feat(z-image): add Seed Variance Enhancer node and Linear UI integration#8753
Merged
Pfannkuchensack merged 10 commits intoinvoke-ai:mainfrom Jan 12, 2026
Merged
Conversation
Add a new conditioning node for Z-Image models that injects seed-based noise into text embeddings to increase visual variation between seeds. Backend: - New invocation: z_image_seed_variance_enhancer.py - Parameters: strength (0-2), randomize_percent (1-100%), seed Frontend: - State management in paramsSlice with selectors and reducers - UI components in SeedVariance/ folder with toggle and sliders - Integration in GenerationSettingsAccordion (Advanced Options) - Graph builder integration in buildZImageGraph.ts - Metadata recall handlers for remix functionality - Translations and tooltip descriptions Based on: github.com/Pfannkuchensack/invokeai-z-image-seed-variance-enhancer
Pfannkuchensack
requested review from
JPPhoto,
blessedcoolant,
dunkeroni and
lstein
as code owners
github-actions
bot
added
python
JPPhoto
reviewed
Jan 10, 2026
JPPhoto
approved these changes
Jan 10, 2026
Collaborator
JPPhoto
left a comment
JPPhoto
left a comment
There was a problem hiding this comment.
Python code looks good. Need to get reviews from frontend prior to merging.
lstein
approved these changes
Jan 11, 2026
Collaborator
lstein
left a comment
lstein
left a comment
There was a problem hiding this comment.
I've been functionally testing this for the past few days and it seems to do the job. My only comment concerns the wording of the noise strength popup.
Updated description for seedVarianceStrength.
JPPhoto
requested changes
Jan 11, 2026
torch.rand() generates [0, 1), so the scaled range excludes 1.
…ub.com/Pfannkuchensack/InvokeAI into feature/z-image-seed-variance-enhancer
JPPhoto
approved these changes
Jan 12, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Add a new conditioning node for Z-Image models that injects seed-based noise into text embeddings to increase visual variation between seeds.
Z-Image-Turbo can produce relatively similar images with different seeds, making it harder to explore variations of a prompt. This feature implements reproducible, seed-based noise injection into text embeddings to increase visual variation while maintaining reproducibility.
Backend:
z_image_seed_variance_enhancer.pystrength(0-2),randomize_percent(1-100%),seedFrontend:
paramsSlicewith selectors and reducersSeedVariance/folder with toggle and slidersbuildZImageGraph.tsBased on: https://github.com/Pfannkuchensack/invokeai-z-image-seed-variance-enhancer
Related Issues / Discussions
Integrates community node as core feature for Z-Image models.
QA Instructions
Workflow Editor:
Merge Plan
No special merge requirements. Schema will be auto-generated on backend start.
Checklist
What's Newcopy (if doing a release after this PR)