Workflows/Create transparent images locally with Qwen-Image 2.1

Images / WORKFLOW

Create transparent images locally with Qwen-Image 2.1

Create a source-linked RGBA image with the official ComfyUI Qwen-Image 2.1 template, then verify the PNG really contains transparency.

EstimatedSource checked 9/29/2026

LOCALRENTED GPU

Freshness 81 out of 100. Repository age: 55 days old, Sep 14, 2026. Measures age, not quality.

Popularity 76 out of 100. 248,964 downloads in 30 days; 3,575 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.

What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

A new release starts at 100. The score halves every 180 days: 100 × 2^(−age in days / 180). When the release date is unavailable, a repository date is a fallback—not proof of when the model was released. Missing dates stay unscored.

Check the date source

Popularity · adoption

Downloads over the past 30 days carry 80% of the score; lifetime likes carry 20%. Both use a logarithmic scale, capped at 10 million downloads and 100,000 likes. If likes are unavailable, downloads carry all the weight. Missing download statistics stay unscored.

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source

Scores as of Nov 8, 2026. Statistics checked Nov 7, 2026. Statistics are marked for refresh after 72 hours.Full scoring method

Before you begin

Hardware

Estimated memory starting points for the complete Qwen-Image 2.1 text-to-image pipeline at 1024×1024. These are not measured render times.

Machine	Qwen-Image 2.1 · ComfyUI text-to-image
NVIDIA GPU · 24 GB	Possible with ComfyUI offload and at least 32 GB system RAM; expect slower rendering.
NVIDIA GPU · 32 GB	Estimated INT8 starting point with better headroom.
AMD Radeon · 24 GB	No verified INT8 ConvRot run path for this exact graph; choose a supported NVIDIA GPU or rental instead.
Intel Arc · 24 GB	No verified INT8 ConvRot run path for this exact graph; choose a supported NVIDIA GPU or rental instead.
Apple Silicon Mac · 64 GB unified	Estimated BF16 text-to-image starting point with limited headroom.
Apple Silicon Mac · 96 GB unified	Estimated BF16 text-to-image starting point with more headroom.
CPU-only · 64 GB RAM	Not recommended because the complete image pipeline and many sampling steps are impractically slow; use a GPU or rental.

The Mac text-to-image path decodes images; do not infer the separate Qwen-Image 2.1 image-edit path is healthy on Apple MPS.

IMAGE / QWEN + COMFYUISOURCE-LINKED SETUP

Make a transparent PNG on your own machine

Use ComfyUI’s official Qwen-Image 2.1 text-to-image graph and the publisher RGBA prompt; choose the Windows, Mac or Linux instructions for your hardware, then verify actual alpha transparency.

Choose a package Load the visual workflow Check the result

BEFORE YOU BEGIN

The documented package

This setup uses the package documented for this task. Review its source and supported platforms before starting.

Save your machine in My Hardware to get an automatic starting choice. You can always choose any package yourself.

CURRENT SETUP

Qwen-Image 2.1 · ComfyUI text-to-image · official

Start with 1024×1024 and 25 steps. The NVIDIA INT8 files total about 27 GB and can offload on 24 GB GPUs with enough RAM; 32 GB VRAM is a safer estimate. Mac BF16 files total about 32 GB before runtime, so start at 64 GB unified memory. Longer/higher-resolution output needs more memory.

Freshness 81 out of 100. Repository age: 54 days old, Sep 15, 2026. Measures age, not quality.

Popularity 92 out of 100. 10,603,096 downloads in 30 days; 1,288 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.

What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

Check the date source

Popularity · adoption

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source

Scores as of Nov 8, 2026. Statistics checked Nov 7, 2026. Statistics are marked for refresh after 72 hours.Full scoring method

View weight source

THE WORKFLOW

Set up Qwen-Image 2.1 · ComfyUI text-to-image on your machine

Choose your computer first, then its GPU path. Follow the visual setup and import the linked workflow.

CHOOSE YOUR SETUPWindows

Use the ComfyUI Windows NVIDIA app and its GPU backend.

Install and open ComfyUI

Install the current ComfyUI Windows NVIDIA app from the official download page. Open the canvas and check the selected GPU is detected. This is the GUI route; a Python environment is not needed.

Install ComfyUI

Put the four exact model files in place

Download every linked file into its named folder under ComfyUI/models. The diffusion model alone cannot generate an image. Restart ComfyUI after the downloads.

