Qwen Image 2.1 text-to-image: establish a comparable first run
Verify the reviewed template’s component selections, 25-step setting and 1MP baseline before moving to 2K. Keep Diffusers defaults separate.
Workflow source record
Open upstream reference ↗This listing points to an upstream reference. It does not mean a JSON workflow file is hosted in this package.
File and download boundary: ComfyUI World provides source references and metadata only. Model and LoRA weights, node packages, and workflow files stay with their original providers; this site does not store, mirror, or proxy them. Downloads and access conditions are controlled by each provider.
Workflow identity
This guide links image_qwen_image_2_1_t2i.json at workflow-template commit 7ce126c4ec77f44cc7a60f93eec52a197efb9ce5. ComfyUI v0.37.0 contains the core support. Later template revisions may require additional frontend features or nodes, so a core version number is not a guarantee for every future graph. Q03 Q05
Import and inspect the component set
Open a copy of the official graph in a backed-up instance. Inspect the subgraph’s UNETLoader, CLIPLoader and VAELoader selections and its TextEncodeQwenImage21 conditioning node. A UUID used for a subgraph is not automatically the name of a missing custom-node package. Resolve graph interpretation before trying to repair model selectors.
Preserve this snapshot’s starting settings
The reviewed named widgets show 25 steps, CFG 1, Euler, simple scheduling and a 1024×1024 baseline; ResolutionSelector starts near one megapixel. The official Diffusers example uses 40 steps. Those settings belong to different execution paths and must not be merged into one supposed universal preset. Q01 Q05
Change only the prompt first
An original diagnostic prompt is: A ceramic teapot on a plain wooden table, side lighting, realistic glaze, no text. Record the seed and graph settings before running. Leave prompt enhancement, LoRAs and additional references out of this initial comparison. Changing several components at once makes a failure difficult to attribute.
Increase the output size deliberately
A native square 2K output is 2048×2048, about four megapixels, not two. Checking dimensions against multiples of 32 is consistent with the reference-image block handling. Q01 Q04 Moving from one to four megapixels changes the workload; the smaller run cannot certify memory use at the larger size.
Completion checks
Confirm that execution finishes, reopen the original PNG and inspect its actual dimensions. Then evaluate prompt adherence, lettering and edge quality separately. A PNG extension does not certify transparency. A fixed seed is useful for controlled comparisons, but changing precision, software or hardware does not guarantee identical pixels.
Route failures by stage
For a missing dropdown value, check the actual instance’s model directory and exact filename. For a missing TextEncodeQwenImage21 class, check core and frontend support. For sampling failures, investigate the component set and memory stage. Save the first complete exception and graph before replacing the environment with a broad upgrade.
Sources and verification
Source and code review on 2026-09-22. No GPU run, weight download or quality benchmark. Procedures are editorial guidance, not a confirmed diagnosis of your environment.
- Q01 · Qwen-Image-2.1 official README
- Q03 · ComfyUI v0.37.0 release
- Q04 · ComfyUI 0.37.0 Qwen nodes
- Q05 · Official Qwen 2.1 text-to-image template
Related reading
Documented dependencies
This page does not inspect your machine or guarantee execution.
File names describe recorded requirements. A linked page may be a tutorial or repository, not a direct download; verify the exact file, revision, access terms and checksum at the publisher.
Was this page helpful?
Anonymous. Only a yes/no counter is stored; no account, IP address or device details.
Sources & references
Source and code review on 2026-09-22. No GPU run, weight download or quality benchmark. Procedures are editorial guidance, not a confirmed diagnosis of your environment.
01Qwen-Image-2.1 official READMESource checked: 2026-09-2202ComfyUI v0.37.0 releaseSource checked: 2026-09-2203ComfyUI 0.37.0 Qwen nodesSource checked: 2026-09-2204Official Qwen 2.1 text-to-image templateSource checked: 2026-09-22