NS WaveSpeed Nano Banana Pro Edit Ultra

Nano Banana Pro Edit Ultra is the "Ultra" rung of the Pro edit ladder, and the "Ultra" is doing real work: where the standard Pro edit tops out at 4K, this node offers 4k and 8k output. Editing an image at 8K native is a job no local setup is casually doing - and here it's a dropdown.

It's part of the Symbiotica pack's neuralsins/WaveSpeed family. If you squint, this is also a hosted upscaling play: feed it a modest source, prompt it to "enhance," and get a genuinely large, clean result back - the same reason the community already leans on Nano Banana Pro as an upscaler. The tooltip quotes the meter: 4k at $0.15, 8k at $0.18. Notably cheap relative to the Pro Edit's $0.24 for 4k - WaveSpeed prices the Ultra tier on its own scale.

How it works

Identical shape to the standard Pro edit: a prompt, an image_1 (required) plus image_2–image_6 (optional references), an aspect_ratio menu spanning 1:1 to 21:9, and output_format (png/jpeg). The one real difference is resolution:

• 4k - $0.15, the everyday big edit.
• 8k - $0.18, and genuinely enormous. That's a 7680px-ish long edge from an edit model.

It returns output_image (IMAGE), POSTing to WaveSpeed's Nano Banana Pro edit path just like the sibling nodes. 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

• 8K is a lot of pixels to move around. The result tensor is huge; downstream save/encode nodes will feel it, and the render takes real time. Use 8K when the deliverable demands it, not to feel fancy.
• "Ultra" doesn't mean unlimited references. Still six image slots, still only image_1 required.
• References are URLs. Standing family rule - hosted, reachable URLs, not local tensors.
• Watch the bill on re-rolls. An 8K re-queue isn't free, and the prompt-wording iteration can quietly add up.

The trap to avoid is treating 8K as the default because it's only three cents more. The quality ceiling is the model's, not the resolution's - an 8K render of a mid prompt is still a mid prompt, just bigger.