within the Prompts from file or textbox script, called something like "Process as parallel batch".

How it would work:

I enter my list of **four completely different prompts.**

I check the "Process as parallel batch" box.

I click Generate.

The system treats the list of prompts as a single batch job, generating all four images simultaneously on the GPU.

The output order is preserved: the first image in the result corresponds to the first prompt in the list.

This would combine the speed of Batch size with the control of the prompt list feature, eliminating the need to choose between performance and workflow.

Additional Enhancement: Integration with Prompt S/R Concept
It would be a powerful enhancement if this new parallel batch feature could also integrate the "search and replace" functionality from the XYZ plot.

How it could work:

A user could define a base prompt in the main field:
a photorealistic image of a {subject}, professional studio lighting

Imagine {subject} is a cat, since thats how s/r works right now.
Then, in the parallel list, they would provide only the values to replace {subject} with:

cat
dog
car
house

The system would then generate the four final prompts in parallel:

a photorealistic image of a cat, professional studio lighting
a photorealistic image of a dog, professional studio lighting
a photorealistic image of a car, professional studio lighting
a photorealistic image of a house, professional studio lighting

This would make the feature incredibly versatile, allowing users to efficiently generate batches of images that share a common style or structure but vary by a specific subject, attribute, or style keyword—all processed simultaneously for maximum speed.

This combines the best aspects of Prompt S/R (template efficiency) and parallel batch processing (GPU speed).

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reading the existing Prompts from file or textbox flow, Batch size handling, and XYZ plot Prompt S/R entry points. Determine how parallel prompt execution, output ordering, and template replacement should interact; done means a prompt list can run in parallel while preserving order and optionally applying the shared template.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.