NS WaveSpeed Qwen Edit Plus

The "Plus" on Qwen Edit Plus is doing two jobs: it takes up to three reference images instead of one, and it lets you set an exact output size. That combination makes it the Qwen edit node you reach for when the edit needs to match something - a face, a style, a product - rather than just follow a sentence.

It's part of the Symbiotica pack's neuralsins/WaveSpeed family. Where the plain NS WaveSpeed Qwen Edit node is a one-image, one-prompt bare edit, this one is the multi-reference workhorse. Qwen-Image-Edit's strength is instruction-following with a vision-language backbone (the KB's inpainting doc covers how it took the editing job), and feeding it references plays directly to that strength.

How it works

prompt in, up to three image_url_1–image_url_3 reference images, and the edited result comes back through output_image (IMAGE). It POSTs to WaveSpeed's /api/v3/wavespeed-ai/qwen-image/edit-plus.

The inputs that matter:

• prompt - the edit instructions. Chinese and English both work.
• image_url_1 - first reference, required in practice. image_url_2 / image_url_3 are optional, max 3 total.
• size - a 7-choice preset menu, default 1328x1328 (1:1), with 16:9, 9:16, 4:3, 3:4, 3:2, and 2:3.
• custom_size - the escape hatch: type 1920*1080 and it overrides the size dropdown. This is the input the basic Qwen nodes don't have, and it's why you'd pick this node for a nonstandard deliverable.
• seed - -1 for random, otherwise reproducible.
• output_format - jpeg (default), png, or webp.
• enable_base64_output - returns the image as a base64 string instead of a URL.

enable_sync_mode defaults on.

Installing

cd ComfyUI/custom_nodes
git clone https://github.com/symbiotica-ai/comfyui-nodes.git symbiotica
pip install -r symbiotica/requirements.txt

Restart or use ComfyUI Manager (search Symbiotica), then feed an NS WaveSpeed Client node carrying your WAVESPEED_API_KEY into client.

Common issues

• custom_size beats the dropdown, always. If both are set, the tooltip is explicit: custom wins. Clear it when you want the presets back.
• "max 3 total" is a hard ceiling. There's no slot four. Need more references? That's the Nano Banana Pro edit's six-slot territory instead.
• References are URLs. The standing family rule - hosted, reachable URLs, not local tensors.
• Edit drift compounds with more references. Every extra image is another thing the model can blend. If the output starts mixing features, drop to one reference and prompt more precisely.

The honest take: this is the Qwen node that actually gets used in production - multi-reference identity work plus exact sizing covers most real edit briefs. The LoRA version of it just adds server-side style steering on top.