Put these files in ComfyUI/models/4 required files

diffusion_models/

qwen_image_2.1_int8_convrot.safetensorsImage diffusion modelGet file

text_encoders/

qwen3vl_8b_int8_convrot.safetensorsPrompt encoderGet file qwen3.5_9b_qwen_image_2.1_pe_t2i.int8_convrot.safetensorsText-to-image prompt encoderGet file

vae/

qwen_image_2.1_vae_bf16.safetensorsPNG image decoderGet file

Open the official Qwen image template

In ComfyUI choose Templates → Image → Qwen-Image 2.1 Text to Image or download the exact official JSON and drag it into the canvas. Select the four downloaded model files in their loader nodes; update ComfyUI if a node is missing.

Official Qwen-Image 2.1 text-to-image JSON

Generate an asset with actual transparency

Replace the prompt with “This is an RGBA format image with transparency. A clean red rocket icon, centered, no lettering. The image has an alpha channel and a transparent background.” Select 1024×1024, 25 steps and CFG 1 in the template, then click Queue Prompt once. These are first-run settings, not quality benchmarks.

Save the PNG and inspect its alpha

Save or export the generated PNG from the ComfyUI output node. Open it in an editor with a checkerboard transparency view, then place it over both a dark and a light background. A white-looking background alone does not prove a real alpha channel.

SUCCESS CHECK · Qwen-Image 2.1 · ComfyUI text-to-image

An actual transparent PNG

The image saves from ComfyUI as PNG and reveals a transparent alpha channel when placed over a checkerboard, dark background and light background.

This guide uses the official ComfyUI text-to-image graph and official package files. The hardware tiers are file-size-based estimates; no YouRunAI speed benchmark is claimed. For unsupported GPU kernels use a rented NVIDIA GPU.

Back to setup steps

When it doesn’t go to plan

The ComfyUI loader is red or cannot find a model.

Update ComfyUI, verify every linked filename and folder in the file tree, then restart the app. Do not substitute a Qwen-Image 2.0 checkpoint.

The image appears on white instead of true transparency.

Keep the publisher RGBA/alpha prompt wording, save a PNG, then inspect its alpha against checkerboard, dark and light backgrounds.

The GPU runs out of memory.

Start at 1024×1024, close other GPU tasks and use ComfyUI model offload or a machine with more VRAM or unified memory. Weight file size is not peak pipeline memory.

REFERENCE LIBRARY

Sources and files

Original instructions, model files, and compatibility notes behind this setup.

5 SOURCES

01Qwen publisher RGBA prompt and Apple MPS notehuggingface.co 02Official Qwen-Image 2.1 ComfyUI fileshuggingface.co 03Official text-to-image template JSONraw.githubusercontent.com 04ComfyUI installercomfy.org 05ComfyUI Apple image-edit issue, not this text-to-image pathgithub.com

The model behind this workflow

Qwen-Image 2.1

Keep exploring

Text

SETUP GUIDE

Qwen3.5 · 9B private chat setup

Source-backed package and platform-route plan for private local chat with Qwen3.5 · 9B. Preserves the official Ollama 9B package, a smaller 4B family fallback, and a clearly labeled community uncensored GGUF option.

Freshness 38 out of 100. Release age: 254 days old, Feb 27, 2026. Measures age, not quality.

Build a private coding assistant with Qwen3.8 27B

Run Qwen3.8 27B with Ollama, connect it to Cline in VS Code, and make one reviewable code change in a tiny local project.

Freshness 69 out of 100. Repository age: 95 days old, Aug 5, 2026. Measures age, not quality.

Arena Open Source Web coding: rank 12 of 62. Ranked model: Qwen3.8-27B. Original model ranking, not a benchmark of this package. Export dated Nov 6, 2026. Source checked Nov 7, 2026. Human preference ranking, not hardware suitability.+3

SOFTWAREOllama + Visual Studio Code + Cline + Python

Create images locally with FLUX.2 Klein

Generate images from text prompts in ComfyUI with the fast distilled or more adjustable base 4B model. Save your results and reuse the visual workflow.

Freshness 32 out of 100. Release age: 297 days old, Jan 15, 2026. Measures age, not quality.

SOFTWAREComfyUI

LOCALRENTED GPU

Verified source