FAL Seedream v5 Lite Edit (Soze)

ByteDance's Seedream lineup splits into tiers, and v5 Lite is the one that trades a little quality for a noticeably smaller bill. FAL Seedream v5 Lite Edit wraps fal-ai/bytedance/seedream/v5/lite/edit: up to ten input images, a prompt, and edited results back in your graph as an IMAGE batch. If you've got a pipeline full of edit jobs where the flagship (v5 Pro) is overkill and the quality gap doesn't matter, this is the volume option.

Same paid hosted model family as the other Seedream nodes here - nothing local, pay per run. The Lite tier is where that pricing starts to make batch-ish workflows feel sane.

How it works

Required inputs:

• prompt - the edit instruction, multiline.
• image_size - the notable options are auto_2K, auto_3K and auto_4K (Lite adds the 3K step the v4.5 tier doesn't have), alongside the classic square/portrait_16_9/landscape_16_9 ratios.
• num_images (1–8) - results per run.
• max_images (1–8) - above 1 enables multi-image generation.
• enable_safety_checker - on by default.

Reference inputs: ten image_N slots or an image_batch (capped at ten frames, individual slots winning). sync_mode returns images as data URIs instead of public URLs.

Outputs: images, image_urls, image_count, seed, and status. One difference worth calling out from the v4.5 edit node: there's no seed input here. This node has a seed output (fal's chosen seed, for logging and reproducibility forensics) but no way to pin it going in - Lite runs are seedless from your side. If you need deterministic reruns, that's a v4.5/Pro capability.

When it's worth it

The case is pretty much "edits at volume, quality ceiling acceptable." Iterating a product shot over many small changes, restyling a batch of images, anything where paying flagship prices per call would hurt. If a job is the hero image and the edit quality decides whether it ships, step up to v5 Pro or v4.5 instead - Lite is the workhorse, not the showpiece.

Getting it running and gotchas

Standard pack install, then the FAL key:

cd ComfyUI/custom_nodes
git clone https://github.com/SozeInc/ComfyUI_Soze.git
pip install -r ComfyUI_Soze/requirements.txt

FAL_KEY in your environment, a .env in the pack root, or config.ini under [API]. Missing it reads "FAL_KEY is not configured" on the node.

The gotchas that bite: blank prompt = clean skip with no image (check status), uploads can fail per-slot and show in status, and the num_images/max_images pair - num_images is how many you want, max_images is the cap that unlocks multi-image mode; if results look like you asked for more than you got, one of them is at the wrong value. And don't hunt for a seed input that isn't there - set expectations accordingly when you need reproducibility.