Wavemaker

Chat operates workflows — guided inputs, costs, and outputs — without starting in a node graph. Open the graph when you want power.

  • ✓ One prompt → a directed multi-scene film, with QC on every clip
  • ✓ Credits never expire — no 90-day forfeiture
  • ✓ Publish a workflow once, earn per run — not a one-time template sale
  • ✓ Import ComfyUI JSON and it runs natively (how)
Try in chat Browse workflows
Chat

Product packshot, soft daylight, white seamless — keep the label sharp.

Using product-packshot. Confirm aspect + reference, then I’ll preflight credits.

Graph optional
Load Sample Save QC

Same session — open when you need power.

Static demo of the chat‖graph thesis — not a live app embed (app shell stays noindex).

Not another model aggregator

Aggregators resell access to other people’s models. Wavemaker directs productions — the models are cast; the pipeline, the QC, and the workflows you own are the product.

Model aggregators Wavemaker
A prompt box per model — you assemble the campaign by hand One prompt → a directed multi-scene production: script, scenes, voiceover, music, and assemble in one run
Ships whatever the model returns QC gates on every clip plus a final QC on the assembled cut — failures re-roll or fail honestly, never silently ship
Thin proxy over partner APIs Native Comfy graphs on our GPUs — open the graph, edit it, import ComfyUI JSON and it runs
Opaque credits, often 90-day expiry Dual-meter preflight (GPU + partner) before Confirm; unused reserve refunds; credits never expire
Templates you consume Workflows you own — publish once, earn per run (royalty or premium pricing)
Generic output; brand pasted on after Brand grounding built in: site research → brand kit → letter-perfect ad type burned by the pipeline
Web app only Every workflow is an API endpoint, plus MCP tools for agents

Same models you’d find anywhere — Kling, Seedance, Veo, Seedream and more — but bound into graphs with quality gates, honest metering, and ownership. Named comparisons stay receipt-backed under Compare.

Golden path

Prompt → guided inputs → outputs. Graph optional.

1. Describe

Tell chat the outcome. We recommend a published workflow.

2. Confirm inputs

Fill promoted fields and review credit preflight.

3. Get outputs

Iterate in chat. Switch to graph only when you need it.

Use cases

Curated collections from the bundled catalog — Try deep-links into the app shell.

Treatments

Outcome-shaped workflows — legacy treatments/skills as catalog graphs. Not 1:1 with all 213 skillTypes (spines + aliases only; see legacy-deliverable-parity.md).

Explore Treatments

Look stills

Look-first still workflows (packshot, lifestyle, portrait, fashion, food, architecture) — catalog graphs, not the Style Catalog (ims_*).

Explore Look stills

Training Studio

Train character LoRAs and manage Datasets → Train → Evaluate → Library. Primary CTA opens /platform/studio?recipe=character (not a dead workflow stub). Catalog slug training-studio is a reference still companion.

Explore Training Studio

Weekly drop

Curated weekly featured workflows — cadence resolved at request time from the weekly-drop schedule (ISO week). Distinct from personal Collections.

Explore Weekly drop

Edit · Video face / character / double-exposure

Closed residue (runtime-v2): video face-swap (reactor+VHS), character-swap (IPAdapter+inpaint), double-exposure. LivePortrait reenact is a separate showcase under video (not this residue list).

Explore Edit · Video face / character / double-exposure

Fashion

Concept, lookbook, on-model, and try-on workflows.

Explore Fashion

Product

Packshot, lifestyle, background swap, and 360 product stills.

Explore Product

Portrait

Headshot, face detailer, and identity sheet workflows.

Explore Portrait

Video

I2V, long-take, foley, and upscale/interpolation clips.

Explore Video

Ads

Directed static/commercial ads, creative packs (multi-AR, carousel, asset-group, vertical-feed), and leaf companions.

Explore Ads

Food

Hero and process food photography workflows.

Explore Food

Floyo · Best recipe ports

Highest-ICP Floyo commercial cookbook shapes recreated clean-room (ADR-023). Product insert, multi-angle, jewelry, try-on/UGC video, talking avatar, SeedVR2, beauty/influencer ads.

Explore Floyo · Best recipe ports

Mode · Video

Legacy composer om:v — standard video spines (general + I2V + UGC + cinema seed).

Explore Mode · Video

Mode · Audio

Legacy composer om:a — Music / Voice / Spot / Podcast artifact paths.

Explore Mode · Audio

Mode · Refine

Legacy refine treatments (recast / outfit / face). Bind reference images in the editor.

Explore Mode · Refine

Format packs

Legacy fpk: overlays — social_mute_first, product_plate, tv_broadcast (adult_structure under Adult formats).

Explore Format packs

Adult formats

Legacy system-only adult_* pacing formats + adult_structure — re-expressed as hybrid NSFW graphs (never commerce).

Explore Adult formats

Parity · Core deliverables

Umbrella set covering legacy non-permissive product modes (video, images, audio, catalog, cinema, animate, refine, format packs).

Explore Parity · Core deliverables

SFW adapters (LTX LoRA presets)

Non-permissive LTX adapter demo graphs — phys-ltx floor, cinematic/product/portrait/locked-camera plus CRT/FPV/cinemagraph/transition/audio-reactive/3dreal community presets. See docs/platform/sfw-adapters.md. IC-LoRAs remain lab-path leaves in adapters-ltx-looks.

