Concurrent Submit | zhenzhen-wan-2.7-spicy-i2v-lowprice
Wan 2.7 'spicy' image-to-video, queued
- first_image
- api_config
- task
"Spicy" here is a content-policy term, not a flavor. The Wan 2.7 spicy image-to-video lane is the one the author built for 宽审核 - looser content review - and the README pitches it at cross-border e-commerce and anyone whose prompts trip the strict channels. It's Wan 2.7 i2v (reference image in, video out) through the low-price domestic channel, with a negative prompt and an optional audio track. Whether you need the looser moderation or not, it's a solid, cheap i2v node; just know the selling point when you're deciding between this and the other Wan 2.7 nodes.
The concurrent wrapper is standard: submit to the shared video pool (ten workers), get a task back, collect in Concurrent Collect Videos. Animate the same first frame with a dozen prompts in parallel - for a model with a strict filter, batching is how you find the prompts that get through.
Inputs that matter
- first_image - required. The first frame; the clip animates from this.
- prompt - the motion/action description.
- seconds - 2 through 15 seconds, default 2.
- resolution -
720p(default) or1080p. - negative_prompt - yes, Wan 2.7 spicy i2v supports one. Write it as normal English/Chinese describing what to avoid, not tag spam.
- audio_url - optional public URL to an audio track for the clip.
- prompt_extend - off by default; turn on for server-side prompt expansion.
- seed - defaults to -1 (randomize each run), unlike most of the pack.
- api_config, skip_error - low-price channel config via the zhenzhen-seedance2-low-price-api-settings node or
SEEDANCE_API_KEY.
Output: a COMFLY_VIDEO_FUTURE task.
How it works
The ComflyConcurrent.py wrapper validates your inputs, pushes the original Comfly_wan_2_7_spicy_i2v_lowprice call onto the bounded video executor (ten workers), and returns the future immediately. The collector waits on all tasks, keeps slot order, and reports per-slot status; failure_mode: placeholder turns a failed slot into a blank clip instead of aborting the batch.
Setup
Install the pack once - ComfyUI Manager (search "Comfyui-zhenzhen"), or:
cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/Comfyui-zhenzhen
No weights - Wan 2.7 has no open checkpoint. Key it through the low-price settings node or SEEDANCE_API_KEY (api.seedance.nz, the domestic discount channel), and every clip is a paid call.
Gotchas
Two honest warnings. The "looser review" claim only goes so far: the moderation is baked into the model vendor's pipeline, and the node can't promise anything the upstream refuses - treat 宽审核 as "less likely to bounce," not "will generate anything." And negative prompts on API video models are advisory, not guaranteed; the first frame does more of the steering than your negative text does. Keep the reference image clean and you'll spend less time fighting both. The seed defaults to randomize, so if a run looks like a rerun, that's the cache-control seed doing its job. Timeouts: check the async task page.
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| first_image | IMAGE | — | |
| prompt | STRING | — | |
| seconds | COMBO | 2 | 14 options: 2, 3, 4, 5, 6, 7, +8 |
| resolution | COMBO | 720p | 2 options: 720p, 1080p |
| negative_prompt | STRING | — | |
| audio_url | STRING | — | |
| prompt_extend | BOOLEAN | false | — |
| seed | INT | -1-1–2147483647 | — |
| api_configopt | ZHENZHEN_SEEDANCE2_CONFIG | — | |
| skip_erroropt | BOOLEAN | false | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| task | COMFLY_VIDEO_FUTURE | — |