Zhenzhen Google Veo3 Lite

Veo quality, smaller price tag. Comfly_Googel_Veo3_Lite is the budget sibling of the full Veo3 node: it wraps Google's veo3.1-lite model - that's the entire model dropdown, one option - for when you want Veo's video-and-audio magic on a draft or a throwaway clip without paying flagship rates.

It's a deliberately thin node, and that's the point. The relay priced the Lite model as the value tier; you trade some fidelity and resolution headroom for a meaningfully smaller bill per clip. If you're iterating on a scene and burning a lot of Veo calls, this is the one to put in the loop.

What's here

• prompt - the video prompt. enhance_prompt does the same Chinese-to-English auto-translation as the full Veo3 node; set it true if your prompt isn't clean English.
• model - veo3.1-lite, fixed.
• aspect_ratio - 16:9 or 9:16.
• image1–image3 - optional reference images for image-to-video.
• enable_upsample, seed, apikey, and the pack-standard skip_error.

Outputs: video (VIDEO tensor), video_url, and response (JSON).

Install and key

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

Restart, or ComfyUI Manager → "Comfyui-zhenzhen". Same shared setup: no model downloads, API key from ai.t8star.org into apikey. Token-group note from the README: Lite rides the veo&grok-备用1 backup group on the site - if your key isn't assigned that group, expect auth errors, and that's a website-side fix, not a node bug.

Bottom line

If you already run the full Veo3 node, this is a one-node addition that gives you a cheaper lane for drafts and tests. If you're new to Veo on this pack, note that "Lite" is still a closed, metered, server-side model - your prompt and reference images leave the machine, Google's moderation still applies, and upstream hiccups still show up as transient 500s that a re-run clears. For storyboard-speed iteration it's the sensible default; for a hero shot you'll want the flagship model.