Explore SFW adapters (LTX LoRA presets)

LTX-2.5 looks + IC / Comfy ports

LTX-2.5 community productization leaves: SFW CRT/FPV/cinemagraph/transition/audio-reactive/3dreal presets, IC capability demos, eros25 eval, and T-08 Comfy bridge ports. See docs/research/ltx-lora-workflow-audit-2026-08-12.md.

Explore LTX-2.5 looks + IC / Comfy ports

Parity · Permissive deliverables

Umbrella set covering legacy NSFW product surfaces — plain/ad-set stills, I2V/video, adult pacing, Grok gateway rail, identity mint (dual-rail LTX + gateway).

Explore Parity · Permissive deliverables

Gear · Default

Legacy system:default — environment defaults. Companion RealVisXL + Gemini + Kling I2V workflows (no forced provider pins).

Explore Gear · Default

Gear · High quality

Legacy system:high-quality — 2K images, pro video tiers, writingModel=fable companions.

Explore Gear · High quality

Gear · In-house

Legacy system:in-house (WAN+Krea2). Platform maps to hybrid LTX /best_seed + LTX distilled video + RealVisXL — not CheckpointLoader krea2 (D-02).

Explore Gear · In-house

Model · Grok Imagine

Official api.x.ai Grok Imagine image + video. Adult Imagine uses WavemakerGrokGateway* / permissive-gateway-* (consumer gateway), not these leaves.

Explore Model · Grok Imagine

Model · Qwen-Image

Qwen-Image base T2I and GGUF quant mirrors (Apache-2.0). Volume residency + GPU verify are ops follow-ups (WAV-877 / WAV-879).

Explore Model · Qwen-Image

Model · Kling

Legacy videoProvider=kling-direct — T2V + I2V (fresh-start default video).

Explore Model · Kling

Model · Partners

Partner video model leaf workflows (Seedance, Veo, Runway, Kling, and more). Trying one binds that leaf graph — it does not change directed SFW Quality Partner presets (Kling/Grok Materialize stacks).

Explore Model · Partners

Model · LTX hybrid

Self-hosted / experimental LTX path — hybrid /best_seed + catalog LTX video (covers legacy ltx + in-house krea2 image pin).

Explore Model · LTX hybrid

Writing · Claude Sonnet

Legacy writingModel=sonnet (claude-sonnet-4-6 on strategize/write_script). Media companions for default gear — Claude is agent-side, not a Comfy node.

Explore Writing · Claude Sonnet

LLM stack · Stable

Legacy admin ModelConfig presets stable-v0.2/v0.3/v0.4 — Claude Sonnet planning + Gemini Flash reviews + Grok Imagine video + ElevenLabs multilingual.

Explore LLM stack · Stable

LLM stack · Fresh Start

Legacy admin ModelConfig fresh-start — Gemini image + Kling video + ElevenLabs v3 + Gemini/Claude orchestrator stack.

Explore LLM stack · Fresh Start

Model · Runway

First-party Runway partner nodes — Gen-4.5 T2V, Aleph video edit, Act-Two performance transfer.

Explore Model · Runway

Model · SkyReels V3

SkyReels V3 R2V / extend / avatar scaffolds on WanVideoWrapper (runtime-v4). Skywork Community License pass-through.

Explore Model · SkyReels V3

Market · Competitor parity ports

Clean-room recreations of the strongest non-Comfy competitor deliverables (2026-08 replication sweep): UGC actor ads, dubbing, viral effects, camera-move packs, article/doc-to-video, upscalers, icon sets, audiobooks, jingles, SFX packs.

Explore Market · Competitor parity ports

Power when you need it

Native Comfy graphs on Modal GPUs — curated packs + partner nodes. Compose, remix, Copy API curl from Simple or Graph, and publish when you’re ready. Public share still requires a verified run. The split above is the thesis: chat operates; graph is the same session.

Honest comparisons

Product-shape pages vs Floyo-class hosts, managed Comfy clouds, and Hub import — receipt claims only; no “beats legacy” until bakeoffs unlock it. Browse comparisons.

Every workflow is an API

One-click Copy API for POST /runs. Dual-meter preflight (GPU + partner APIs). Agents use POST /mcp — ~50 tools (search, apply, edit graph, run, files, wallet, partner nodes). OpenAPI: 57 operations.

Docs: /developers#platform · node library · api-mcp guide · billing rates.

Trust

Verified Private hosting and commercial-use honesty live in the Trust Center. Claims never exceed the public registry.

Credits ↔ models

Before Confirm you see GPU class and Partner API meters for the bound leaf (e.g. Kling video-seconds, ElevenLabs characters). Edit the graph for free; generation reserves credits and refunds the unused remainder. Credits never expire.

See pricing and the billing rates guide.

FAQ

Is Wavemaker just another AI model aggregator?

No. Aggregators resell prompt boxes to partner models. Wavemaker directs productions: one prompt becomes a multi-scene run with script, scenes, voiceover, music, QC gates on every clip, and a final assembled cut. The models are cast — the pipeline, the QC, and the workflows you own are the product.

Do Wavemaker credits expire?

No. Credits never expire — unlike 90-day forfeiture plans common elsewhere.

Can I pick Kling, Veo, or Seedance by name?

Yes. Partner marketing pages under /platform/ai-models deep-link Try into catalog leaves. Directed ads may also meter partners under Quality presets.

How does credit spend map to a model run?

Dual-meter preflight shows GPU class and Partner API units before Confirm. Partner seconds/characters are priced in billing-rates; unused reserve refunds automatically.

Is face swap available on video?

Image face-swap only (refine-face-swap) with likeness consent. Video face-swap is not offered.