SF WaveSpeed Nano Banana Pro Edit
Nano Banana Pro Edit — the community favorite edit model, in your graph
- client
- output_image
Nano Banana Pro is the flagship of the Nano Banana line - Gemini 3 Pro Image - and for editing it's the model people actually get excited about: 4K-native output, a "thinking" pass before it generates, and the instruction-following that made the base Nano Banana famous, but better. SF WaveSpeed Nano Banana Pro Edit is that model as a node. Describe the edit, feed it up to six reference images, and it returns a IMAGE tensor. If the base Nano Banana Edit node is your quick tool, this is the one you reach for when the edit actually matters.
How it works
The node collects your prompt and up to six image URLs (image_1 is required, the rest optional), bundles them into an images array, and posts to WaveSpeed's google/nano-banana-pro/edit endpoint. Returning a real tensor means it slots into ComfyUI like any local generation - preview it, save it, chain it into a video node. The pricing is right in the tooltip: 1k and 2k both run $0.14, 4k costs $0.24. For a closed flagship model, that's cheap enough that iterating doesn't hurt, which is the whole appeal of renting it per-call.
The inputs that matter
- client - from SF WaveSpeed Client, wired first.
- prompt - the edit instruction. The tooltip's own example ("Replace the cloudy sky with a clear sunset") is the right shape: concrete, visual, specific.
- image_1 … image_6 - public URLs.
image_1is the primary edit target; the rest are references for identity, style, or additional subjects. Use them when you need the model to keep a person consistent across the edit. - aspect_ratio - default 1:1, pick what fits the composition.
- resolution - 1k / 2k / 4k. Start at 1k for iteration, spend the $0.24 only on finals.
- output_format - PNG (default) for transparency support.
- enable_sync_mode - default true.
Output: output_image (IMAGE).
Installing it
In the SF ComfyUI Nodes pack from Stillfront. ComfyUI Manager → "SF ComfyUI Nodes" → Install → restart, or:
cd ComfyUI/custom_nodes
git clone https://github.com/Stillfront/comfyui-sf-nodes.git
cd comfyui-sf-nodes
pip install -r requirements.txt
Plus a WaveSpeed key via SF WaveSpeed Client.
Common issues
Same public-URL rule as every WaveSpeed image node - image_1 and any references must be fetchable by WaveSpeed or the call dies up front. Then manage expectations on the model itself: Nano Banana Pro is powerful but has Google's safety policy welded into its weights, so certain prompts get refused or come back bland no matter how clever you are; that's the model, not the node. One practical note on references: more images isn't automatically better - throwing six references at a simple edit can confuse the result. Start with image_1 plus one reference and only escalate if the identity isn't holding.
Inputs (12)
| Name | Type | Default | Description |
|---|---|---|---|
| client | WAVESPEED_AI_API_CLIENT | — | |
| prompt | STRING | Text description of the edit to perform (e.g., 'Replace the cloudy sky with a clear sunset') | |
| image_1 | STRING | Primary input image URL (connect from Upload Image node) | |
| image_2 | STRING | Second input image URL (optional) | |
| image_3 | STRING | Third input image URL (optional) | |
| image_4 | STRING | Fourth input image URL (optional) | |
| image_5 | STRING | Fifth input image URL (optional) | |
| image_6 | STRING | Sixth input image URL (optional) | |
| aspect_ratio | COMBO | 1:1 | Aspect ratio of the output image |
| resolution | COMBO | 1k | Output resolution: 1k ($0.14), 2k ($0.14), 4k ($0.24) |
| output_format | COMBO | png | Output format - use PNG for transparency support |
| enable_sync_mode | BOOLEAN | true | Wait for generation to complete before returning |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| output_image | IMAGE | — |