FSL Gemini Generate Image V8

FSLGeminiGenerateImageV8 is the previous version of the advanced Gemini image node. If you're loading an old FSL workflow you'll hit it; if you're starting fresh, use FSLGeminiGenerateImage instead. The README is blunt about it - V8 is flagged [Legacy], and the changelog notes the standalone maintained version is where fixes land. There's no feature here the current node doesn't do.

What it still has

The shape is identical to its replacement: prompt + model, width/height, temperature, seed, the three use_init_image / use_images / use_mask toggles, base_for_inpaint, optional init_image / images / mask / mask_image inputs, and the same nine outputs (image, prompt, model, width, height, temperature, seed, seed_mode, metadata). Same mask-feathering and mask-strength controls, same base-selection logic.

What differs is maintenance. The model dropdown still lists gemini-2.5-flash-image-preview and gemini-2.0-flash-exp - both deprecated by Google - and the current node's cleanup (removing dead models, the metadata-only width/height tooltips, the quality selector) never came back to V8. It also has a refresh_models toggle that the newer node dropped.

When to use it

Two cases. First, you inherited a workflow that references it - V8 will still run if you don't care about the current model lineup, but check the model dropdown against what Google actually offers today. Second, you're on an older pin of the pack where the current node isn't installed. That's about it.

Install & gotchas

Same install as everything else in this pack: ComfyUI Manager → "ComfyUI FSL Nodes", or:

cd ComfyUI/custom_nodes
git clone https://github.com/fredlef/Comfyui_FSL_Nodes
cd Comfyui_FSL_Nodes
pip install -r requirements.txt

Then restart. No model downloads; API key in the widget or the GEMINI_API_KEY env var.

The practical trap is the stale model list. If you pick gemini-2.5-flash-image-preview or gemini-2.0-flash-exp today, Google will return an error and the node raises a RuntimeError with the API message - that's not a bug in your setup, that's a model that no longer exists. Swap to a current model in the dropdown or migrate the workflow to the maintained node. Since the two share the same outputs, migrating is usually just a right-click → convert-and-delete, then re-picking the model.