视频｜HappyHorse 1.1/1.0 图生视频 I2V

HappyHorse is a video-generation model you probably haven't heard of, and that's the point of this node - it's one of the more obscure models the Tikpan pack resells, aimed squarely at the "make my product shot / illustration / screenshot move" use case. This I2V node takes a static image as its first frame and generates 3–15 seconds of motion at 720p or 1080p. For e-commerce content - a product photo that suddenly turns and glows, an illustration that comes alive - that's a genuinely useful trick that's hard to do well with the big names.

You get both HappyHorse versions here (happyhorse-1.1-i2v and the older happyhorse-1.0-i2v, kept for old-workflow compatibility). The 1.1 default is what you want; 1.0 is there so existing graphs don't break. Same metered-cloud rules as everything else in this pack: sk- key from https://tikpan.com, pay per task, your image leaves your machine.

How it works

Feed it an image and a prompt describing the motion, and the node does the standard Tikpan video dance: submit the task, poll until it's done, download the clip, and return five outputs - 🎬_视频输出 (VIDEO), 📁_本地保存路径, 🔗_视频云端链接, 🆔_任务ID, and 📄_完整日志. Two things differ from the Grok video nodes: you can pick between 同步 (block until the video is ready and downloaded) and 异步 (submit and return immediately, collect the result later with the pack's async task node - the right call for long or batch jobs), and there's a 水印 toggle to keep the video clean or add Tikpan's watermark.

For the input image you have two routes. The 首帧图片 IMAGE port works, but the tooltip and docs push 图片URL - a public URL for an image that's already hosted. It skips the upload step entirely, and uploads are where local failures tend to happen.

The inputs that matter

• 模型 - 1.1 (new) vs 1.0 (legacy).
• 生成指令 - how the image should move. The default ("a cat running across the grass...") shows the level of description that works.
• 执行方式 - sync for a single test, async for batches.
• 清晰度 - 1080P is the default and the point of the node; 720P if you're iterating.
• 视频时长 - 3–15 seconds; longer is slower and pricier.
• 随机种子 - fixed default for reproducible runs.
• 最长等待秒数 - default 600; the tooltip says raise it for 1080P or long clips.

Install and first run

cd ComfyUI/custom_nodes
git clone https://github.com/htrert/ComfyUI-Tikpan-Pro

Restart, or ComfyUI Manager → "Tikpan". Paste your key, drop an image in, set a short duration and 720P for the smoke test. HappyHorse jobs are cloud tasks, so expect polling, not instant results - that's normal.

Common issues

The upload path is where this node bites people. Big images fail the upload more often than small ones - compress first, and prefer 图片URL when the image is already hosted. If a sync job times out, keep the 任务ID and use the pack's task-query node rather than resubmitting (double charge). And if 🎬_视频输出 won't preview but the 本地保存路径 exists, the render succeeded - your ComfyUI just lacks a video-preview extension. Standard 401/402/429 key-and-quota errors apply as always.