Nodes/ERPK Collection/Bytedance Seedream V4.5 Sequential
ComfyUI Node

Bytedance Seedream V4.5 Sequential

One prompt, a batch of consistent images, 15 max

By eRepublik-Labs·Created 11 months ago·Updated 20 days ago· 1
Bytedance Seedream V4.5 Sequential
  • client
  • images
prompt
max_images4
size_presetCustom
seed-1
width2048
height2048
show_aspect_ratiotrue
enable_sync_modefalse
enable_base64_outputfalse

The normal version of Seedream V4.5 gives you one image. The Sequential version gives you 1 to 15 images from a single prompt, with cross-image consistency - same character, same style, same scene logic across the whole batch. That's a storyboard, a character sheet, a spread of UI variants, or a contact sheet of design directions, generated in one call and billed at $0.027 per image.

It's part of the ERPK Collection, the community pack wrapping the WaveSpeed AI API. Same typography strengths as the single V4.5 node, same cloud billing, no local compute.

How it works

You write one prompt and set max_images. Here's the implementation detail the tooltip warns about: the node automatically appends the image count to your prompt for API compliance. So "product shots of a sneaker" becomes "product shots of a sneaker, 4 images" behind the scenes - don't fight it, and don't write the count yourself.

The node sends the batch request to WaveSpeed, waits, and returns an images IMAGE tensor - a batch of frames, not a single image. Wire that into a batch-aware save or preview node and you get all N images out. max_images defaults to 4; the range is 1–15.

The inputs that matter

  • prompt - the shared concept for the whole batch. Cross-image consistency means the prompt is the contract: change the prompt, change the family.
  • max_images - 1–15, default 4. The count is auto-appended to the prompt, so set it once and let the node handle it.
  • size_preset - the resolution presets (up to 4K), or Custom for manual width/height (1024–4096, default 2048).
  • seed - cache control, randomizes by default. Fixed = reuse cached batch, randomize = fresh billed run.
  • enable_sync_mode / enable_base64_output - wait for completion inline / get base64 instead of URLs. Leave them alone unless you have a reason.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/eRepublik-Labs/comfyui-nodes-erpk.git erpk
cd erpk
pip install -r requirements.txt

Or ComfyUI Manager → search erpkERPK Custom Nodes. WaveSpeed key with credits via right-click canvas > ERPK Settings (Settings > ERPK > API Keys).

Common issues

  • Images that don't look like a consistent set - Sequential consistency is real but not absolute; faces still drift across a 15-image batch. Keep the prompt character-descriptive (hair, outfit, setting) to hold it together.
  • A weird count in your prompt - that's the auto-append at work. It's intentional, so don't "fix" it by deleting the number or double-appending your own.
  • Cost creep - $0.027 × N images adds up fast. Start at 4, not 15.
  • Batch save confusion - the images output is a batch tensor; a plain single-image Save node won't handle it. Use a batch-aware saver or split the batch.

The rule of thumb: need one hero image → the single Seedream V4.5 node. Need a consistent family of outputs from one brief → this is the node. The per-image price is the same deal as the sequential siblings in this pack, just with V4.5's typography skills on top.

CategoryERPK/WaveSpeedAI

Inputs (10)

NameTypeDefaultDescription
promptSTRINGText description for image generation. The node automatically appends the image count to your prompt.
max_imagesINT41–15Number of images to generate (1-15). Automatically added to prompt for API compliance.
size_presetCOMBOCustomRecommended resolution presets. Select 'Custom' to use manual width/height.
seedINT-1-1–2147483647Seed for cache control. Randomizes by default.
clientoptWAVESPEED_AI_API_CLIENTWaveSpeed API client (optional if API key is configured in Settings)
widthoptINT20481024–4096Custom width (only used when size_preset is 'Custom')
heightoptINT20481024–4096Custom height (only used when size_preset is 'Custom')
show_aspect_ratiooptBOOLEANtrueShow aspect ratio in node title
enable_sync_modeoptBOOLEANfalseWait for result generation before returning response
enable_base64_outputoptBOOLEANfalseReturn BASE64 encoded output instead of URLs

Outputs (1)

NameTypeDescription
imagesIMAGE