Symbiotica Wavespeed Nano Banana 2 Edit
The edit node the Symbiotica template pipeline hands its prompts to
- images
- image
- task_id
- result_url
The text-to-image sibling of this node gets you a new Nano Banana 2 image from a prompt. This one edits ones you already have. Give it one or more images plus a prompt, and Google's Gemini 3.1 Flash image model rewrites them to match - which is exactly what the Symbiotica template pipeline needs, because that pipeline produces template sheets and region prompts and then asks a model to fill the regions in. Symbiotica Template Prompt builds the numbered per-region edit prompt; this is the node that executes it.
Like its siblings, it's a Wavespeed API wrapper - submit, poll, return. Endpoint: google/nano-banana-2/edit. Closed model, per-call billing, prompt and images leave your machine. Same caveats, same appeal.
The inputs that matter
images- the IMAGE(s) to edit. Up to 14, which is Nano Banana 2's ceiling; a bigger batch is truncated with a console note rather than rejected. Wire a single sheet in, or a batch for multi-reference edits.prompt- what to change. Non-empty, or it raises.aspect_ratio-autoor the same 15 ratios as the text-to-image node.resolution-0.5k,1k(default),2k,4k.enable_web_search/enable_image_search- grounding toggles, and again mutually exclusive: both on is an immediate error.output_format-png(default) orjpeg.api_key- optional; falls back to Settings → Symbiotica orWAVESPEED_API_KEY.timeout- poll timeout, 300s default.
Outputs
image- the edited result as an IMAGE tensor.task_id- the Wavespeed job id.result_url- hosted URL of the full-size result.
Install
Part of the Symbiotica pack:
cd ComfyUI/custom_nodes
git clone https://github.com/symbiotica-ai/comfyui-nodes.git symbiotica
pip install -r symbiotica/requirements.txt
Restart ComfyUI, or ComfyUI Manager → search Symbiotica → install. Then a Wavespeed key in Settings → Symbiotica (or WAVESPEED_API_KEY). No models download for this node.
Common issues
Mostly the family complaints: the both-search-toggles error, the timeout on big batches or 4K, and a task-failed error whose real cause is in the included response body. Two edit-specific gotchas worth knowing:
- Truncation at 14 images - it happens silently (well, with a console note). If you fed a 20-frame batch expecting all of it, you got the first 14. For template work, feed the sheets, not the raw folders.
- The edit prompt is the whole job - Nano Banana 2 is a Gemini image model, so it's a strong instruction-follower but also applies Google's content filter. If a template prompt gets refused, that's the vendor's filter, not the node; there's nothing to patch locally.
If you're running the template pipeline end to end: Template Editor → Template Prompt → this node, and you've got a fill-the-regions edit loop. The Fast edit variant below is the same thing with fewer knobs and a shorter default timeout.
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| images | IMAGE | — | |
| prompt | STRING | — | |
| aspect_ratio | COMBO | auto | 15 options: auto, 1:1, 3:2, 2:3, 3:4, 4:3, +9 |
| resolution | COMBO | 1k | 4 options: 0.5k, 1k, 2k, 4k |
| enable_web_search | BOOLEAN | false | — |
| enable_image_search | BOOLEAN | false | — |
| output_format | COMBO | png | 2 options: png, jpeg |
| api_keyopt | STRING | — | |
| timeoutopt | INT | 30030–1800 | — |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |
| task_id | STRING | — |
| result_url | STRING | — |