NS WaveSpeed Seedream V4 Sequential

The "Sequential" in this node's name is doing real work. NSWaveSpeedSeedreamV4Sequential from symbiotica-ai/comfyui-nodes generates a whole series of images - up to 15 of them - from a single prompt, with no reference images at all. It's ByteDance Seedream V4's answer to the awkward workflow where you'd otherwise queue one text-to-image node fifteen times and pray the character stayed consistent across runs. Here the series is the unit of work: one call, one batch of images that are supposed to read as one set.

How it works

This is the text-to-image sibling of the Edit and Edit Sequential nodes, same hosted engine on WaveSpeed's API. The twist is what the API does with your prompt. Seedream V4 has a mode where a prompt like "the same woman in four outfits, studio lighting" returns four coordinated images rather than four rolls of the dice. The node's max_images field (1–15) tells the API how many frames the series should have, and the prompt's job is to specify the count and the content - "generate a group of three images showing…" - because the number you set and the number you describe have to line up.

The tooltip is blunt about this: "Must clearly specify the number of images in the prompt." Say "a series of six" in text, set max_images to 6, and the model holds the subject consistent across the set; the consistency is the whole reason you'd use this over a plain batch. If the prompt and the field disagree, you get a shrug in image form.

The inputs that matter

• prompt - the entire job: what to draw, how many, and how the series should relate. This is where you earn your keep.
• max_images - the count (1–15), which must match the number your prompt names.
• size_preset - nine Seedream V4 presets, 1:1, cinematic ratios, up to 4096x4096. Series at 4K is a lot of pixels.
• seed - 0 = random on this API; a non-zero value reproduces the same series.
• client - the WAVESPEED_AI_API_CLIENT socket from the NS WaveSpeed Client node.

Output is output_images (IMAGE, batched) - each series frame is a separate image in the batch, ready for a save loop or a contact sheet.

Install and keys

One install covers every node in the pack:

# ComfyUI Manager: search "Symbiotica", or:
cd ComfyUI/custom_nodes
git clone https://github.com/symbiotica-ai/comfyui-nodes.git symbiotica
pip install -r symbiotica/requirements.txt

Restart. Then add the NS WaveSpeed Client node, drop your wavespeed.ai key in its api_key widget (or set it once in ComfyUI Settings → Symbiotica / WAVESPEED_API_KEY env), and wire client into this node. No model downloads - requests and pillow are the only deps that matter here.

Common issues

Result count ignores you. Check that max_images and the number written in the prompt agree. This is the node's #1 gotcha and the API is strict about it.

"API_KEY is empty." The key lives on the client node, not here. Every WaveSpeed node in the pack needs that client connection.

Cost surprise. Fifteen images in one click is fifteen billable images. Series generation is the node you reach for when you need a consistent set - character sheets, storyboards, "the same product in every color" - and skip when one good image would do. It's also cloud-only: your prompt leaves the machine and ByteDance's moderation applies to every frame (external-api-nodes.md). For one-off variety, the plain single-image node is the cheaper reflex.