zhenzhen-image-g2-lowprice

Comfly_zhenzhen_image_g2_lowprice is one of those "why is this node named that way" moments. The G-2 in the name is the model family, and the README's launch note spells out what you're getting: text-to-image (zhenzhen-image-g2-t2i) and image editing (zhenzhen-image-g2-i2i) in one node, with the resolution fixed at 1K and a pick-your-ratio dropdown. It was released as a deliberately stable, separate node so it wouldn't disturb the older seedream node, which tells you the author treats it as the dependable workhorse of the image family.

It runs on Zhenzhen's domestic storefront with the whole low-price family's plumbing. Cloud inference, cheap credits, no GPU needed.

How it works

Choose a model, write a prompt, pick a ratio, and for editing connect reference images. The node submits the job, polls, and returns image, image_url, task_id, and response. Straightforward.

The inputs that matter:

• model - zhenzhen-image-g2-t2i for generation, zhenzhen-image-g2-i2i for editing. This is the big switch.
• prompt - for t2i, describe the image. For i2i with references, describe the change.
• ratio - adaptive plus 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / 21:9. Adaptive follows your reference images when editing.
• resolution - locked to 1k. That's the trade for the low price: you don't get 2K/4K here. If you need big, the pack has other image nodes for that.
• image1–image10 - reference images for editing. Ten is a lot of room; most edits want one or two.

Installing and setting up

Install through ComfyUI Manager (search Comfyui-zhenzhen) or:

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/Comfyui-zhenzhen

Restart ComfyUI. No model files. Connect the Comfly Seedance2 Low Price Settings node's api_config, or set SEEDANCE_API_KEY. One settings node serves the entire low-price family.

Troubleshooting

The family tripwire, again: no api_config connection and you get "empty api_key." Beyond that, the G-2 node's specific quirk is the fixed 1K - people paste in a 2K reference, get a 1K output, and think it broke. It didn't; that's the tier. And editing works best when you actually change something in the prompt; "same but better" is the weakest possible edit instruction for any of these models. skip_error is available for batch safety.

Honestly, this is the node I'd point a beginner at in this pack's image family: one model switch, one ratio dropdown, fixed resolution so the price is predictable. Start here, and only climb to the bigger nodes once you know exactly what 1K is costing you.