WorkflowListedExternal

Reference Fill (flux-fill-dev + deepseek-Janus-Pro) — v1.0

If you can't run, try in Nordy

Sourcecivitai.com/models/1297490

TaskMultiCapabilitiesImage edit · Image to imageFamilyFLUXEstimated VRAMNot yet characterizedBgrade

Runtime profile

Access: External workflow
Status: Listed
Source: Civitai
Updated: Oct 2, 2026

Source description

If you can't run, try in Nordy

🚀 Ultra-Precise Reference-Based Image Generation !

Struggling to get the perfect results with IP-Adapter or fine-tuned LoRA? Try this workflow!

Now, you can easily generate images with an astonishingly similar feel to your reference —without any additional training or IP-Adapter.

✅ Effortlessly create images that match your reference with extreme accuracy

✅ High-quality results without LoRA or extra fine-tuning

✅ Ideal for restoring and extending reference image details

AI-generated commentary

AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.

Purpose

Reference Fill (flux-fill-dev + deepseek-Janus-Pro) is for generating and editing images from a reference image and text prompt, including restoring or extending reference details; it accepts a reference image and text prompt and sends the generated image to preview or saving.

Flow

The named stages include image loading and resizing, mask conversion and cutting, inpainting or outpainting preparation, Flux guidance, sampling, and decoding.

Inputs and outputs

Use a reference image with LoadImage; masked edits use MaskToImage and Cut By Mask.
Text prompts are handled by CLIPTextEncode, and the resulting image can be sent to PreviewImage or SaveImage.

Setup

Published by Nouvo Ai, this workflow uses Flux.1 D as its base model.

Dependencies

The workflow names these model files: ae.safetensors, clip_l.safetensors, flux1-fill-dev.safetensors, and t5xxl_fp8_e4m3fn.safetensors. It also names deepseek-ai/Janus-Pro-1B.
Image and mask stages named include LoadImage, ImageResize+, MaskToImage, Cut By Mask, ImagePadForOutpaint, ImageRemoveAlpha+, and InpaintModelConditioning.
Other named stages include CLIPTextEncode, DualCLIPLoader, FluxGuidance, DifferentialDiffusion, JanusImageUnderstanding, JanusModelLoader, KSampler, UNETLoader, VAEDecode, VAELoader, PreviewImage, and SaveImage.

Configuration

The published description presents reference-based generation without additional training, IP-Adapter, LoRA, or extra fine-tuning.
For restoration or extension, consider ImagePadForOutpaint, InpaintModelConditioning, DifferentialDiffusion, and ImageResize+.
Suggestion · not verified

Cautions

Before running, verify that your setup includes deepseek-ai/Janus-Pro-1B, ae.safetensors, CLIPTextEncode, and SaveImage.
Suggestion · not verified
No license is stated for this workflow.

Does this need to be edited?

Sources

1 source

CivitaiPrimary

civitai.com/models/1297490

Verified Jul 20

Source excerpts

2 excerpts

NoteJul 20civitai.com/models/1297490

Source context: 310 downloads · Type Workflows · Base model Flux.1 D

NoteJul 16civitai.com/models/1297490